diff --git "a/checkpoint-107155/trainer_state.json" "b/checkpoint-107155/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-107155/trainer_state.json"
@@ -0,0 +1,128652 @@
+{
+  "best_metric": 0.03627169877290726,
+  "best_model_checkpoint": "wav2vec2-base-pem123-32-960h-la/checkpoint-64293",
+  "epoch": 5.0,
+  "global_step": 107155,
+  "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
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8084009188504993e-05,
+      "loss": 0.2647,
+      "step": 21435
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8083540387229853e-05,
+      "loss": 0.0701,
+      "step": 21440
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8083071585954713e-05,
+      "loss": 0.0847,
+      "step": 21445
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8082602784679576e-05,
+      "loss": 0.152,
+      "step": 21450
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8082133983404436e-05,
+      "loss": 0.1055,
+      "step": 21455
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8081665182129296e-05,
+      "loss": 0.1901,
+      "step": 21460
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8081196380854156e-05,
+      "loss": 0.2548,
+      "step": 21465
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.808072757957902e-05,
+      "loss": 0.2324,
+      "step": 21470
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.808025877830388e-05,
+      "loss": 0.2663,
+      "step": 21475
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.807978997702874e-05,
+      "loss": 0.3625,
+      "step": 21480
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.80793211757536e-05,
+      "loss": 0.3101,
+      "step": 21485
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.807885237447846e-05,
+      "loss": 0.081,
+      "step": 21490
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.807838357320332e-05,
+      "loss": 0.1012,
+      "step": 21495
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8077914771928183e-05,
+      "loss": 0.1227,
+      "step": 21500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8077445970653043e-05,
+      "loss": 0.1404,
+      "step": 21505
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8076977169377902e-05,
+      "loss": 0.1724,
+      "step": 21510
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8076508368102762e-05,
+      "loss": 0.2312,
+      "step": 21515
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8076039566827622e-05,
+      "loss": 0.2044,
+      "step": 21520
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8075570765552482e-05,
+      "loss": 0.2757,
+      "step": 21525
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8075101964277346e-05,
+      "loss": 0.3255,
+      "step": 21530
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8074633163002206e-05,
+      "loss": 0.2708,
+      "step": 21535
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8074164361727065e-05,
+      "loss": 0.1083,
+      "step": 21540
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8073695560451925e-05,
+      "loss": 0.0432,
+      "step": 21545
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.807322675917679e-05,
+      "loss": 0.0933,
+      "step": 21550
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.807275795790165e-05,
+      "loss": 0.134,
+      "step": 21555
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.807228915662651e-05,
+      "loss": 0.2026,
+      "step": 21560
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.807182035535137e-05,
+      "loss": 0.1708,
+      "step": 21565
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.807135155407623e-05,
+      "loss": 0.1225,
+      "step": 21570
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.807088275280109e-05,
+      "loss": 0.2929,
+      "step": 21575
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8070413951525948e-05,
+      "loss": 0.3245,
+      "step": 21580
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8069945150250808e-05,
+      "loss": 0.2348,
+      "step": 21585
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8069476348975668e-05,
+      "loss": 0.0968,
+      "step": 21590
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.806900754770053e-05,
+      "loss": 0.1464,
+      "step": 21595
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.806853874642539e-05,
+      "loss": 0.0781,
+      "step": 21600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.806806994515025e-05,
+      "loss": 0.1357,
+      "step": 21605
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8067601143875115e-05,
+      "loss": 0.1264,
+      "step": 21610
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8067132342599975e-05,
+      "loss": 0.2146,
+      "step": 21615
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8066663541324835e-05,
+      "loss": 0.1569,
+      "step": 21620
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8066194740049694e-05,
+      "loss": 0.1623,
+      "step": 21625
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8065725938774554e-05,
+      "loss": 0.5652,
+      "step": 21630
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8065257137499414e-05,
+      "loss": 0.347,
+      "step": 21635
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8064788336224278e-05,
+      "loss": 0.047,
+      "step": 21640
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8064319534949138e-05,
+      "loss": 0.0867,
+      "step": 21645
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8063850733673998e-05,
+      "loss": 0.0751,
+      "step": 21650
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8063381932398857e-05,
+      "loss": 0.061,
+      "step": 21655
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8062913131123717e-05,
+      "loss": 0.1454,
+      "step": 21660
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8062444329848577e-05,
+      "loss": 0.1297,
+      "step": 21665
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8061975528573437e-05,
+      "loss": 0.1807,
+      "step": 21670
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.80615067272983e-05,
+      "loss": 0.1411,
+      "step": 21675
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.806103792602316e-05,
+      "loss": 0.3258,
+      "step": 21680
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.806056912474802e-05,
+      "loss": 0.3581,
+      "step": 21685
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8060100323472884e-05,
+      "loss": 0.0666,
+      "step": 21690
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8059631522197744e-05,
+      "loss": 0.0615,
+      "step": 21695
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8059162720922604e-05,
+      "loss": 0.0849,
+      "step": 21700
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8058693919647464e-05,
+      "loss": 0.1099,
+      "step": 21705
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8058225118372324e-05,
+      "loss": 0.1481,
+      "step": 21710
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8057756317097183e-05,
+      "loss": 0.149,
+      "step": 21715
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8057287515822043e-05,
+      "loss": 0.2365,
+      "step": 21720
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8056818714546903e-05,
+      "loss": 0.216,
+      "step": 21725
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8056349913271763e-05,
+      "loss": 0.3717,
+      "step": 21730
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8055881111996627e-05,
+      "loss": 0.3129,
+      "step": 21735
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8055412310721487e-05,
+      "loss": 0.0945,
+      "step": 21740
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8054943509446346e-05,
+      "loss": 0.0414,
+      "step": 21745
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8054474708171206e-05,
+      "loss": 0.0931,
+      "step": 21750
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.805400590689607e-05,
+      "loss": 0.1486,
+      "step": 21755
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.805353710562093e-05,
+      "loss": 0.1353,
+      "step": 21760
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.805306830434579e-05,
+      "loss": 0.2255,
+      "step": 21765
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.805259950307065e-05,
+      "loss": 0.2076,
+      "step": 21770
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.805213070179551e-05,
+      "loss": 0.2506,
+      "step": 21775
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8051661900520373e-05,
+      "loss": 0.2631,
+      "step": 21780
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8051193099245233e-05,
+      "loss": 0.3428,
+      "step": 21785
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8050724297970093e-05,
+      "loss": 0.0835,
+      "step": 21790
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8050255496694953e-05,
+      "loss": 0.1205,
+      "step": 21795
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8049786695419812e-05,
+      "loss": 0.106,
+      "step": 21800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8049317894144672e-05,
+      "loss": 0.1363,
+      "step": 21805
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8048849092869532e-05,
+      "loss": 0.1173,
+      "step": 21810
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8048380291594392e-05,
+      "loss": 0.1431,
+      "step": 21815
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8047911490319256e-05,
+      "loss": 0.1746,
+      "step": 21820
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8047442689044116e-05,
+      "loss": 0.2475,
+      "step": 21825
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8046973887768975e-05,
+      "loss": 0.4129,
+      "step": 21830
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.804650508649384e-05,
+      "loss": 0.3637,
+      "step": 21835
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.80460362852187e-05,
+      "loss": 0.0778,
+      "step": 21840
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.804556748394356e-05,
+      "loss": 0.0479,
+      "step": 21845
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.804509868266842e-05,
+      "loss": 0.1032,
+      "step": 21850
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.804462988139328e-05,
+      "loss": 0.1429,
+      "step": 21855
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.804416108011814e-05,
+      "loss": 0.1455,
+      "step": 21860
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8043692278843e-05,
+      "loss": 0.2018,
+      "step": 21865
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.804322347756786e-05,
+      "loss": 0.1994,
+      "step": 21870
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.804275467629272e-05,
+      "loss": 0.2921,
+      "step": 21875
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.804228587501758e-05,
+      "loss": 0.2647,
+      "step": 21880
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.804181707374244e-05,
+      "loss": 0.3341,
+      "step": 21885
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.80413482724673e-05,
+      "loss": 0.0478,
+      "step": 21890
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8040879471192165e-05,
+      "loss": 0.0548,
+      "step": 21895
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8040410669917025e-05,
+      "loss": 0.0963,
+      "step": 21900
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8039941868641885e-05,
+      "loss": 0.1291,
+      "step": 21905
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8039473067366745e-05,
+      "loss": 0.0829,
+      "step": 21910
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8039004266091605e-05,
+      "loss": 0.1468,
+      "step": 21915
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8038535464816468e-05,
+      "loss": 0.1108,
+      "step": 21920
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8038066663541328e-05,
+      "loss": 0.291,
+      "step": 21925
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8037597862266188e-05,
+      "loss": 0.5618,
+      "step": 21930
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8037129060991048e-05,
+      "loss": 0.2874,
+      "step": 21935
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8036660259715908e-05,
+      "loss": 0.1016,
+      "step": 21940
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8036191458440768e-05,
+      "loss": 0.0921,
+      "step": 21945
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8035722657165627e-05,
+      "loss": 0.094,
+      "step": 21950
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8035253855890487e-05,
+      "loss": 0.1329,
+      "step": 21955
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.803478505461535e-05,
+      "loss": 0.1354,
+      "step": 21960
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.803431625334021e-05,
+      "loss": 0.1359,
+      "step": 21965
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.803384745206507e-05,
+      "loss": 0.1957,
+      "step": 21970
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8033378650789934e-05,
+      "loss": 0.2334,
+      "step": 21975
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8032909849514794e-05,
+      "loss": 0.4382,
+      "step": 21980
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8032441048239654e-05,
+      "loss": 0.2213,
+      "step": 21985
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8031972246964514e-05,
+      "loss": 0.0824,
+      "step": 21990
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8031503445689374e-05,
+      "loss": 0.0964,
+      "step": 21995
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8031034644414234e-05,
+      "loss": 0.0895,
+      "step": 22000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8030565843139093e-05,
+      "loss": 0.1174,
+      "step": 22005
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8030097041863953e-05,
+      "loss": 0.2309,
+      "step": 22010
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8029628240588817e-05,
+      "loss": 0.2318,
+      "step": 22015
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8029159439313677e-05,
+      "loss": 0.1976,
+      "step": 22020
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8028690638038537e-05,
+      "loss": 0.2326,
+      "step": 22025
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8028221836763397e-05,
+      "loss": 0.4568,
+      "step": 22030
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8027753035488256e-05,
+      "loss": 0.3818,
+      "step": 22035
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.802728423421312e-05,
+      "loss": 0.0766,
+      "step": 22040
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.802681543293798e-05,
+      "loss": 0.0628,
+      "step": 22045
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.802634663166284e-05,
+      "loss": 0.1297,
+      "step": 22050
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.80258778303877e-05,
+      "loss": 0.1041,
+      "step": 22055
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8025409029112563e-05,
+      "loss": 0.1585,
+      "step": 22060
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8024940227837423e-05,
+      "loss": 0.1287,
+      "step": 22065
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8024471426562283e-05,
+      "loss": 0.1883,
+      "step": 22070
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8024002625287143e-05,
+      "loss": 0.3461,
+      "step": 22075
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8023533824012003e-05,
+      "loss": 0.4398,
+      "step": 22080
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8023065022736863e-05,
+      "loss": 0.2017,
+      "step": 22085
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8022596221461723e-05,
+      "loss": 0.0727,
+      "step": 22090
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8022127420186582e-05,
+      "loss": 0.0453,
+      "step": 22095
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8021658618911442e-05,
+      "loss": 0.0808,
+      "step": 22100
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8021189817636306e-05,
+      "loss": 0.1692,
+      "step": 22105
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8020721016361166e-05,
+      "loss": 0.1024,
+      "step": 22110
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8020252215086026e-05,
+      "loss": 0.2401,
+      "step": 22115
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.801978341381089e-05,
+      "loss": 0.1554,
+      "step": 22120
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.801931461253575e-05,
+      "loss": 0.3393,
+      "step": 22125
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.801884581126061e-05,
+      "loss": 0.4696,
+      "step": 22130
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.801837700998547e-05,
+      "loss": 0.3602,
+      "step": 22135
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.801790820871033e-05,
+      "loss": 0.1322,
+      "step": 22140
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.801743940743519e-05,
+      "loss": 0.0692,
+      "step": 22145
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8016970606160052e-05,
+      "loss": 0.1042,
+      "step": 22150
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8016501804884912e-05,
+      "loss": 0.1385,
+      "step": 22155
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8016033003609772e-05,
+      "loss": 0.3227,
+      "step": 22160
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.801556420233463e-05,
+      "loss": 0.2212,
+      "step": 22165
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.801509540105949e-05,
+      "loss": 0.2446,
+      "step": 22170
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.801462659978435e-05,
+      "loss": 0.3732,
+      "step": 22175
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.801415779850921e-05,
+      "loss": 0.3812,
+      "step": 22180
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8013688997234075e-05,
+      "loss": 0.4351,
+      "step": 22185
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8013220195958935e-05,
+      "loss": 0.0613,
+      "step": 22190
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8012751394683795e-05,
+      "loss": 0.0991,
+      "step": 22195
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8012282593408658e-05,
+      "loss": 0.1864,
+      "step": 22200
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8011813792133518e-05,
+      "loss": 0.1468,
+      "step": 22205
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8011344990858378e-05,
+      "loss": 0.1907,
+      "step": 22210
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8010876189583238e-05,
+      "loss": 0.1372,
+      "step": 22215
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8010407388308098e-05,
+      "loss": 0.2023,
+      "step": 22220
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8009938587032958e-05,
+      "loss": 0.2943,
+      "step": 22225
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8009469785757818e-05,
+      "loss": 0.2892,
+      "step": 22230
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8009000984482678e-05,
+      "loss": 0.2727,
+      "step": 22235
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8008532183207537e-05,
+      "loss": 0.0482,
+      "step": 22240
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.80080633819324e-05,
+      "loss": 0.0691,
+      "step": 22245
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.800759458065726e-05,
+      "loss": 0.0779,
+      "step": 22250
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.800712577938212e-05,
+      "loss": 0.1529,
+      "step": 22255
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.800665697810698e-05,
+      "loss": 0.1618,
+      "step": 22260
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8006188176831844e-05,
+      "loss": 0.1802,
+      "step": 22265
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8005719375556704e-05,
+      "loss": 0.2725,
+      "step": 22270
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8005250574281564e-05,
+      "loss": 0.3094,
+      "step": 22275
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8004781773006424e-05,
+      "loss": 0.4009,
+      "step": 22280
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8004312971731284e-05,
+      "loss": 0.267,
+      "step": 22285
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8003844170456147e-05,
+      "loss": 0.0792,
+      "step": 22290
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8003375369181007e-05,
+      "loss": 0.1097,
+      "step": 22295
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8002906567905867e-05,
+      "loss": 0.1334,
+      "step": 22300
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8002437766630727e-05,
+      "loss": 0.1154,
+      "step": 22305
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8001968965355587e-05,
+      "loss": 0.1646,
+      "step": 22310
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8001500164080447e-05,
+      "loss": 0.2812,
+      "step": 22315
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8001031362805307e-05,
+      "loss": 0.1536,
+      "step": 22320
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.800056256153017e-05,
+      "loss": 0.1213,
+      "step": 22325
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.800009376025503e-05,
+      "loss": 0.4102,
+      "step": 22330
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.799962495897989e-05,
+      "loss": 0.2346,
+      "step": 22335
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.799915615770475e-05,
+      "loss": 0.0383,
+      "step": 22340
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.7998687356429613e-05,
+      "loss": 0.2049,
+      "step": 22345
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.7998218555154473e-05,
+      "loss": 0.0988,
+      "step": 22350
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.7997749753879333e-05,
+      "loss": 0.1249,
+      "step": 22355
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.7997280952604193e-05,
+      "loss": 0.0932,
+      "step": 22360
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.7996812151329053e-05,
+      "loss": 0.1902,
+      "step": 22365
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.7996343350053913e-05,
+      "loss": 0.2656,
+      "step": 22370
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.7995874548778773e-05,
+      "loss": 0.2387,
+      "step": 22375
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.7995405747503633e-05,
+      "loss": 0.3128,
+      "step": 22380
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.7994936946228496e-05,
+      "loss": 0.2632,
+      "step": 22385
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.7994468144953356e-05,
+      "loss": 0.0293,
+      "step": 22390
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.7993999343678216e-05,
+      "loss": 0.1036,
+      "step": 22395
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.7993530542403076e-05,
+      "loss": 0.1302,
+      "step": 22400
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.799306174112794e-05,
+      "loss": 0.0677,
+      "step": 22405
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.79925929398528e-05,
+      "loss": 0.1077,
+      "step": 22410
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.799212413857766e-05,
+      "loss": 0.2445,
+      "step": 22415
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.799165533730252e-05,
+      "loss": 0.2922,
+      "step": 22420
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.799118653602738e-05,
+      "loss": 0.3946,
+      "step": 22425
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.7990717734752242e-05,
+      "loss": 0.3417,
+      "step": 22430
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.7990248933477102e-05,
+      "loss": 0.2651,
+      "step": 22435
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.7989780132201962e-05,
+      "loss": 0.0262,
+      "step": 22440
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.7989311330926822e-05,
+      "loss": 0.1012,
+      "step": 22445
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.7988842529651682e-05,
+      "loss": 0.1389,
+      "step": 22450
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.7988373728376542e-05,
+      "loss": 0.105,
+      "step": 22455
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.79879049271014e-05,
+      "loss": 0.1181,
+      "step": 22460
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.798743612582626e-05,
+      "loss": 0.1825,
+      "step": 22465
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.7986967324551125e-05,
+      "loss": 0.2169,
+      "step": 22470
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.7986498523275985e-05,
+      "loss": 0.2781,
+      "step": 22475
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.7986029722000845e-05,
+      "loss": 0.3195,
+      "step": 22480
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.7985560920725708e-05,
+      "loss": 0.2792,
+      "step": 22485
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.7985092119450568e-05,
+      "loss": 0.0311,
+      "step": 22490
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.7984623318175428e-05,
+      "loss": 0.0569,
+      "step": 22495
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.7984154516900288e-05,
+      "loss": 0.122,
+      "step": 22500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.7983685715625148e-05,
+      "loss": 0.1595,
+      "step": 22505
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.7983216914350008e-05,
+      "loss": 0.1147,
+      "step": 22510
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.7982748113074868e-05,
+      "loss": 0.1325,
+      "step": 22515
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.7982279311799728e-05,
+      "loss": 0.2697,
+      "step": 22520
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.798181051052459e-05,
+      "loss": 0.3659,
+      "step": 22525
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.798134170924945e-05,
+      "loss": 0.3407,
+      "step": 22530
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.798087290797431e-05,
+      "loss": 0.2422,
+      "step": 22535
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.798040410669917e-05,
+      "loss": 0.103,
+      "step": 22540
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.797993530542403e-05,
+      "loss": 0.1043,
+      "step": 22545
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.7979466504148894e-05,
+      "loss": 0.1243,
+      "step": 22550
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.7978997702873754e-05,
+      "loss": 0.1655,
+      "step": 22555
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.7978528901598614e-05,
+      "loss": 0.1039,
+      "step": 22560
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.7978060100323474e-05,
+      "loss": 0.1785,
+      "step": 22565
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.7977591299048337e-05,
+      "loss": 0.1834,
+      "step": 22570
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.7977122497773197e-05,
+      "loss": 0.2746,
+      "step": 22575
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.7976653696498057e-05,
+      "loss": 0.3149,
+      "step": 22580
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.7976184895222917e-05,
+      "loss": 0.2779,
+      "step": 22585
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.7975716093947777e-05,
+      "loss": 0.1215,
+      "step": 22590
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.7975247292672637e-05,
+      "loss": 0.0716,
+      "step": 22595
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.7974778491397497e-05,
+      "loss": 0.0738,
+      "step": 22600
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.7974309690122357e-05,
+      "loss": 0.1404,
+      "step": 22605
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.7973840888847217e-05,
+      "loss": 0.1767,
+      "step": 22610
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.797337208757208e-05,
+      "loss": 0.1615,
+      "step": 22615
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.797290328629694e-05,
+      "loss": 0.1144,
+      "step": 22620
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.79724344850218e-05,
+      "loss": 0.3295,
+      "step": 22625
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.7971965683746663e-05,
+      "loss": 0.3561,
+      "step": 22630
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.7971496882471523e-05,
+      "loss": 0.3627,
+      "step": 22635
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.7971028081196383e-05,
+      "loss": 0.0388,
+      "step": 22640
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.7970559279921243e-05,
+      "loss": 0.0624,
+      "step": 22645
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.7970090478646103e-05,
+      "loss": 0.1464,
+      "step": 22650
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.7969621677370963e-05,
+      "loss": 0.1578,
+      "step": 22655
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.7969152876095823e-05,
+      "loss": 0.2061,
+      "step": 22660
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.7968684074820686e-05,
+      "loss": 0.1574,
+      "step": 22665
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.7968215273545546e-05,
+      "loss": 0.293,
+      "step": 22670
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.7967746472270406e-05,
+      "loss": 0.3353,
+      "step": 22675
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.7967277670995266e-05,
+      "loss": 0.387,
+      "step": 22680
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.7966808869720126e-05,
+      "loss": 0.3516,
+      "step": 22685
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.7966340068444986e-05,
+      "loss": 0.0655,
+      "step": 22690
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.796587126716985e-05,
+      "loss": 0.0642,
+      "step": 22695
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.796540246589471e-05,
+      "loss": 0.12,
+      "step": 22700
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.796493366461957e-05,
+      "loss": 0.1054,
+      "step": 22705
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.7964464863344432e-05,
+      "loss": 0.1462,
+      "step": 22710
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.7963996062069292e-05,
+      "loss": 0.1599,
+      "step": 22715
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.7963527260794152e-05,
+      "loss": 0.2888,
+      "step": 22720
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.7963058459519012e-05,
+      "loss": 0.1973,
+      "step": 22725
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.7962589658243872e-05,
+      "loss": 0.3846,
+      "step": 22730
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.7962120856968732e-05,
+      "loss": 0.3066,
+      "step": 22735
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.7961652055693592e-05,
+      "loss": 0.0643,
+      "step": 22740
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.7961183254418452e-05,
+      "loss": 0.0761,
+      "step": 22745
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.796071445314331e-05,
+      "loss": 0.0798,
+      "step": 22750
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.7960245651868175e-05,
+      "loss": 0.16,
+      "step": 22755
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.7959776850593035e-05,
+      "loss": 0.155,
+      "step": 22760
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.7959308049317895e-05,
+      "loss": 0.2153,
+      "step": 22765
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.7958839248042755e-05,
+      "loss": 0.191,
+      "step": 22770
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.7958370446767618e-05,
+      "loss": 0.2506,
+      "step": 22775
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.7957901645492478e-05,
+      "loss": 0.4182,
+      "step": 22780
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.7957432844217338e-05,
+      "loss": 0.3069,
+      "step": 22785
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.7956964042942198e-05,
+      "loss": 0.0769,
+      "step": 22790
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.7956495241667058e-05,
+      "loss": 0.1017,
+      "step": 22795
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.795602644039192e-05,
+      "loss": 0.1326,
+      "step": 22800
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.795555763911678e-05,
+      "loss": 0.0708,
+      "step": 22805
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.795508883784164e-05,
+      "loss": 0.2177,
+      "step": 22810
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.79546200365665e-05,
+      "loss": 0.1882,
+      "step": 22815
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.795415123529136e-05,
+      "loss": 0.1609,
+      "step": 22820
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.795368243401622e-05,
+      "loss": 0.1834,
+      "step": 22825
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.795321363274108e-05,
+      "loss": 0.4985,
+      "step": 22830
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.7952744831465944e-05,
+      "loss": 0.2931,
+      "step": 22835
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.7952276030190804e-05,
+      "loss": 0.036,
+      "step": 22840
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.7951807228915664e-05,
+      "loss": 0.0431,
+      "step": 22845
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.7951338427640524e-05,
+      "loss": 0.1114,
+      "step": 22850
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.7950869626365387e-05,
+      "loss": 0.1402,
+      "step": 22855
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.7950400825090247e-05,
+      "loss": 0.1071,
+      "step": 22860
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.7949932023815107e-05,
+      "loss": 0.2206,
+      "step": 22865
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.7949463222539967e-05,
+      "loss": 0.1546,
+      "step": 22870
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.7948994421264827e-05,
+      "loss": 0.2827,
+      "step": 22875
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.7948525619989687e-05,
+      "loss": 0.384,
+      "step": 22880
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.7948056818714547e-05,
+      "loss": 0.162,
+      "step": 22885
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.7947588017439407e-05,
+      "loss": 0.0751,
+      "step": 22890
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.794711921616427e-05,
+      "loss": 0.0633,
+      "step": 22895
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.794665041488913e-05,
+      "loss": 0.097,
+      "step": 22900
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.794618161361399e-05,
+      "loss": 0.0881,
+      "step": 22905
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.794571281233885e-05,
+      "loss": 0.1653,
+      "step": 22910
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.7945244011063713e-05,
+      "loss": 0.212,
+      "step": 22915
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.7944775209788573e-05,
+      "loss": 0.217,
+      "step": 22920
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.7944306408513433e-05,
+      "loss": 0.2547,
+      "step": 22925
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.7943837607238293e-05,
+      "loss": 0.3459,
+      "step": 22930
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.7943368805963153e-05,
+      "loss": 0.3144,
+      "step": 22935
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.7942900004688016e-05,
+      "loss": 0.0644,
+      "step": 22940
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.7942431203412876e-05,
+      "loss": 0.0695,
+      "step": 22945
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.7941962402137736e-05,
+      "loss": 0.0989,
+      "step": 22950
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.7941493600862596e-05,
+      "loss": 0.0713,
+      "step": 22955
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.7941024799587456e-05,
+      "loss": 0.1471,
+      "step": 22960
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.7940555998312316e-05,
+      "loss": 0.7201,
+      "step": 22965
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.7940087197037176e-05,
+      "loss": 0.3036,
+      "step": 22970
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.7939618395762036e-05,
+      "loss": 0.2699,
+      "step": 22975
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.79391495944869e-05,
+      "loss": 0.3579,
+      "step": 22980
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.793868079321176e-05,
+      "loss": 0.2728,
+      "step": 22985
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.793821199193662e-05,
+      "loss": 0.0657,
+      "step": 22990
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.7937743190661482e-05,
+      "loss": 0.0762,
+      "step": 22995
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.7937274389386342e-05,
+      "loss": 0.146,
+      "step": 23000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.7936805588111202e-05,
+      "loss": 0.1352,
+      "step": 23005
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.7936336786836062e-05,
+      "loss": 0.126,
+      "step": 23010
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.7935867985560922e-05,
+      "loss": 0.1249,
+      "step": 23015
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.7935399184285782e-05,
+      "loss": 0.1431,
+      "step": 23020
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.7934930383010642e-05,
+      "loss": 0.1323,
+      "step": 23025
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.7934461581735502e-05,
+      "loss": 0.3606,
+      "step": 23030
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.7933992780460365e-05,
+      "loss": 0.2748,
+      "step": 23035
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.7933523979185225e-05,
+      "loss": 0.0852,
+      "step": 23040
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.7933055177910085e-05,
+      "loss": 0.0636,
+      "step": 23045
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.7932586376634945e-05,
+      "loss": 0.1535,
+      "step": 23050
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.7932117575359805e-05,
+      "loss": 0.146,
+      "step": 23055
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.7931648774084668e-05,
+      "loss": 0.1514,
+      "step": 23060
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.7931179972809528e-05,
+      "loss": 0.1452,
+      "step": 23065
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.7930711171534388e-05,
+      "loss": 0.1974,
+      "step": 23070
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.7930242370259248e-05,
+      "loss": 0.338,
+      "step": 23075
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.792977356898411e-05,
+      "loss": 0.3308,
+      "step": 23080
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.792930476770897e-05,
+      "loss": 0.2528,
+      "step": 23085
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.792883596643383e-05,
+      "loss": 0.0529,
+      "step": 23090
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.792836716515869e-05,
+      "loss": 0.0403,
+      "step": 23095
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.792789836388355e-05,
+      "loss": 0.1068,
+      "step": 23100
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.792742956260841e-05,
+      "loss": 0.1184,
+      "step": 23105
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.792696076133327e-05,
+      "loss": 0.1366,
+      "step": 23110
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.792649196005813e-05,
+      "loss": 0.1412,
+      "step": 23115
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.792602315878299e-05,
+      "loss": 0.3482,
+      "step": 23120
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.7925554357507854e-05,
+      "loss": 0.2595,
+      "step": 23125
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.7925085556232714e-05,
+      "loss": 0.3428,
+      "step": 23130
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.7924616754957574e-05,
+      "loss": 0.3361,
+      "step": 23135
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.7924147953682437e-05,
+      "loss": 0.0959,
+      "step": 23140
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.7923679152407297e-05,
+      "loss": 0.0802,
+      "step": 23145
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.7923210351132157e-05,
+      "loss": 0.1528,
+      "step": 23150
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.7922741549857017e-05,
+      "loss": 0.0776,
+      "step": 23155
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.7922272748581877e-05,
+      "loss": 0.1456,
+      "step": 23160
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.7921803947306737e-05,
+      "loss": 0.1474,
+      "step": 23165
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.7921335146031597e-05,
+      "loss": 0.1562,
+      "step": 23170
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.792086634475646e-05,
+      "loss": 0.2387,
+      "step": 23175
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.792039754348132e-05,
+      "loss": 0.4654,
+      "step": 23180
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.791992874220618e-05,
+      "loss": 0.2778,
+      "step": 23185
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.791945994093104e-05,
+      "loss": 0.0927,
+      "step": 23190
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.79189911396559e-05,
+      "loss": 0.0786,
+      "step": 23195
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.791852233838076e-05,
+      "loss": 0.0507,
+      "step": 23200
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.7918053537105623e-05,
+      "loss": 0.1528,
+      "step": 23205
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.7917584735830483e-05,
+      "loss": 0.146,
+      "step": 23210
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.7917115934555343e-05,
+      "loss": 0.1655,
+      "step": 23215
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.7916647133280206e-05,
+      "loss": 0.1505,
+      "step": 23220
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.7916178332005066e-05,
+      "loss": 0.2023,
+      "step": 23225
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.7915709530729926e-05,
+      "loss": 0.348,
+      "step": 23230
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.7915240729454786e-05,
+      "loss": 0.3989,
+      "step": 23235
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.7914771928179646e-05,
+      "loss": 0.0733,
+      "step": 23240
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.7914303126904506e-05,
+      "loss": 0.0842,
+      "step": 23245
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.7913834325629366e-05,
+      "loss": 0.0824,
+      "step": 23250
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.7913365524354226e-05,
+      "loss": 0.1814,
+      "step": 23255
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.7912896723079086e-05,
+      "loss": 0.088,
+      "step": 23260
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.791242792180395e-05,
+      "loss": 0.1784,
+      "step": 23265
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.791195912052881e-05,
+      "loss": 0.1825,
+      "step": 23270
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.791149031925367e-05,
+      "loss": 0.1527,
+      "step": 23275
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.7911021517978532e-05,
+      "loss": 0.2648,
+      "step": 23280
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.7910552716703392e-05,
+      "loss": 0.4263,
+      "step": 23285
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.7910083915428252e-05,
+      "loss": 0.062,
+      "step": 23290
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.7909615114153112e-05,
+      "loss": 0.0452,
+      "step": 23295
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.7909146312877972e-05,
+      "loss": 0.0897,
+      "step": 23300
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.7908677511602832e-05,
+      "loss": 0.1544,
+      "step": 23305
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.7908208710327692e-05,
+      "loss": 0.1552,
+      "step": 23310
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.7907739909052555e-05,
+      "loss": 0.2193,
+      "step": 23315
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.7907271107777415e-05,
+      "loss": 0.1309,
+      "step": 23320
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.7906802306502275e-05,
+      "loss": 0.2026,
+      "step": 23325
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.7906333505227135e-05,
+      "loss": 0.3341,
+      "step": 23330
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.7905864703951995e-05,
+      "loss": 0.3806,
+      "step": 23335
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.7905395902676855e-05,
+      "loss": 0.09,
+      "step": 23340
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.790492710140172e-05,
+      "loss": 0.079,
+      "step": 23345
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.7904458300126578e-05,
+      "loss": 0.155,
+      "step": 23350
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.7903989498851438e-05,
+      "loss": 0.2088,
+      "step": 23355
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.79035206975763e-05,
+      "loss": 0.1606,
+      "step": 23360
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.790305189630116e-05,
+      "loss": 0.109,
+      "step": 23365
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.790258309502602e-05,
+      "loss": 0.1766,
+      "step": 23370
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.790211429375088e-05,
+      "loss": 0.1729,
+      "step": 23375
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.790164549247574e-05,
+      "loss": 0.3656,
+      "step": 23380
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.79011766912006e-05,
+      "loss": 0.2559,
+      "step": 23385
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.790070788992546e-05,
+      "loss": 0.0238,
+      "step": 23390
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.790023908865032e-05,
+      "loss": 0.0779,
+      "step": 23395
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.789977028737518e-05,
+      "loss": 0.1486,
+      "step": 23400
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.7899301486100044e-05,
+      "loss": 0.1425,
+      "step": 23405
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.7898832684824904e-05,
+      "loss": 0.1606,
+      "step": 23410
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.7898363883549764e-05,
+      "loss": 0.1494,
+      "step": 23415
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.7897895082274624e-05,
+      "loss": 0.1622,
+      "step": 23420
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.7897426280999487e-05,
+      "loss": 0.2253,
+      "step": 23425
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.7896957479724347e-05,
+      "loss": 0.3518,
+      "step": 23430
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.7896488678449207e-05,
+      "loss": 0.2974,
+      "step": 23435
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.7896019877174067e-05,
+      "loss": 0.0553,
+      "step": 23440
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.7895551075898927e-05,
+      "loss": 0.0576,
+      "step": 23445
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.7895082274623787e-05,
+      "loss": 0.1191,
+      "step": 23450
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.789461347334865e-05,
+      "loss": 0.1455,
+      "step": 23455
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.789414467207351e-05,
+      "loss": 0.1459,
+      "step": 23460
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.789367587079837e-05,
+      "loss": 0.19,
+      "step": 23465
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.789320706952323e-05,
+      "loss": 0.2075,
+      "step": 23470
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.789273826824809e-05,
+      "loss": 0.2089,
+      "step": 23475
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.789226946697295e-05,
+      "loss": 0.6432,
+      "step": 23480
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.789180066569781e-05,
+      "loss": 0.293,
+      "step": 23485
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7891331864422673e-05,
+      "loss": 0.148,
+      "step": 23490
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7890863063147533e-05,
+      "loss": 0.0799,
+      "step": 23495
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7890394261872393e-05,
+      "loss": 0.1131,
+      "step": 23500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7889925460597257e-05,
+      "loss": 0.1146,
+      "step": 23505
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7889456659322116e-05,
+      "loss": 0.0917,
+      "step": 23510
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7888987858046976e-05,
+      "loss": 0.2351,
+      "step": 23515
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7888519056771836e-05,
+      "loss": 0.1878,
+      "step": 23520
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7888050255496696e-05,
+      "loss": 0.3739,
+      "step": 23525
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7887581454221556e-05,
+      "loss": 0.339,
+      "step": 23530
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7887112652946416e-05,
+      "loss": 0.2131,
+      "step": 23535
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7886643851671276e-05,
+      "loss": 0.0588,
+      "step": 23540
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.788617505039614e-05,
+      "loss": 0.0581,
+      "step": 23545
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7885706249121e-05,
+      "loss": 0.0815,
+      "step": 23550
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.788523744784586e-05,
+      "loss": 0.116,
+      "step": 23555
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.788476864657072e-05,
+      "loss": 0.0722,
+      "step": 23560
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.788429984529558e-05,
+      "loss": 0.1832,
+      "step": 23565
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7883831044020442e-05,
+      "loss": 0.2784,
+      "step": 23570
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7883362242745302e-05,
+      "loss": 0.2925,
+      "step": 23575
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7882893441470162e-05,
+      "loss": 0.3034,
+      "step": 23580
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7882424640195022e-05,
+      "loss": 0.306,
+      "step": 23585
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7881955838919886e-05,
+      "loss": 0.042,
+      "step": 23590
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7881487037644746e-05,
+      "loss": 0.0754,
+      "step": 23595
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7881018236369605e-05,
+      "loss": 0.0822,
+      "step": 23600
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7880549435094465e-05,
+      "loss": 0.1275,
+      "step": 23605
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7880080633819325e-05,
+      "loss": 0.2322,
+      "step": 23610
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7879611832544185e-05,
+      "loss": 0.1159,
+      "step": 23615
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7879143031269045e-05,
+      "loss": 0.0986,
+      "step": 23620
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7878674229993905e-05,
+      "loss": 0.3391,
+      "step": 23625
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.787820542871877e-05,
+      "loss": 0.4291,
+      "step": 23630
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.787773662744363e-05,
+      "loss": 0.3598,
+      "step": 23635
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7877267826168488e-05,
+      "loss": 0.0618,
+      "step": 23640
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7876799024893348e-05,
+      "loss": 0.0916,
+      "step": 23645
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.787633022361821e-05,
+      "loss": 0.1183,
+      "step": 23650
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.787586142234307e-05,
+      "loss": 0.2357,
+      "step": 23655
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.787539262106793e-05,
+      "loss": 0.1575,
+      "step": 23660
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.787492381979279e-05,
+      "loss": 0.1602,
+      "step": 23665
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.787445501851765e-05,
+      "loss": 0.2106,
+      "step": 23670
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.787398621724251e-05,
+      "loss": 0.2623,
+      "step": 23675
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.787351741596737e-05,
+      "loss": 0.3984,
+      "step": 23680
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7873048614692234e-05,
+      "loss": 0.4425,
+      "step": 23685
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7872579813417094e-05,
+      "loss": 0.0526,
+      "step": 23690
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7872111012141954e-05,
+      "loss": 0.0902,
+      "step": 23695
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7871642210866814e-05,
+      "loss": 0.0561,
+      "step": 23700
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7871173409591674e-05,
+      "loss": 0.1389,
+      "step": 23705
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7870704608316538e-05,
+      "loss": 0.1042,
+      "step": 23710
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7870235807041397e-05,
+      "loss": 0.2266,
+      "step": 23715
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7869767005766257e-05,
+      "loss": 0.2452,
+      "step": 23720
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7869298204491117e-05,
+      "loss": 0.2247,
+      "step": 23725
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.786882940321598e-05,
+      "loss": 0.472,
+      "step": 23730
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.786836060194084e-05,
+      "loss": 0.268,
+      "step": 23735
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.78678918006657e-05,
+      "loss": 0.1071,
+      "step": 23740
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.786742299939056e-05,
+      "loss": 0.0653,
+      "step": 23745
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.786695419811542e-05,
+      "loss": 0.089,
+      "step": 23750
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.786648539684028e-05,
+      "loss": 0.1511,
+      "step": 23755
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.786601659556514e-05,
+      "loss": 0.2224,
+      "step": 23760
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.786554779429e-05,
+      "loss": 0.1669,
+      "step": 23765
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.786507899301486e-05,
+      "loss": 0.2287,
+      "step": 23770
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7864610191739723e-05,
+      "loss": 0.2261,
+      "step": 23775
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7864141390464583e-05,
+      "loss": 0.2919,
+      "step": 23780
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7863672589189443e-05,
+      "loss": 0.2836,
+      "step": 23785
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7863203787914307e-05,
+      "loss": 0.0762,
+      "step": 23790
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7862734986639167e-05,
+      "loss": 0.0781,
+      "step": 23795
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7862266185364026e-05,
+      "loss": 0.0974,
+      "step": 23800
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7861797384088886e-05,
+      "loss": 0.0707,
+      "step": 23805
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7861328582813746e-05,
+      "loss": 0.1144,
+      "step": 23810
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7860859781538606e-05,
+      "loss": 0.2205,
+      "step": 23815
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7860390980263466e-05,
+      "loss": 0.2307,
+      "step": 23820
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.785992217898833e-05,
+      "loss": 0.2626,
+      "step": 23825
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.785945337771319e-05,
+      "loss": 0.372,
+      "step": 23830
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.785898457643805e-05,
+      "loss": 0.2216,
+      "step": 23835
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.785851577516291e-05,
+      "loss": 0.0576,
+      "step": 23840
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.785804697388777e-05,
+      "loss": 0.0833,
+      "step": 23845
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.785757817261263e-05,
+      "loss": 0.1436,
+      "step": 23850
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7857109371337493e-05,
+      "loss": 0.1121,
+      "step": 23855
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7856640570062352e-05,
+      "loss": 0.139,
+      "step": 23860
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7856171768787212e-05,
+      "loss": 0.2333,
+      "step": 23865
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7855702967512076e-05,
+      "loss": 0.2352,
+      "step": 23870
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7855234166236936e-05,
+      "loss": 0.2574,
+      "step": 23875
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7854765364961796e-05,
+      "loss": 0.5096,
+      "step": 23880
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7854296563686656e-05,
+      "loss": 0.3584,
+      "step": 23885
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7853827762411515e-05,
+      "loss": 0.0982,
+      "step": 23890
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7853358961136375e-05,
+      "loss": 0.0527,
+      "step": 23895
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.7852890159861235e-05,
+      "loss": 0.0734,
+      "step": 23900
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.7852421358586095e-05,
+      "loss": 0.0671,
+      "step": 23905
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.7851952557310955e-05,
+      "loss": 0.1738,
+      "step": 23910
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.785148375603582e-05,
+      "loss": 0.1695,
+      "step": 23915
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.785101495476068e-05,
+      "loss": 0.2176,
+      "step": 23920
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.785054615348554e-05,
+      "loss": 0.2628,
+      "step": 23925
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.78500773522104e-05,
+      "loss": 0.3754,
+      "step": 23930
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.784960855093526e-05,
+      "loss": 0.3783,
+      "step": 23935
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.784913974966012e-05,
+      "loss": 0.0661,
+      "step": 23940
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.784867094838498e-05,
+      "loss": 0.0745,
+      "step": 23945
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.784820214710984e-05,
+      "loss": 0.0752,
+      "step": 23950
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.78477333458347e-05,
+      "loss": 0.0765,
+      "step": 23955
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.784726454455956e-05,
+      "loss": 0.177,
+      "step": 23960
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.7846795743284425e-05,
+      "loss": 0.2159,
+      "step": 23965
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.7846326942009285e-05,
+      "loss": 0.1894,
+      "step": 23970
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.7845858140734145e-05,
+      "loss": 0.1678,
+      "step": 23975
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.7845389339459004e-05,
+      "loss": 0.3089,
+      "step": 23980
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.7844920538183864e-05,
+      "loss": 0.2526,
+      "step": 23985
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.7844451736908724e-05,
+      "loss": 0.0912,
+      "step": 23990
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.7843982935633584e-05,
+      "loss": 0.098,
+      "step": 23995
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.7843514134358448e-05,
+      "loss": 0.0601,
+      "step": 24000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.7843045333083307e-05,
+      "loss": 0.1781,
+      "step": 24005
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.7842576531808167e-05,
+      "loss": 0.1413,
+      "step": 24010
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.784210773053303e-05,
+      "loss": 0.2184,
+      "step": 24015
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.784163892925789e-05,
+      "loss": 0.2072,
+      "step": 24020
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.784117012798275e-05,
+      "loss": 0.2547,
+      "step": 24025
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.784070132670761e-05,
+      "loss": 0.328,
+      "step": 24030
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.784023252543247e-05,
+      "loss": 0.3129,
+      "step": 24035
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.783976372415733e-05,
+      "loss": 0.0616,
+      "step": 24040
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.783929492288219e-05,
+      "loss": 0.0432,
+      "step": 24045
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.783882612160705e-05,
+      "loss": 0.0913,
+      "step": 24050
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.7838357320331914e-05,
+      "loss": 0.1446,
+      "step": 24055
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.7837888519056774e-05,
+      "loss": 0.1433,
+      "step": 24060
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.7837419717781633e-05,
+      "loss": 0.1023,
+      "step": 24065
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.7836950916506493e-05,
+      "loss": 0.188,
+      "step": 24070
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.7836482115231353e-05,
+      "loss": 0.1815,
+      "step": 24075
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.7836013313956217e-05,
+      "loss": 0.4238,
+      "step": 24080
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.7835544512681077e-05,
+      "loss": 0.3328,
+      "step": 24085
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.7835075711405937e-05,
+      "loss": 0.1068,
+      "step": 24090
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.7834606910130796e-05,
+      "loss": 0.0785,
+      "step": 24095
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.7834138108855656e-05,
+      "loss": 0.1101,
+      "step": 24100
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.783366930758052e-05,
+      "loss": 0.169,
+      "step": 24105
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.783320050630538e-05,
+      "loss": 0.1494,
+      "step": 24110
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.783273170503024e-05,
+      "loss": 0.1267,
+      "step": 24115
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.78322629037551e-05,
+      "loss": 0.1311,
+      "step": 24120
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.783179410247996e-05,
+      "loss": 0.299,
+      "step": 24125
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.783132530120482e-05,
+      "loss": 0.2228,
+      "step": 24130
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.783085649992968e-05,
+      "loss": 0.309,
+      "step": 24135
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.7830387698654543e-05,
+      "loss": 0.0578,
+      "step": 24140
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.7829918897379403e-05,
+      "loss": 0.0749,
+      "step": 24145
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.7829450096104263e-05,
+      "loss": 0.1276,
+      "step": 24150
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.7828981294829122e-05,
+      "loss": 0.1273,
+      "step": 24155
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.7828512493553986e-05,
+      "loss": 0.1065,
+      "step": 24160
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.7828043692278846e-05,
+      "loss": 0.2525,
+      "step": 24165
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.7827574891003706e-05,
+      "loss": 0.1666,
+      "step": 24170
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.7827106089728566e-05,
+      "loss": 0.2416,
+      "step": 24175
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.7826637288453426e-05,
+      "loss": 0.5039,
+      "step": 24180
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.7826168487178285e-05,
+      "loss": 0.3964,
+      "step": 24185
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.7825699685903145e-05,
+      "loss": 0.0342,
+      "step": 24190
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.782523088462801e-05,
+      "loss": 0.0864,
+      "step": 24195
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.782476208335287e-05,
+      "loss": 0.1232,
+      "step": 24200
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.782429328207773e-05,
+      "loss": 0.149,
+      "step": 24205
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.782382448080259e-05,
+      "loss": 0.1592,
+      "step": 24210
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.782335567952745e-05,
+      "loss": 0.2233,
+      "step": 24215
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.7822886878252312e-05,
+      "loss": 0.1183,
+      "step": 24220
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.782241807697717e-05,
+      "loss": 0.1203,
+      "step": 24225
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.782194927570203e-05,
+      "loss": 0.2905,
+      "step": 24230
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.782148047442689e-05,
+      "loss": 0.2715,
+      "step": 24235
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.7821011673151755e-05,
+      "loss": 0.026,
+      "step": 24240
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.7820542871876615e-05,
+      "loss": 0.0983,
+      "step": 24245
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.7820074070601475e-05,
+      "loss": 0.0667,
+      "step": 24250
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.7819605269326335e-05,
+      "loss": 0.1388,
+      "step": 24255
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.7819136468051195e-05,
+      "loss": 0.0779,
+      "step": 24260
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.7818667666776055e-05,
+      "loss": 0.2576,
+      "step": 24265
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.7818198865500914e-05,
+      "loss": 0.2524,
+      "step": 24270
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.7817730064225774e-05,
+      "loss": 0.328,
+      "step": 24275
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.7817261262950634e-05,
+      "loss": 0.4354,
+      "step": 24280
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.7816792461675498e-05,
+      "loss": 0.362,
+      "step": 24285
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.7816323660400358e-05,
+      "loss": 0.0855,
+      "step": 24290
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.7815854859125218e-05,
+      "loss": 0.0941,
+      "step": 24295
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.781538605785008e-05,
+      "loss": 0.1006,
+      "step": 24300
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.781491725657494e-05,
+      "loss": 0.1203,
+      "step": 24305
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.78144484552998e-05,
+      "loss": 0.1505,
+      "step": 24310
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.781397965402466e-05,
+      "loss": 0.3013,
+      "step": 24315
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.781351085274952e-05,
+      "loss": 0.1516,
+      "step": 24320
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.781304205147438e-05,
+      "loss": 0.1622,
+      "step": 24325
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.781257325019924e-05,
+      "loss": 0.3296,
+      "step": 24330
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.7812104448924104e-05,
+      "loss": 0.2447,
+      "step": 24335
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.7811635647648964e-05,
+      "loss": 0.0976,
+      "step": 24340
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.7811166846373824e-05,
+      "loss": 0.0757,
+      "step": 24345
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.7810698045098684e-05,
+      "loss": 0.0458,
+      "step": 24350
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.7810229243823544e-05,
+      "loss": 0.1581,
+      "step": 24355
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.7809760442548403e-05,
+      "loss": 0.1213,
+      "step": 24360
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.7809291641273267e-05,
+      "loss": 0.1537,
+      "step": 24365
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.7808822839998127e-05,
+      "loss": 0.2788,
+      "step": 24370
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.7808354038722987e-05,
+      "loss": 0.2862,
+      "step": 24375
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.780788523744785e-05,
+      "loss": 0.3774,
+      "step": 24380
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.780741643617271e-05,
+      "loss": 0.2727,
+      "step": 24385
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.780694763489757e-05,
+      "loss": 0.0509,
+      "step": 24390
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.780647883362243e-05,
+      "loss": 0.0936,
+      "step": 24395
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.780601003234729e-05,
+      "loss": 0.0708,
+      "step": 24400
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.780554123107215e-05,
+      "loss": 0.0753,
+      "step": 24405
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.780507242979701e-05,
+      "loss": 0.1762,
+      "step": 24410
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.780460362852187e-05,
+      "loss": 0.1752,
+      "step": 24415
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.780413482724673e-05,
+      "loss": 0.2408,
+      "step": 24420
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.780366602597159e-05,
+      "loss": 0.2928,
+      "step": 24425
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.7803197224696453e-05,
+      "loss": 0.4798,
+      "step": 24430
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.7802728423421313e-05,
+      "loss": 0.407,
+      "step": 24435
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.7802259622146173e-05,
+      "loss": 0.1121,
+      "step": 24440
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.7801790820871036e-05,
+      "loss": 0.0546,
+      "step": 24445
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.7801322019595896e-05,
+      "loss": 0.078,
+      "step": 24450
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.7800853218320756e-05,
+      "loss": 0.1126,
+      "step": 24455
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.7800384417045616e-05,
+      "loss": 0.2212,
+      "step": 24460
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.7799915615770476e-05,
+      "loss": 0.1623,
+      "step": 24465
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.7799446814495336e-05,
+      "loss": 0.3043,
+      "step": 24470
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.77989780132202e-05,
+      "loss": 0.3083,
+      "step": 24475
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.779850921194506e-05,
+      "loss": 0.3038,
+      "step": 24480
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.779804041066992e-05,
+      "loss": 0.3422,
+      "step": 24485
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.779757160939478e-05,
+      "loss": 0.0868,
+      "step": 24490
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.779710280811964e-05,
+      "loss": 0.0834,
+      "step": 24495
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.77966340068445e-05,
+      "loss": 0.0701,
+      "step": 24500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.779616520556936e-05,
+      "loss": 0.1719,
+      "step": 24505
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.7795696404294222e-05,
+      "loss": 0.0946,
+      "step": 24510
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.7795227603019082e-05,
+      "loss": 0.1765,
+      "step": 24515
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.779475880174394e-05,
+      "loss": 0.256,
+      "step": 24520
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.7794290000468805e-05,
+      "loss": 0.326,
+      "step": 24525
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.7793821199193665e-05,
+      "loss": 0.2897,
+      "step": 24530
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.7793352397918525e-05,
+      "loss": 0.2758,
+      "step": 24535
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7792883596643385e-05,
+      "loss": 0.0219,
+      "step": 24540
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7792414795368245e-05,
+      "loss": 0.1285,
+      "step": 24545
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7791945994093105e-05,
+      "loss": 0.1647,
+      "step": 24550
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7791477192817965e-05,
+      "loss": 0.0803,
+      "step": 24555
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7791008391542825e-05,
+      "loss": 0.0733,
+      "step": 24560
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7790539590267688e-05,
+      "loss": 0.1594,
+      "step": 24565
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7790070788992548e-05,
+      "loss": 0.1789,
+      "step": 24570
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7789601987717408e-05,
+      "loss": 0.2286,
+      "step": 24575
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7789133186442268e-05,
+      "loss": 0.269,
+      "step": 24580
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7788664385167128e-05,
+      "loss": 0.2611,
+      "step": 24585
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.778819558389199e-05,
+      "loss": 0.0783,
+      "step": 24590
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.778772678261685e-05,
+      "loss": 0.0946,
+      "step": 24595
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.778725798134171e-05,
+      "loss": 0.0734,
+      "step": 24600
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.778678918006657e-05,
+      "loss": 0.1414,
+      "step": 24605
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.778632037879143e-05,
+      "loss": 0.2184,
+      "step": 24610
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7785851577516294e-05,
+      "loss": 0.129,
+      "step": 24615
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7785382776241154e-05,
+      "loss": 0.1087,
+      "step": 24620
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7784913974966014e-05,
+      "loss": 0.2552,
+      "step": 24625
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7784445173690874e-05,
+      "loss": 0.2631,
+      "step": 24630
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7783976372415734e-05,
+      "loss": 0.2573,
+      "step": 24635
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7783507571140594e-05,
+      "loss": 0.1247,
+      "step": 24640
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7783038769865454e-05,
+      "loss": 0.0605,
+      "step": 24645
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7782569968590317e-05,
+      "loss": 0.0973,
+      "step": 24650
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7782101167315177e-05,
+      "loss": 0.1361,
+      "step": 24655
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7781632366040037e-05,
+      "loss": 0.1205,
+      "step": 24660
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7781163564764897e-05,
+      "loss": 0.1463,
+      "step": 24665
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.778069476348976e-05,
+      "loss": 0.2136,
+      "step": 24670
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.778022596221462e-05,
+      "loss": 0.2491,
+      "step": 24675
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.777975716093948e-05,
+      "loss": 0.4697,
+      "step": 24680
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.777928835966434e-05,
+      "loss": 0.296,
+      "step": 24685
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.77788195583892e-05,
+      "loss": 0.0517,
+      "step": 24690
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.777835075711406e-05,
+      "loss": 0.0591,
+      "step": 24695
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.777788195583892e-05,
+      "loss": 0.0752,
+      "step": 24700
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7777413154563783e-05,
+      "loss": 0.1123,
+      "step": 24705
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7776944353288643e-05,
+      "loss": 0.2295,
+      "step": 24710
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7776475552013503e-05,
+      "loss": 0.1714,
+      "step": 24715
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7776006750738363e-05,
+      "loss": 0.18,
+      "step": 24720
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7775537949463223e-05,
+      "loss": 0.2655,
+      "step": 24725
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7775069148188086e-05,
+      "loss": 0.2824,
+      "step": 24730
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7774600346912946e-05,
+      "loss": 0.2489,
+      "step": 24735
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7774131545637806e-05,
+      "loss": 0.0462,
+      "step": 24740
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7773662744362666e-05,
+      "loss": 0.0724,
+      "step": 24745
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7773193943087526e-05,
+      "loss": 0.1316,
+      "step": 24750
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.777272514181239e-05,
+      "loss": 0.1558,
+      "step": 24755
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.777225634053725e-05,
+      "loss": 0.1752,
+      "step": 24760
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.777178753926211e-05,
+      "loss": 0.171,
+      "step": 24765
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.777131873798697e-05,
+      "loss": 0.1807,
+      "step": 24770
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.777084993671183e-05,
+      "loss": 0.3059,
+      "step": 24775
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.777038113543669e-05,
+      "loss": 0.3799,
+      "step": 24780
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.776991233416155e-05,
+      "loss": 0.2705,
+      "step": 24785
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.776944353288641e-05,
+      "loss": 0.0925,
+      "step": 24790
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7768974731611272e-05,
+      "loss": 0.1392,
+      "step": 24795
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7768505930336132e-05,
+      "loss": 0.1734,
+      "step": 24800
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7768037129060992e-05,
+      "loss": 0.0997,
+      "step": 24805
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7767568327785855e-05,
+      "loss": 0.1075,
+      "step": 24810
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7767099526510715e-05,
+      "loss": 0.1196,
+      "step": 24815
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7766630725235575e-05,
+      "loss": 0.2809,
+      "step": 24820
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7766161923960435e-05,
+      "loss": 0.2512,
+      "step": 24825
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7765693122685295e-05,
+      "loss": 0.3496,
+      "step": 24830
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7765224321410155e-05,
+      "loss": 0.3198,
+      "step": 24835
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7764755520135015e-05,
+      "loss": 0.0283,
+      "step": 24840
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7764286718859878e-05,
+      "loss": 0.1024,
+      "step": 24845
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7763817917584738e-05,
+      "loss": 0.117,
+      "step": 24850
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7763349116309598e-05,
+      "loss": 0.1471,
+      "step": 24855
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7762880315034458e-05,
+      "loss": 0.1288,
+      "step": 24860
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7762411513759318e-05,
+      "loss": 0.1377,
+      "step": 24865
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7761942712484178e-05,
+      "loss": 0.252,
+      "step": 24870
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.776147391120904e-05,
+      "loss": 0.401,
+      "step": 24875
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.77610051099339e-05,
+      "loss": 0.3525,
+      "step": 24880
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.776053630865876e-05,
+      "loss": 0.187,
+      "step": 24885
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7760067507383624e-05,
+      "loss": 0.0264,
+      "step": 24890
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7759598706108484e-05,
+      "loss": 0.0683,
+      "step": 24895
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7759129904833344e-05,
+      "loss": 0.0647,
+      "step": 24900
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7758661103558204e-05,
+      "loss": 0.1196,
+      "step": 24905
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7758192302283064e-05,
+      "loss": 0.1604,
+      "step": 24910
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7757723501007924e-05,
+      "loss": 0.1187,
+      "step": 24915
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7757254699732784e-05,
+      "loss": 0.2915,
+      "step": 24920
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7756785898457644e-05,
+      "loss": 0.268,
+      "step": 24925
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7756317097182504e-05,
+      "loss": 0.3793,
+      "step": 24930
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7755848295907364e-05,
+      "loss": 0.4032,
+      "step": 24935
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7755379494632227e-05,
+      "loss": 0.1255,
+      "step": 24940
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7754910693357087e-05,
+      "loss": 0.0745,
+      "step": 24945
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7754441892081947e-05,
+      "loss": 0.121,
+      "step": 24950
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.775397309080681e-05,
+      "loss": 0.1057,
+      "step": 24955
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.775350428953167e-05,
+      "loss": 0.1091,
+      "step": 24960
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.775303548825653e-05,
+      "loss": 0.1403,
+      "step": 24965
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.775256668698139e-05,
+      "loss": 0.2515,
+      "step": 24970
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.775209788570625e-05,
+      "loss": 0.3576,
+      "step": 24975
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.775162908443111e-05,
+      "loss": 0.3469,
+      "step": 24980
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7751160283155973e-05,
+      "loss": 0.3998,
+      "step": 24985
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7750691481880833e-05,
+      "loss": 0.1192,
+      "step": 24990
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7750222680605693e-05,
+      "loss": 0.0596,
+      "step": 24995
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7749753879330553e-05,
+      "loss": 0.0592,
+      "step": 25000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7749285078055413e-05,
+      "loss": 0.0797,
+      "step": 25005
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7748816276780273e-05,
+      "loss": 0.1481,
+      "step": 25010
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7748347475505136e-05,
+      "loss": 0.1276,
+      "step": 25015
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7747878674229996e-05,
+      "loss": 0.145,
+      "step": 25020
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7747409872954856e-05,
+      "loss": 0.3773,
+      "step": 25025
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7746941071679716e-05,
+      "loss": 0.3071,
+      "step": 25030
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.774647227040458e-05,
+      "loss": 0.2949,
+      "step": 25035
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.774600346912944e-05,
+      "loss": 0.0725,
+      "step": 25040
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.77455346678543e-05,
+      "loss": 0.0693,
+      "step": 25045
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.774506586657916e-05,
+      "loss": 0.0615,
+      "step": 25050
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.774459706530402e-05,
+      "loss": 0.1353,
+      "step": 25055
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.774412826402888e-05,
+      "loss": 0.1324,
+      "step": 25060
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.774365946275374e-05,
+      "loss": 0.1406,
+      "step": 25065
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.77431906614786e-05,
+      "loss": 0.1729,
+      "step": 25070
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.774272186020346e-05,
+      "loss": 0.2353,
+      "step": 25075
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7742253058928322e-05,
+      "loss": 0.3157,
+      "step": 25080
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7741784257653182e-05,
+      "loss": 0.3223,
+      "step": 25085
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7741315456378042e-05,
+      "loss": 0.0615,
+      "step": 25090
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7740846655102905e-05,
+      "loss": 0.124,
+      "step": 25095
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7740377853827765e-05,
+      "loss": 0.0724,
+      "step": 25100
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7739909052552625e-05,
+      "loss": 0.077,
+      "step": 25105
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7739440251277485e-05,
+      "loss": 0.1978,
+      "step": 25110
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7738971450002345e-05,
+      "loss": 0.1213,
+      "step": 25115
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7738502648727205e-05,
+      "loss": 0.2148,
+      "step": 25120
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7738033847452068e-05,
+      "loss": 0.3011,
+      "step": 25125
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7737565046176928e-05,
+      "loss": 0.3277,
+      "step": 25130
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7737096244901788e-05,
+      "loss": 0.2688,
+      "step": 25135
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7736627443626648e-05,
+      "loss": 0.0598,
+      "step": 25140
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7736158642351508e-05,
+      "loss": 0.041,
+      "step": 25145
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7735689841076368e-05,
+      "loss": 0.1021,
+      "step": 25150
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7735221039801228e-05,
+      "loss": 0.1441,
+      "step": 25155
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.773475223852609e-05,
+      "loss": 0.1063,
+      "step": 25160
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.773428343725095e-05,
+      "loss": 0.109,
+      "step": 25165
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.773381463597581e-05,
+      "loss": 0.2057,
+      "step": 25170
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7733345834700674e-05,
+      "loss": 0.2269,
+      "step": 25175
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7732877033425534e-05,
+      "loss": 0.428,
+      "step": 25180
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7732408232150394e-05,
+      "loss": 0.3289,
+      "step": 25185
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7731939430875254e-05,
+      "loss": 0.0549,
+      "step": 25190
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7731470629600114e-05,
+      "loss": 0.1167,
+      "step": 25195
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7731001828324974e-05,
+      "loss": 0.0702,
+      "step": 25200
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7730533027049834e-05,
+      "loss": 0.1164,
+      "step": 25205
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7730064225774694e-05,
+      "loss": 0.1205,
+      "step": 25210
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7729595424499557e-05,
+      "loss": 0.1234,
+      "step": 25215
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7729126623224417e-05,
+      "loss": 0.1769,
+      "step": 25220
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7728657821949277e-05,
+      "loss": 0.2162,
+      "step": 25225
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7728189020674137e-05,
+      "loss": 0.3294,
+      "step": 25230
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7727720219398997e-05,
+      "loss": 0.2073,
+      "step": 25235
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.772725141812386e-05,
+      "loss": 0.0587,
+      "step": 25240
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.772678261684872e-05,
+      "loss": 0.0679,
+      "step": 25245
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.772631381557358e-05,
+      "loss": 0.1205,
+      "step": 25250
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.772584501429844e-05,
+      "loss": 0.0763,
+      "step": 25255
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.77253762130233e-05,
+      "loss": 0.1503,
+      "step": 25260
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7724907411748163e-05,
+      "loss": 0.208,
+      "step": 25265
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7724438610473023e-05,
+      "loss": 0.1812,
+      "step": 25270
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7723969809197883e-05,
+      "loss": 0.2399,
+      "step": 25275
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7723501007922743e-05,
+      "loss": 0.276,
+      "step": 25280
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7723032206647603e-05,
+      "loss": 0.3312,
+      "step": 25285
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7722563405372463e-05,
+      "loss": 0.0309,
+      "step": 25290
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7722094604097323e-05,
+      "loss": 0.0898,
+      "step": 25295
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7721625802822183e-05,
+      "loss": 0.1028,
+      "step": 25300
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7721157001547046e-05,
+      "loss": 0.1351,
+      "step": 25305
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7720688200271906e-05,
+      "loss": 0.1858,
+      "step": 25310
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7720219398996766e-05,
+      "loss": 0.0857,
+      "step": 25315
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.771975059772163e-05,
+      "loss": 0.2156,
+      "step": 25320
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.771928179644649e-05,
+      "loss": 0.3597,
+      "step": 25325
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.771881299517135e-05,
+      "loss": 0.2196,
+      "step": 25330
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.771834419389621e-05,
+      "loss": 0.2814,
+      "step": 25335
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.771787539262107e-05,
+      "loss": 0.0829,
+      "step": 25340
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.771740659134593e-05,
+      "loss": 0.0587,
+      "step": 25345
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.771693779007079e-05,
+      "loss": 0.1198,
+      "step": 25350
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7716468988795652e-05,
+      "loss": 0.1103,
+      "step": 25355
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7716000187520512e-05,
+      "loss": 0.0928,
+      "step": 25360
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7715531386245372e-05,
+      "loss": 0.1933,
+      "step": 25365
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7715062584970232e-05,
+      "loss": 0.246,
+      "step": 25370
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7714593783695092e-05,
+      "loss": 0.2511,
+      "step": 25375
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7714124982419952e-05,
+      "loss": 0.3878,
+      "step": 25380
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7713656181144815e-05,
+      "loss": 0.2689,
+      "step": 25385
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7713187379869675e-05,
+      "loss": 0.0576,
+      "step": 25390
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7712718578594535e-05,
+      "loss": 0.0802,
+      "step": 25395
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7712249777319395e-05,
+      "loss": 0.1151,
+      "step": 25400
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.771178097604426e-05,
+      "loss": 0.1614,
+      "step": 25405
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7711312174769118e-05,
+      "loss": 0.2646,
+      "step": 25410
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7710843373493978e-05,
+      "loss": 0.1363,
+      "step": 25415
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7710374572218838e-05,
+      "loss": 0.271,
+      "step": 25420
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7709905770943698e-05,
+      "loss": 0.2105,
+      "step": 25425
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7709436969668558e-05,
+      "loss": 0.4278,
+      "step": 25430
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7708968168393418e-05,
+      "loss": 0.304,
+      "step": 25435
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7708499367118278e-05,
+      "loss": 0.096,
+      "step": 25440
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.770803056584314e-05,
+      "loss": 0.0942,
+      "step": 25445
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7707561764568e-05,
+      "loss": 0.0986,
+      "step": 25450
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.770709296329286e-05,
+      "loss": 0.1379,
+      "step": 25455
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.770662416201772e-05,
+      "loss": 0.1218,
+      "step": 25460
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7706155360742584e-05,
+      "loss": 0.172,
+      "step": 25465
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7705686559467444e-05,
+      "loss": 0.175,
+      "step": 25470
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7705217758192304e-05,
+      "loss": 0.1812,
+      "step": 25475
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7704748956917164e-05,
+      "loss": 0.4416,
+      "step": 25480
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7704280155642024e-05,
+      "loss": 0.3549,
+      "step": 25485
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7703811354366884e-05,
+      "loss": 0.0537,
+      "step": 25490
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7703342553091747e-05,
+      "loss": 0.0699,
+      "step": 25495
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7702873751816607e-05,
+      "loss": 0.1461,
+      "step": 25500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7702404950541467e-05,
+      "loss": 0.181,
+      "step": 25505
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7701936149266327e-05,
+      "loss": 0.2071,
+      "step": 25510
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7701467347991187e-05,
+      "loss": 0.1548,
+      "step": 25515
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7700998546716047e-05,
+      "loss": 0.2228,
+      "step": 25520
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.770052974544091e-05,
+      "loss": 0.232,
+      "step": 25525
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.770006094416577e-05,
+      "loss": 0.3807,
+      "step": 25530
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.769959214289063e-05,
+      "loss": 0.3537,
+      "step": 25535
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.769912334161549e-05,
+      "loss": 0.0648,
+      "step": 25540
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7698654540340353e-05,
+      "loss": 0.0903,
+      "step": 25545
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7698185739065213e-05,
+      "loss": 0.1323,
+      "step": 25550
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7697716937790073e-05,
+      "loss": 0.106,
+      "step": 25555
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7697248136514933e-05,
+      "loss": 0.0727,
+      "step": 25560
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7696779335239793e-05,
+      "loss": 0.1354,
+      "step": 25565
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7696310533964653e-05,
+      "loss": 0.2474,
+      "step": 25570
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7695841732689513e-05,
+      "loss": 0.2366,
+      "step": 25575
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7695372931414373e-05,
+      "loss": 0.3618,
+      "step": 25580
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7694904130139233e-05,
+      "loss": 0.294,
+      "step": 25585
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7694435328864096e-05,
+      "loss": 0.0909,
+      "step": 25590
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7693966527588956e-05,
+      "loss": 0.0789,
+      "step": 25595
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7693497726313816e-05,
+      "loss": 0.1543,
+      "step": 25600
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.769302892503868e-05,
+      "loss": 0.1547,
+      "step": 25605
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.769256012376354e-05,
+      "loss": 0.1539,
+      "step": 25610
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.76920913224884e-05,
+      "loss": 0.1479,
+      "step": 25615
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.769162252121326e-05,
+      "loss": 0.2259,
+      "step": 25620
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.769115371993812e-05,
+      "loss": 0.237,
+      "step": 25625
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.769068491866298e-05,
+      "loss": 0.2004,
+      "step": 25630
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7690216117387842e-05,
+      "loss": 0.4021,
+      "step": 25635
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7689747316112702e-05,
+      "loss": 0.102,
+      "step": 25640
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7689278514837562e-05,
+      "loss": 0.0637,
+      "step": 25645
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7688809713562422e-05,
+      "loss": 0.0587,
+      "step": 25650
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7688340912287282e-05,
+      "loss": 0.112,
+      "step": 25655
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7687872111012142e-05,
+      "loss": 0.1684,
+      "step": 25660
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7687403309737002e-05,
+      "loss": 0.2316,
+      "step": 25665
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7686934508461865e-05,
+      "loss": 0.2574,
+      "step": 25670
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7686465707186725e-05,
+      "loss": 0.1927,
+      "step": 25675
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7685996905911585e-05,
+      "loss": 0.3456,
+      "step": 25680
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.768552810463645e-05,
+      "loss": 0.2925,
+      "step": 25685
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.768505930336131e-05,
+      "loss": 0.056,
+      "step": 25690
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.768459050208617e-05,
+      "loss": 0.1489,
+      "step": 25695
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7684121700811028e-05,
+      "loss": 0.0954,
+      "step": 25700
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7683652899535888e-05,
+      "loss": 0.079,
+      "step": 25705
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7683184098260748e-05,
+      "loss": 0.0937,
+      "step": 25710
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7682715296985608e-05,
+      "loss": 0.1415,
+      "step": 25715
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7682246495710468e-05,
+      "loss": 0.1691,
+      "step": 25720
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7681777694435328e-05,
+      "loss": 0.3064,
+      "step": 25725
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.768130889316019e-05,
+      "loss": 0.3589,
+      "step": 25730
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.768084009188505e-05,
+      "loss": 0.2006,
+      "step": 25735
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.768037129060991e-05,
+      "loss": 0.0595,
+      "step": 25740
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.767990248933477e-05,
+      "loss": 0.0753,
+      "step": 25745
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7679433688059634e-05,
+      "loss": 0.0849,
+      "step": 25750
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7678964886784494e-05,
+      "loss": 0.1281,
+      "step": 25755
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7678496085509354e-05,
+      "loss": 0.1522,
+      "step": 25760
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7678027284234214e-05,
+      "loss": 0.0967,
+      "step": 25765
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7677558482959074e-05,
+      "loss": 0.1523,
+      "step": 25770
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7677089681683937e-05,
+      "loss": 0.1729,
+      "step": 25775
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7676620880408797e-05,
+      "loss": 0.4448,
+      "step": 25780
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7676152079133657e-05,
+      "loss": 0.3308,
+      "step": 25785
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7675683277858517e-05,
+      "loss": 0.0479,
+      "step": 25790
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7675214476583377e-05,
+      "loss": 0.0691,
+      "step": 25795
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7674745675308237e-05,
+      "loss": 0.0718,
+      "step": 25800
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7674276874033097e-05,
+      "loss": 0.1133,
+      "step": 25805
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7673808072757957e-05,
+      "loss": 0.1421,
+      "step": 25810
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.767333927148282e-05,
+      "loss": 0.2539,
+      "step": 25815
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.767287047020768e-05,
+      "loss": 0.2135,
+      "step": 25820
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.767240166893254e-05,
+      "loss": 0.311,
+      "step": 25825
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7671932867657404e-05,
+      "loss": 0.3515,
+      "step": 25830
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7671464066382263e-05,
+      "loss": 0.2823,
+      "step": 25835
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7670995265107123e-05,
+      "loss": 0.0406,
+      "step": 25840
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7670526463831983e-05,
+      "loss": 0.0476,
+      "step": 25845
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7670057662556843e-05,
+      "loss": 0.158,
+      "step": 25850
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7669588861281703e-05,
+      "loss": 0.1091,
+      "step": 25855
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7669120060006563e-05,
+      "loss": 0.1941,
+      "step": 25860
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7668651258731426e-05,
+      "loss": 0.1813,
+      "step": 25865
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7668182457456286e-05,
+      "loss": 0.2163,
+      "step": 25870
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7667713656181146e-05,
+      "loss": 0.2309,
+      "step": 25875
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7667244854906006e-05,
+      "loss": 0.3899,
+      "step": 25880
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7666776053630866e-05,
+      "loss": 0.409,
+      "step": 25885
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7666307252355726e-05,
+      "loss": 0.0603,
+      "step": 25890
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.766583845108059e-05,
+      "loss": 0.0614,
+      "step": 25895
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.766536964980545e-05,
+      "loss": 0.0755,
+      "step": 25900
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.766490084853031e-05,
+      "loss": 0.0907,
+      "step": 25905
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.766443204725517e-05,
+      "loss": 0.2172,
+      "step": 25910
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7663963245980033e-05,
+      "loss": 0.1953,
+      "step": 25915
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7663494444704892e-05,
+      "loss": 0.2066,
+      "step": 25920
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7663025643429752e-05,
+      "loss": 0.2019,
+      "step": 25925
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7662556842154612e-05,
+      "loss": 0.2587,
+      "step": 25930
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7662088040879472e-05,
+      "loss": 0.328,
+      "step": 25935
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7661619239604332e-05,
+      "loss": 0.0652,
+      "step": 25940
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7661150438329192e-05,
+      "loss": 0.1316,
+      "step": 25945
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7660681637054052e-05,
+      "loss": 0.1019,
+      "step": 25950
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7660212835778915e-05,
+      "loss": 0.1534,
+      "step": 25955
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7659744034503775e-05,
+      "loss": 0.1238,
+      "step": 25960
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7659275233228635e-05,
+      "loss": 0.1582,
+      "step": 25965
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7658806431953495e-05,
+      "loss": 0.2585,
+      "step": 25970
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.765833763067836e-05,
+      "loss": 0.3096,
+      "step": 25975
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.765786882940322e-05,
+      "loss": 0.4421,
+      "step": 25980
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.765740002812808e-05,
+      "loss": 0.2735,
+      "step": 25985
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.765693122685294e-05,
+      "loss": 0.0522,
+      "step": 25990
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7656462425577798e-05,
+      "loss": 0.0711,
+      "step": 25995
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7655993624302658e-05,
+      "loss": 0.0588,
+      "step": 26000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.765552482302752e-05,
+      "loss": 0.1255,
+      "step": 26005
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.765505602175238e-05,
+      "loss": 0.0974,
+      "step": 26010
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.765458722047724e-05,
+      "loss": 0.162,
+      "step": 26015
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.76541184192021e-05,
+      "loss": 0.2493,
+      "step": 26020
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.765364961792696e-05,
+      "loss": 0.228,
+      "step": 26025
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.765318081665182e-05,
+      "loss": 0.3248,
+      "step": 26030
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7652712015376684e-05,
+      "loss": 0.3221,
+      "step": 26035
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7652243214101544e-05,
+      "loss": 0.0706,
+      "step": 26040
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7651774412826404e-05,
+      "loss": 0.1151,
+      "step": 26045
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7651305611551264e-05,
+      "loss": 0.0961,
+      "step": 26050
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7650836810276128e-05,
+      "loss": 0.1412,
+      "step": 26055
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7650368009000988e-05,
+      "loss": 0.1789,
+      "step": 26060
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7649899207725847e-05,
+      "loss": 0.1407,
+      "step": 26065
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7649430406450707e-05,
+      "loss": 0.2843,
+      "step": 26070
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7648961605175567e-05,
+      "loss": 0.2238,
+      "step": 26075
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7648492803900427e-05,
+      "loss": 0.4524,
+      "step": 26080
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7648024002625287e-05,
+      "loss": 0.2933,
+      "step": 26085
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7647555201350147e-05,
+      "loss": 0.0408,
+      "step": 26090
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7647086400075007e-05,
+      "loss": 0.1177,
+      "step": 26095
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.764661759879987e-05,
+      "loss": 0.1175,
+      "step": 26100
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.764614879752473e-05,
+      "loss": 0.1405,
+      "step": 26105
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.764567999624959e-05,
+      "loss": 0.1418,
+      "step": 26110
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7645211194974454e-05,
+      "loss": 0.1756,
+      "step": 26115
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7644742393699314e-05,
+      "loss": 0.1311,
+      "step": 26120
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7644273592424173e-05,
+      "loss": 0.2413,
+      "step": 26125
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7643804791149033e-05,
+      "loss": 0.2902,
+      "step": 26130
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7643335989873893e-05,
+      "loss": 0.4399,
+      "step": 26135
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7642867188598753e-05,
+      "loss": 0.0452,
+      "step": 26140
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7642398387323617e-05,
+      "loss": 0.0868,
+      "step": 26145
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7641929586048477e-05,
+      "loss": 0.0898,
+      "step": 26150
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7641460784773336e-05,
+      "loss": 0.1588,
+      "step": 26155
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7640991983498196e-05,
+      "loss": 0.1818,
+      "step": 26160
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7640523182223056e-05,
+      "loss": 0.241,
+      "step": 26165
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7640054380947916e-05,
+      "loss": 0.1719,
+      "step": 26170
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7639585579672776e-05,
+      "loss": 0.219,
+      "step": 26175
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.763911677839764e-05,
+      "loss": 0.3367,
+      "step": 26180
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.76386479771225e-05,
+      "loss": 0.2374,
+      "step": 26185
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.763817917584736e-05,
+      "loss": 0.0655,
+      "step": 26190
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7637710374572223e-05,
+      "loss": 0.0818,
+      "step": 26195
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7637241573297083e-05,
+      "loss": 0.1566,
+      "step": 26200
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7636772772021943e-05,
+      "loss": 0.1525,
+      "step": 26205
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7636303970746803e-05,
+      "loss": 0.1549,
+      "step": 26210
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7635835169471662e-05,
+      "loss": 0.2122,
+      "step": 26215
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7635366368196522e-05,
+      "loss": 0.1392,
+      "step": 26220
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7634897566921382e-05,
+      "loss": 0.3111,
+      "step": 26225
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7634428765646242e-05,
+      "loss": 0.4944,
+      "step": 26230
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7633959964371102e-05,
+      "loss": 0.3476,
+      "step": 26235
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7633491163095965e-05,
+      "loss": 0.0758,
+      "step": 26240
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7633022361820825e-05,
+      "loss": 0.1091,
+      "step": 26245
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7632553560545685e-05,
+      "loss": 0.1075,
+      "step": 26250
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7632084759270545e-05,
+      "loss": 0.0975,
+      "step": 26255
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.763161595799541e-05,
+      "loss": 0.1404,
+      "step": 26260
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.763114715672027e-05,
+      "loss": 0.1439,
+      "step": 26265
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.763067835544513e-05,
+      "loss": 0.1963,
+      "step": 26270
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.763020955416999e-05,
+      "loss": 0.2833,
+      "step": 26275
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.762974075289485e-05,
+      "loss": 0.4187,
+      "step": 26280
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.762927195161971e-05,
+      "loss": 0.3837,
+      "step": 26285
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.762880315034457e-05,
+      "loss": 0.0572,
+      "step": 26290
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.762833434906943e-05,
+      "loss": 0.1154,
+      "step": 26295
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.762786554779429e-05,
+      "loss": 0.1305,
+      "step": 26300
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.762739674651915e-05,
+      "loss": 0.0764,
+      "step": 26305
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.762692794524401e-05,
+      "loss": 0.172,
+      "step": 26310
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.762645914396887e-05,
+      "loss": 0.2716,
+      "step": 26315
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.762599034269373e-05,
+      "loss": 0.1888,
+      "step": 26320
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7625521541418595e-05,
+      "loss": 0.2317,
+      "step": 26325
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7625052740143454e-05,
+      "loss": 0.3846,
+      "step": 26330
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7624583938868314e-05,
+      "loss": 0.3627,
+      "step": 26335
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7624115137593178e-05,
+      "loss": 0.0344,
+      "step": 26340
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7623646336318038e-05,
+      "loss": 0.0382,
+      "step": 26345
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7623177535042898e-05,
+      "loss": 0.0994,
+      "step": 26350
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7622708733767758e-05,
+      "loss": 0.0887,
+      "step": 26355
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7622239932492617e-05,
+      "loss": 0.107,
+      "step": 26360
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7621771131217477e-05,
+      "loss": 0.2014,
+      "step": 26365
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7621302329942337e-05,
+      "loss": 0.1593,
+      "step": 26370
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7620833528667197e-05,
+      "loss": 0.2043,
+      "step": 26375
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.762036472739206e-05,
+      "loss": 0.2195,
+      "step": 26380
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.761989592611692e-05,
+      "loss": 0.2368,
+      "step": 26385
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.761942712484178e-05,
+      "loss": 0.0934,
+      "step": 26390
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.761895832356664e-05,
+      "loss": 0.1525,
+      "step": 26395
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.76184895222915e-05,
+      "loss": 0.1338,
+      "step": 26400
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7618020721016364e-05,
+      "loss": 0.1834,
+      "step": 26405
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7617551919741224e-05,
+      "loss": 0.1386,
+      "step": 26410
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7617083118466084e-05,
+      "loss": 0.2247,
+      "step": 26415
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7616614317190943e-05,
+      "loss": 0.2742,
+      "step": 26420
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7616145515915807e-05,
+      "loss": 0.2639,
+      "step": 26425
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7615676714640667e-05,
+      "loss": 0.2932,
+      "step": 26430
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7615207913365527e-05,
+      "loss": 0.2663,
+      "step": 26435
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7614739112090387e-05,
+      "loss": 0.0708,
+      "step": 26440
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7614270310815246e-05,
+      "loss": 0.0757,
+      "step": 26445
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7613801509540106e-05,
+      "loss": 0.0615,
+      "step": 26450
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7613332708264966e-05,
+      "loss": 0.1459,
+      "step": 26455
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7612863906989826e-05,
+      "loss": 0.1307,
+      "step": 26460
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.761239510571469e-05,
+      "loss": 0.1669,
+      "step": 26465
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.761192630443955e-05,
+      "loss": 0.2281,
+      "step": 26470
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.761145750316441e-05,
+      "loss": 0.2489,
+      "step": 26475
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7610988701889273e-05,
+      "loss": 0.3256,
+      "step": 26480
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7610519900614133e-05,
+      "loss": 0.2399,
+      "step": 26485
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7610051099338993e-05,
+      "loss": 0.0681,
+      "step": 26490
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7609582298063853e-05,
+      "loss": 0.1018,
+      "step": 26495
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7609113496788713e-05,
+      "loss": 0.09,
+      "step": 26500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7608644695513572e-05,
+      "loss": 0.1208,
+      "step": 26505
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7608175894238432e-05,
+      "loss": 0.1643,
+      "step": 26510
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7607707092963296e-05,
+      "loss": 0.2492,
+      "step": 26515
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7607238291688156e-05,
+      "loss": 0.2035,
+      "step": 26520
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7606769490413016e-05,
+      "loss": 0.1378,
+      "step": 26525
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7606300689137876e-05,
+      "loss": 0.3092,
+      "step": 26530
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7605831887862735e-05,
+      "loss": 0.2494,
+      "step": 26535
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7605363086587595e-05,
+      "loss": 0.0539,
+      "step": 26540
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.760489428531246e-05,
+      "loss": 0.1035,
+      "step": 26545
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.760442548403732e-05,
+      "loss": 0.079,
+      "step": 26550
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.760395668276218e-05,
+      "loss": 0.0964,
+      "step": 26555
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.760348788148704e-05,
+      "loss": 0.1856,
+      "step": 26560
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7603019080211902e-05,
+      "loss": 0.1496,
+      "step": 26565
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7602550278936762e-05,
+      "loss": 0.1282,
+      "step": 26570
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7602081477661622e-05,
+      "loss": 0.2813,
+      "step": 26575
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.760161267638648e-05,
+      "loss": 0.3293,
+      "step": 26580
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.760114387511134e-05,
+      "loss": 0.1481,
+      "step": 26585
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.76006750738362e-05,
+      "loss": 0.0804,
+      "step": 26590
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.760020627256106e-05,
+      "loss": 0.0621,
+      "step": 26595
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.759973747128592e-05,
+      "loss": 0.1511,
+      "step": 26600
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.759926867001078e-05,
+      "loss": 0.1558,
+      "step": 26605
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7598799868735645e-05,
+      "loss": 0.1994,
+      "step": 26610
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7598331067460505e-05,
+      "loss": 0.1426,
+      "step": 26615
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7597862266185364e-05,
+      "loss": 0.2883,
+      "step": 26620
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7597393464910228e-05,
+      "loss": 0.1739,
+      "step": 26625
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7596924663635088e-05,
+      "loss": 0.3145,
+      "step": 26630
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7596455862359948e-05,
+      "loss": 0.3456,
+      "step": 26635
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7595987061084808e-05,
+      "loss": 0.0483,
+      "step": 26640
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7595518259809668e-05,
+      "loss": 0.0308,
+      "step": 26645
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7595049458534527e-05,
+      "loss": 0.0719,
+      "step": 26650
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.759458065725939e-05,
+      "loss": 0.0575,
+      "step": 26655
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.759411185598425e-05,
+      "loss": 0.123,
+      "step": 26660
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.759364305470911e-05,
+      "loss": 0.17,
+      "step": 26665
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.759317425343397e-05,
+      "loss": 0.2294,
+      "step": 26670
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.759270545215883e-05,
+      "loss": 0.357,
+      "step": 26675
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.759223665088369e-05,
+      "loss": 0.32,
+      "step": 26680
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.759176784960855e-05,
+      "loss": 0.2794,
+      "step": 26685
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7591299048333414e-05,
+      "loss": 0.0375,
+      "step": 26690
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7590830247058274e-05,
+      "loss": 0.0531,
+      "step": 26695
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7590361445783134e-05,
+      "loss": 0.1184,
+      "step": 26700
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7589892644507997e-05,
+      "loss": 0.0585,
+      "step": 26705
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7589423843232857e-05,
+      "loss": 0.1184,
+      "step": 26710
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7588955041957717e-05,
+      "loss": 0.2323,
+      "step": 26715
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7588486240682577e-05,
+      "loss": 0.174,
+      "step": 26720
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7588017439407437e-05,
+      "loss": 0.2886,
+      "step": 26725
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7587548638132297e-05,
+      "loss": 0.3716,
+      "step": 26730
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7587079836857157e-05,
+      "loss": 0.3872,
+      "step": 26735
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7586611035582016e-05,
+      "loss": 0.0421,
+      "step": 26740
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7586142234306876e-05,
+      "loss": 0.1046,
+      "step": 26745
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.758567343303174e-05,
+      "loss": 0.081,
+      "step": 26750
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.75852046317566e-05,
+      "loss": 0.0735,
+      "step": 26755
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.758473583048146e-05,
+      "loss": 0.1882,
+      "step": 26760
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.758426702920632e-05,
+      "loss": 0.1912,
+      "step": 26765
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7583798227931183e-05,
+      "loss": 0.1503,
+      "step": 26770
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7583329426656043e-05,
+      "loss": 0.2477,
+      "step": 26775
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7582860625380903e-05,
+      "loss": 0.3636,
+      "step": 26780
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7582391824105763e-05,
+      "loss": 0.3664,
+      "step": 26785
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7581923022830623e-05,
+      "loss": 0.0495,
+      "step": 26790
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7581454221555486e-05,
+      "loss": 0.1003,
+      "step": 26795
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7580985420280346e-05,
+      "loss": 0.0936,
+      "step": 26800
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7580516619005206e-05,
+      "loss": 0.1304,
+      "step": 26805
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7580047817730066e-05,
+      "loss": 0.1174,
+      "step": 26810
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7579579016454926e-05,
+      "loss": 0.1183,
+      "step": 26815
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7579110215179786e-05,
+      "loss": 0.3376,
+      "step": 26820
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7578641413904645e-05,
+      "loss": 0.212,
+      "step": 26825
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.757817261262951e-05,
+      "loss": 0.3103,
+      "step": 26830
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.757770381135437e-05,
+      "loss": 0.204,
+      "step": 26835
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.757723501007923e-05,
+      "loss": 0.0467,
+      "step": 26840
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.757676620880409e-05,
+      "loss": 0.0806,
+      "step": 26845
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7576297407528952e-05,
+      "loss": 0.1039,
+      "step": 26850
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7575828606253812e-05,
+      "loss": 0.1226,
+      "step": 26855
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7575359804978672e-05,
+      "loss": 0.0761,
+      "step": 26860
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7574891003703532e-05,
+      "loss": 0.197,
+      "step": 26865
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.757442220242839e-05,
+      "loss": 0.2467,
+      "step": 26870
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.757395340115325e-05,
+      "loss": 0.303,
+      "step": 26875
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.757348459987811e-05,
+      "loss": 0.4572,
+      "step": 26880
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.757301579860297e-05,
+      "loss": 0.3963,
+      "step": 26885
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7572546997327835e-05,
+      "loss": 0.0959,
+      "step": 26890
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7572078196052695e-05,
+      "loss": 0.147,
+      "step": 26895
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7571609394777555e-05,
+      "loss": 0.078,
+      "step": 26900
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7571140593502415e-05,
+      "loss": 0.1146,
+      "step": 26905
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7570671792227278e-05,
+      "loss": 0.1608,
+      "step": 26910
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7570202990952138e-05,
+      "loss": 0.1956,
+      "step": 26915
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7569734189676998e-05,
+      "loss": 0.2593,
+      "step": 26920
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7569265388401858e-05,
+      "loss": 0.3484,
+      "step": 26925
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7568796587126718e-05,
+      "loss": 0.2616,
+      "step": 26930
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.756832778585158e-05,
+      "loss": 0.3386,
+      "step": 26935
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.756785898457644e-05,
+      "loss": 0.0567,
+      "step": 26940
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.75673901833013e-05,
+      "loss": 0.0479,
+      "step": 26945
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.756692138202616e-05,
+      "loss": 0.1017,
+      "step": 26950
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.756645258075102e-05,
+      "loss": 0.1064,
+      "step": 26955
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.756598377947588e-05,
+      "loss": 0.1581,
+      "step": 26960
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.756551497820074e-05,
+      "loss": 0.0933,
+      "step": 26965
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.75650461769256e-05,
+      "loss": 0.1222,
+      "step": 26970
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7564577375650464e-05,
+      "loss": 0.275,
+      "step": 26975
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7564108574375324e-05,
+      "loss": 0.3865,
+      "step": 26980
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7563639773100184e-05,
+      "loss": 0.3438,
+      "step": 26985
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7563170971825047e-05,
+      "loss": 0.0642,
+      "step": 26990
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7562702170549907e-05,
+      "loss": 0.0703,
+      "step": 26995
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7562233369274767e-05,
+      "loss": 0.0959,
+      "step": 27000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7561764567999627e-05,
+      "loss": 0.0822,
+      "step": 27005
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7561295766724487e-05,
+      "loss": 0.1104,
+      "step": 27010
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7560826965449347e-05,
+      "loss": 0.1466,
+      "step": 27015
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7560358164174207e-05,
+      "loss": 0.2263,
+      "step": 27020
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7559889362899067e-05,
+      "loss": 0.3172,
+      "step": 27025
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.755942056162393e-05,
+      "loss": 0.3317,
+      "step": 27030
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.755895176034879e-05,
+      "loss": 0.3048,
+      "step": 27035
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.755848295907365e-05,
+      "loss": 0.0425,
+      "step": 27040
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.755801415779851e-05,
+      "loss": 0.077,
+      "step": 27045
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.755754535652337e-05,
+      "loss": 0.1112,
+      "step": 27050
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7557076555248233e-05,
+      "loss": 0.0855,
+      "step": 27055
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7556607753973093e-05,
+      "loss": 0.1392,
+      "step": 27060
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7556138952697953e-05,
+      "loss": 0.1075,
+      "step": 27065
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7555670151422813e-05,
+      "loss": 0.2517,
+      "step": 27070
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7555201350147676e-05,
+      "loss": 0.3461,
+      "step": 27075
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7554732548872536e-05,
+      "loss": 0.4033,
+      "step": 27080
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7554263747597396e-05,
+      "loss": 0.3499,
+      "step": 27085
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7553794946322256e-05,
+      "loss": 0.0371,
+      "step": 27090
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7553326145047116e-05,
+      "loss": 0.0903,
+      "step": 27095
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7552857343771976e-05,
+      "loss": 0.0927,
+      "step": 27100
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7552388542496836e-05,
+      "loss": 0.1295,
+      "step": 27105
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7551919741221696e-05,
+      "loss": 0.1342,
+      "step": 27110
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7551450939946556e-05,
+      "loss": 0.2217,
+      "step": 27115
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.755098213867142e-05,
+      "loss": 0.1572,
+      "step": 27120
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.755051333739628e-05,
+      "loss": 0.2619,
+      "step": 27125
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.755004453612114e-05,
+      "loss": 0.303,
+      "step": 27130
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7549575734846002e-05,
+      "loss": 0.3301,
+      "step": 27135
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7549106933570862e-05,
+      "loss": 0.0658,
+      "step": 27140
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7548638132295722e-05,
+      "loss": 0.0525,
+      "step": 27145
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7548169331020582e-05,
+      "loss": 0.1088,
+      "step": 27150
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7547700529745442e-05,
+      "loss": 0.1151,
+      "step": 27155
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7547231728470302e-05,
+      "loss": 0.1102,
+      "step": 27160
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7546762927195165e-05,
+      "loss": 0.1528,
+      "step": 27165
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7546294125920025e-05,
+      "loss": 0.1441,
+      "step": 27170
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7545825324644885e-05,
+      "loss": 0.2677,
+      "step": 27175
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7545356523369745e-05,
+      "loss": 0.3605,
+      "step": 27180
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7544887722094605e-05,
+      "loss": 0.2491,
+      "step": 27185
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7544418920819465e-05,
+      "loss": 0.0406,
+      "step": 27190
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7543950119544325e-05,
+      "loss": 0.0386,
+      "step": 27195
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7543481318269188e-05,
+      "loss": 0.0862,
+      "step": 27200
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7543012516994048e-05,
+      "loss": 0.1498,
+      "step": 27205
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7542543715718908e-05,
+      "loss": 0.098,
+      "step": 27210
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.754207491444377e-05,
+      "loss": 0.1449,
+      "step": 27215
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.754160611316863e-05,
+      "loss": 0.1432,
+      "step": 27220
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.754113731189349e-05,
+      "loss": 0.2482,
+      "step": 27225
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.754066851061835e-05,
+      "loss": 0.5192,
+      "step": 27230
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.754019970934321e-05,
+      "loss": 0.2192,
+      "step": 27235
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.753973090806807e-05,
+      "loss": 0.0538,
+      "step": 27240
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.753926210679293e-05,
+      "loss": 0.0637,
+      "step": 27245
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.753879330551779e-05,
+      "loss": 0.0584,
+      "step": 27250
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.753832450424265e-05,
+      "loss": 0.1314,
+      "step": 27255
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7537855702967514e-05,
+      "loss": 0.0987,
+      "step": 27260
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7537386901692374e-05,
+      "loss": 0.101,
+      "step": 27265
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7536918100417234e-05,
+      "loss": 0.117,
+      "step": 27270
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7536449299142094e-05,
+      "loss": 0.2577,
+      "step": 27275
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7535980497866957e-05,
+      "loss": 0.3543,
+      "step": 27280
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7535511696591817e-05,
+      "loss": 0.2811,
+      "step": 27285
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7535042895316677e-05,
+      "loss": 0.0357,
+      "step": 27290
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7534574094041537e-05,
+      "loss": 0.0465,
+      "step": 27295
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7534105292766397e-05,
+      "loss": 0.1089,
+      "step": 27300
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.753363649149126e-05,
+      "loss": 0.1224,
+      "step": 27305
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.753316769021612e-05,
+      "loss": 0.1423,
+      "step": 27310
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.753269888894098e-05,
+      "loss": 0.1598,
+      "step": 27315
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.753223008766584e-05,
+      "loss": 0.1652,
+      "step": 27320
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.75317612863907e-05,
+      "loss": 0.2521,
+      "step": 27325
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.753129248511556e-05,
+      "loss": 0.3785,
+      "step": 27330
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.753082368384042e-05,
+      "loss": 0.3144,
+      "step": 27335
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7530354882565283e-05,
+      "loss": 0.0735,
+      "step": 27340
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7529886081290143e-05,
+      "loss": 0.0404,
+      "step": 27345
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7529417280015003e-05,
+      "loss": 0.1386,
+      "step": 27350
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7528948478739863e-05,
+      "loss": 0.0787,
+      "step": 27355
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7528479677464726e-05,
+      "loss": 0.1407,
+      "step": 27360
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7528010876189586e-05,
+      "loss": 0.1503,
+      "step": 27365
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7527542074914446e-05,
+      "loss": 0.2271,
+      "step": 27370
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7527073273639306e-05,
+      "loss": 0.2996,
+      "step": 27375
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7526604472364166e-05,
+      "loss": 0.4044,
+      "step": 27380
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7526135671089026e-05,
+      "loss": 0.1628,
+      "step": 27385
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7525666869813886e-05,
+      "loss": 0.0612,
+      "step": 27390
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7525198068538746e-05,
+      "loss": 0.0614,
+      "step": 27395
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.752472926726361e-05,
+      "loss": 0.0768,
+      "step": 27400
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.752426046598847e-05,
+      "loss": 0.1372,
+      "step": 27405
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.752379166471333e-05,
+      "loss": 0.1258,
+      "step": 27410
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.752332286343819e-05,
+      "loss": 0.1211,
+      "step": 27415
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7522854062163052e-05,
+      "loss": 0.1808,
+      "step": 27420
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7522385260887912e-05,
+      "loss": 0.3216,
+      "step": 27425
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7521916459612772e-05,
+      "loss": 0.3866,
+      "step": 27430
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7521447658337632e-05,
+      "loss": 0.214,
+      "step": 27435
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7520978857062492e-05,
+      "loss": 0.0791,
+      "step": 27440
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7520510055787355e-05,
+      "loss": 0.0851,
+      "step": 27445
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7520041254512215e-05,
+      "loss": 0.1055,
+      "step": 27450
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7519572453237075e-05,
+      "loss": 0.0761,
+      "step": 27455
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7519103651961935e-05,
+      "loss": 0.1543,
+      "step": 27460
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7518634850686795e-05,
+      "loss": 0.1972,
+      "step": 27465
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7518166049411655e-05,
+      "loss": 0.2102,
+      "step": 27470
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7517697248136515e-05,
+      "loss": 0.3063,
+      "step": 27475
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7517228446861375e-05,
+      "loss": 0.4658,
+      "step": 27480
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7516759645586238e-05,
+      "loss": 0.2809,
+      "step": 27485
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7516290844311098e-05,
+      "loss": 0.0563,
+      "step": 27490
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7515822043035958e-05,
+      "loss": 0.077,
+      "step": 27495
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.751535324176082e-05,
+      "loss": 0.093,
+      "step": 27500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.751488444048568e-05,
+      "loss": 0.11,
+      "step": 27505
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.751441563921054e-05,
+      "loss": 0.1683,
+      "step": 27510
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.75139468379354e-05,
+      "loss": 0.1233,
+      "step": 27515
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.751347803666026e-05,
+      "loss": 0.207,
+      "step": 27520
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.751300923538512e-05,
+      "loss": 0.3287,
+      "step": 27525
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.751254043410998e-05,
+      "loss": 0.4735,
+      "step": 27530
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.751207163283484e-05,
+      "loss": 0.2638,
+      "step": 27535
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7511602831559704e-05,
+      "loss": 0.0586,
+      "step": 27540
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7511134030284564e-05,
+      "loss": 0.0821,
+      "step": 27545
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7510665229009424e-05,
+      "loss": 0.1407,
+      "step": 27550
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7510196427734284e-05,
+      "loss": 0.1208,
+      "step": 27555
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7509727626459144e-05,
+      "loss": 0.1009,
+      "step": 27560
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7509258825184007e-05,
+      "loss": 0.196,
+      "step": 27565
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7508790023908867e-05,
+      "loss": 0.17,
+      "step": 27570
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7508321222633727e-05,
+      "loss": 0.3356,
+      "step": 27575
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7507852421358587e-05,
+      "loss": 0.3552,
+      "step": 27580
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.750738362008345e-05,
+      "loss": 0.2562,
+      "step": 27585
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.750691481880831e-05,
+      "loss": 0.0318,
+      "step": 27590
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.750644601753317e-05,
+      "loss": 0.0508,
+      "step": 27595
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.750597721625803e-05,
+      "loss": 0.1133,
+      "step": 27600
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.750550841498289e-05,
+      "loss": 0.1333,
+      "step": 27605
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.750503961370775e-05,
+      "loss": 0.1305,
+      "step": 27610
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.750457081243261e-05,
+      "loss": 0.1097,
+      "step": 27615
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.750410201115747e-05,
+      "loss": 0.1544,
+      "step": 27620
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.750363320988233e-05,
+      "loss": 0.2137,
+      "step": 27625
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7503164408607193e-05,
+      "loss": 0.4484,
+      "step": 27630
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7502695607332053e-05,
+      "loss": 0.2642,
+      "step": 27635
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7502226806056913e-05,
+      "loss": 0.0451,
+      "step": 27640
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7501758004781776e-05,
+      "loss": 0.0739,
+      "step": 27645
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7501289203506636e-05,
+      "loss": 0.0869,
+      "step": 27650
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7500820402231496e-05,
+      "loss": 0.0963,
+      "step": 27655
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7500351600956356e-05,
+      "loss": 0.0773,
+      "step": 27660
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7499882799681216e-05,
+      "loss": 0.1242,
+      "step": 27665
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7499413998406076e-05,
+      "loss": 0.1775,
+      "step": 27670
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7498945197130936e-05,
+      "loss": 0.3349,
+      "step": 27675
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.74984763958558e-05,
+      "loss": 0.2802,
+      "step": 27680
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.749800759458066e-05,
+      "loss": 0.3577,
+      "step": 27685
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.749753879330552e-05,
+      "loss": 0.0712,
+      "step": 27690
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.749706999203038e-05,
+      "loss": 0.0935,
+      "step": 27695
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.749660119075524e-05,
+      "loss": 0.0822,
+      "step": 27700
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.74961323894801e-05,
+      "loss": 0.1642,
+      "step": 27705
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7495663588204962e-05,
+      "loss": 0.2227,
+      "step": 27710
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7495194786929822e-05,
+      "loss": 0.1858,
+      "step": 27715
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7494725985654682e-05,
+      "loss": 0.199,
+      "step": 27720
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7494257184379545e-05,
+      "loss": 0.2302,
+      "step": 27725
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7493788383104405e-05,
+      "loss": 0.3215,
+      "step": 27730
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7493319581829265e-05,
+      "loss": 0.2974,
+      "step": 27735
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7492850780554125e-05,
+      "loss": 0.0448,
+      "step": 27740
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7492381979278985e-05,
+      "loss": 0.1718,
+      "step": 27745
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7491913178003845e-05,
+      "loss": 0.1088,
+      "step": 27750
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7491444376728705e-05,
+      "loss": 0.1133,
+      "step": 27755
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7490975575453565e-05,
+      "loss": 0.1354,
+      "step": 27760
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7490506774178425e-05,
+      "loss": 0.1955,
+      "step": 27765
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7490037972903288e-05,
+      "loss": 0.1759,
+      "step": 27770
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7489569171628148e-05,
+      "loss": 0.2415,
+      "step": 27775
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7489100370353008e-05,
+      "loss": 0.5153,
+      "step": 27780
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7488631569077868e-05,
+      "loss": 0.3479,
+      "step": 27785
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.748816276780273e-05,
+      "loss": 0.0432,
+      "step": 27790
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.748769396652759e-05,
+      "loss": 0.0697,
+      "step": 27795
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.748722516525245e-05,
+      "loss": 0.084,
+      "step": 27800
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.748675636397731e-05,
+      "loss": 0.1467,
+      "step": 27805
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.748628756270217e-05,
+      "loss": 0.1314,
+      "step": 27810
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7485818761427034e-05,
+      "loss": 0.176,
+      "step": 27815
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7485349960151894e-05,
+      "loss": 0.1543,
+      "step": 27820
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7484881158876754e-05,
+      "loss": 0.1403,
+      "step": 27825
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7484412357601614e-05,
+      "loss": 0.2803,
+      "step": 27830
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7483943556326474e-05,
+      "loss": 0.2421,
+      "step": 27835
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7483474755051334e-05,
+      "loss": 0.0877,
+      "step": 27840
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7483005953776194e-05,
+      "loss": 0.034,
+      "step": 27845
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7482537152501057e-05,
+      "loss": 0.1936,
+      "step": 27850
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7482068351225917e-05,
+      "loss": 0.0793,
+      "step": 27855
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7481599549950777e-05,
+      "loss": 0.1592,
+      "step": 27860
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.748113074867564e-05,
+      "loss": 0.1525,
+      "step": 27865
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.74806619474005e-05,
+      "loss": 0.0891,
+      "step": 27870
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.748019314612536e-05,
+      "loss": 0.3361,
+      "step": 27875
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.747972434485022e-05,
+      "loss": 0.3712,
+      "step": 27880
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.747925554357508e-05,
+      "loss": 0.256,
+      "step": 27885
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.747878674229994e-05,
+      "loss": 0.0428,
+      "step": 27890
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.74783179410248e-05,
+      "loss": 0.0539,
+      "step": 27895
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.747784913974966e-05,
+      "loss": 0.0732,
+      "step": 27900
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.747738033847452e-05,
+      "loss": 0.0836,
+      "step": 27905
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7476911537199383e-05,
+      "loss": 0.096,
+      "step": 27910
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7476442735924243e-05,
+      "loss": 0.125,
+      "step": 27915
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7475973934649103e-05,
+      "loss": 0.1961,
+      "step": 27920
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7475505133373963e-05,
+      "loss": 0.2617,
+      "step": 27925
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7475036332098826e-05,
+      "loss": 0.282,
+      "step": 27930
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7474567530823686e-05,
+      "loss": 0.2166,
+      "step": 27935
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7474098729548546e-05,
+      "loss": 0.0175,
+      "step": 27940
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7473629928273406e-05,
+      "loss": 0.101,
+      "step": 27945
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7473161126998266e-05,
+      "loss": 0.1078,
+      "step": 27950
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.747269232572313e-05,
+      "loss": 0.1029,
+      "step": 27955
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.747222352444799e-05,
+      "loss": 0.1267,
+      "step": 27960
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.747175472317285e-05,
+      "loss": 0.1834,
+      "step": 27965
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.747128592189771e-05,
+      "loss": 0.2201,
+      "step": 27970
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.747081712062257e-05,
+      "loss": 0.233,
+      "step": 27975
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.747034831934743e-05,
+      "loss": 0.3193,
+      "step": 27980
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.746987951807229e-05,
+      "loss": 0.2488,
+      "step": 27985
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.746941071679715e-05,
+      "loss": 0.101,
+      "step": 27990
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7468941915522012e-05,
+      "loss": 0.0682,
+      "step": 27995
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7468473114246872e-05,
+      "loss": 0.0767,
+      "step": 28000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7468004312971732e-05,
+      "loss": 0.1169,
+      "step": 28005
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7467535511696595e-05,
+      "loss": 0.147,
+      "step": 28010
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7467066710421455e-05,
+      "loss": 0.1447,
+      "step": 28015
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7466597909146315e-05,
+      "loss": 0.2328,
+      "step": 28020
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7466129107871175e-05,
+      "loss": 0.1731,
+      "step": 28025
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7465660306596035e-05,
+      "loss": 0.4335,
+      "step": 28030
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7465191505320895e-05,
+      "loss": 0.2766,
+      "step": 28035
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7464722704045755e-05,
+      "loss": 0.0367,
+      "step": 28040
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7464253902770615e-05,
+      "loss": 0.0953,
+      "step": 28045
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.746378510149548e-05,
+      "loss": 0.0909,
+      "step": 28050
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7463316300220338e-05,
+      "loss": 0.087,
+      "step": 28055
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7462847498945198e-05,
+      "loss": 0.0996,
+      "step": 28060
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7462378697670058e-05,
+      "loss": 0.2102,
+      "step": 28065
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7461909896394918e-05,
+      "loss": 0.2142,
+      "step": 28070
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.746144109511978e-05,
+      "loss": 0.2751,
+      "step": 28075
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.746097229384464e-05,
+      "loss": 0.2934,
+      "step": 28080
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.74605034925695e-05,
+      "loss": 0.2695,
+      "step": 28085
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.746003469129436e-05,
+      "loss": 0.0472,
+      "step": 28090
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7459565890019224e-05,
+      "loss": 0.0919,
+      "step": 28095
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7459097088744084e-05,
+      "loss": 0.1356,
+      "step": 28100
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7458628287468944e-05,
+      "loss": 0.1815,
+      "step": 28105
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7458159486193804e-05,
+      "loss": 0.1405,
+      "step": 28110
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7457690684918664e-05,
+      "loss": 0.1442,
+      "step": 28115
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7457221883643524e-05,
+      "loss": 0.1353,
+      "step": 28120
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7456753082368384e-05,
+      "loss": 0.3492,
+      "step": 28125
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7456284281093244e-05,
+      "loss": 0.1896,
+      "step": 28130
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7455815479818104e-05,
+      "loss": 0.2213,
+      "step": 28135
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7455346678542967e-05,
+      "loss": 0.0696,
+      "step": 28140
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7454877877267827e-05,
+      "loss": 0.0769,
+      "step": 28145
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7454409075992687e-05,
+      "loss": 0.0635,
+      "step": 28150
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.745394027471755e-05,
+      "loss": 0.1047,
+      "step": 28155
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.745347147344241e-05,
+      "loss": 0.1078,
+      "step": 28160
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.745300267216727e-05,
+      "loss": 0.2376,
+      "step": 28165
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.745253387089213e-05,
+      "loss": 0.1726,
+      "step": 28170
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.745206506961699e-05,
+      "loss": 0.2706,
+      "step": 28175
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.745159626834185e-05,
+      "loss": 0.3923,
+      "step": 28180
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.745112746706671e-05,
+      "loss": 0.2519,
+      "step": 28185
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7450658665791573e-05,
+      "loss": 0.0364,
+      "step": 28190
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7450189864516433e-05,
+      "loss": 0.1435,
+      "step": 28195
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7449721063241293e-05,
+      "loss": 0.0601,
+      "step": 28200
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7449252261966153e-05,
+      "loss": 0.0721,
+      "step": 28205
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7448783460691013e-05,
+      "loss": 0.1386,
+      "step": 28210
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7448314659415876e-05,
+      "loss": 0.1472,
+      "step": 28215
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7447845858140736e-05,
+      "loss": 0.2471,
+      "step": 28220
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7447377056865596e-05,
+      "loss": 0.2434,
+      "step": 28225
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7446908255590456e-05,
+      "loss": 0.3995,
+      "step": 28230
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.744643945431532e-05,
+      "loss": 0.4138,
+      "step": 28235
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.744597065304018e-05,
+      "loss": 0.091,
+      "step": 28240
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.744550185176504e-05,
+      "loss": 0.0882,
+      "step": 28245
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.74450330504899e-05,
+      "loss": 0.0817,
+      "step": 28250
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.744456424921476e-05,
+      "loss": 0.1233,
+      "step": 28255
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.744409544793962e-05,
+      "loss": 0.0671,
+      "step": 28260
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.744362664666448e-05,
+      "loss": 0.1374,
+      "step": 28265
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.744315784538934e-05,
+      "loss": 0.1614,
+      "step": 28270
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.74426890441142e-05,
+      "loss": 0.1907,
+      "step": 28275
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7442220242839062e-05,
+      "loss": 0.385,
+      "step": 28280
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7441751441563922e-05,
+      "loss": 0.2444,
+      "step": 28285
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7441282640288782e-05,
+      "loss": 0.0702,
+      "step": 28290
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7440813839013646e-05,
+      "loss": 0.1009,
+      "step": 28295
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7440345037738505e-05,
+      "loss": 0.0929,
+      "step": 28300
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7439876236463365e-05,
+      "loss": 0.1733,
+      "step": 28305
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7439407435188225e-05,
+      "loss": 0.1582,
+      "step": 28310
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7438938633913085e-05,
+      "loss": 0.1792,
+      "step": 28315
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7438469832637945e-05,
+      "loss": 0.1222,
+      "step": 28320
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7438001031362805e-05,
+      "loss": 0.3091,
+      "step": 28325
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.743753223008767e-05,
+      "loss": 0.4052,
+      "step": 28330
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.743706342881253e-05,
+      "loss": 0.198,
+      "step": 28335
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.743659462753739e-05,
+      "loss": 0.0879,
+      "step": 28340
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7436125826262248e-05,
+      "loss": 0.0845,
+      "step": 28345
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7435657024987108e-05,
+      "loss": 0.0569,
+      "step": 28350
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7435188223711968e-05,
+      "loss": 0.1203,
+      "step": 28355
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.743471942243683e-05,
+      "loss": 0.1062,
+      "step": 28360
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.743425062116169e-05,
+      "loss": 0.2103,
+      "step": 28365
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.743378181988655e-05,
+      "loss": 0.2832,
+      "step": 28370
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7433313018611415e-05,
+      "loss": 0.2313,
+      "step": 28375
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7432844217336275e-05,
+      "loss": 0.4534,
+      "step": 28380
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7432375416061135e-05,
+      "loss": 0.238,
+      "step": 28385
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7431906614785994e-05,
+      "loss": 0.0604,
+      "step": 28390
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7431437813510854e-05,
+      "loss": 0.1129,
+      "step": 28395
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7430969012235714e-05,
+      "loss": 0.0935,
+      "step": 28400
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7430500210960574e-05,
+      "loss": 0.0649,
+      "step": 28405
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7430031409685434e-05,
+      "loss": 0.0958,
+      "step": 28410
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7429562608410294e-05,
+      "loss": 0.1529,
+      "step": 28415
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7429093807135157e-05,
+      "loss": 0.2254,
+      "step": 28420
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7428625005860017e-05,
+      "loss": 0.3171,
+      "step": 28425
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7428156204584877e-05,
+      "loss": 0.3572,
+      "step": 28430
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7427687403309737e-05,
+      "loss": 0.2508,
+      "step": 28435
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.74272186020346e-05,
+      "loss": 0.0471,
+      "step": 28440
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.742674980075946e-05,
+      "loss": 0.097,
+      "step": 28445
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.742628099948432e-05,
+      "loss": 0.039,
+      "step": 28450
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.742581219820918e-05,
+      "loss": 0.1062,
+      "step": 28455
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.742534339693404e-05,
+      "loss": 0.0994,
+      "step": 28460
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7424874595658904e-05,
+      "loss": 0.0935,
+      "step": 28465
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7424405794383764e-05,
+      "loss": 0.1496,
+      "step": 28470
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7423936993108623e-05,
+      "loss": 0.2379,
+      "step": 28475
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7423468191833483e-05,
+      "loss": 0.1887,
+      "step": 28480
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7422999390558343e-05,
+      "loss": 0.1909,
+      "step": 28485
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7422530589283203e-05,
+      "loss": 0.0401,
+      "step": 28490
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7422061788008063e-05,
+      "loss": 0.0426,
+      "step": 28495
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7421592986732923e-05,
+      "loss": 0.0825,
+      "step": 28500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7421124185457786e-05,
+      "loss": 0.1346,
+      "step": 28505
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7420655384182646e-05,
+      "loss": 0.1569,
+      "step": 28510
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7420186582907506e-05,
+      "loss": 0.1373,
+      "step": 28515
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.741971778163237e-05,
+      "loss": 0.1554,
+      "step": 28520
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.741924898035723e-05,
+      "loss": 0.223,
+      "step": 28525
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.741878017908209e-05,
+      "loss": 0.5315,
+      "step": 28530
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.741831137780695e-05,
+      "loss": 0.2349,
+      "step": 28535
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.741784257653181e-05,
+      "loss": 0.0541,
+      "step": 28540
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.741737377525667e-05,
+      "loss": 0.086,
+      "step": 28545
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.741690497398153e-05,
+      "loss": 0.08,
+      "step": 28550
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.741643617270639e-05,
+      "loss": 0.0762,
+      "step": 28555
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7415967371431253e-05,
+      "loss": 0.1393,
+      "step": 28560
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7415498570156112e-05,
+      "loss": 0.169,
+      "step": 28565
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7415029768880972e-05,
+      "loss": 0.1328,
+      "step": 28570
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7414560967605832e-05,
+      "loss": 0.2354,
+      "step": 28575
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7414092166330692e-05,
+      "loss": 0.4134,
+      "step": 28580
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7413623365055556e-05,
+      "loss": 0.3811,
+      "step": 28585
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7413154563780416e-05,
+      "loss": 0.0618,
+      "step": 28590
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7412685762505275e-05,
+      "loss": 0.0739,
+      "step": 28595
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7412216961230135e-05,
+      "loss": 0.1536,
+      "step": 28600
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7411748159955e-05,
+      "loss": 0.1513,
+      "step": 28605
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.741127935867986e-05,
+      "loss": 0.1786,
+      "step": 28610
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.741081055740472e-05,
+      "loss": 0.1871,
+      "step": 28615
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.741034175612958e-05,
+      "loss": 0.1789,
+      "step": 28620
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.740987295485444e-05,
+      "loss": 0.247,
+      "step": 28625
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.74094041535793e-05,
+      "loss": 0.2658,
+      "step": 28630
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.740893535230416e-05,
+      "loss": 0.2673,
+      "step": 28635
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7408466551029018e-05,
+      "loss": 0.0536,
+      "step": 28640
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.740799774975388e-05,
+      "loss": 0.0496,
+      "step": 28645
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.740752894847874e-05,
+      "loss": 0.1276,
+      "step": 28650
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.74070601472036e-05,
+      "loss": 0.0875,
+      "step": 28655
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.740659134592846e-05,
+      "loss": 0.1346,
+      "step": 28660
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7406122544653325e-05,
+      "loss": 0.1021,
+      "step": 28665
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7405653743378185e-05,
+      "loss": 0.1522,
+      "step": 28670
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7405184942103045e-05,
+      "loss": 0.3115,
+      "step": 28675
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7404716140827904e-05,
+      "loss": 0.3448,
+      "step": 28680
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7404247339552764e-05,
+      "loss": 0.3397,
+      "step": 28685
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7403778538277624e-05,
+      "loss": 0.1061,
+      "step": 28690
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7403309737002484e-05,
+      "loss": 0.0856,
+      "step": 28695
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7402840935727348e-05,
+      "loss": 0.0919,
+      "step": 28700
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7402372134452208e-05,
+      "loss": 0.2105,
+      "step": 28705
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7401903333177067e-05,
+      "loss": 0.1149,
+      "step": 28710
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7401434531901927e-05,
+      "loss": 0.1549,
+      "step": 28715
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7400965730626787e-05,
+      "loss": 0.201,
+      "step": 28720
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.740049692935165e-05,
+      "loss": 0.235,
+      "step": 28725
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.740002812807651e-05,
+      "loss": 0.2196,
+      "step": 28730
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.739955932680137e-05,
+      "loss": 0.2805,
+      "step": 28735
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.739909052552623e-05,
+      "loss": 0.0878,
+      "step": 28740
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7398621724251094e-05,
+      "loss": 0.1229,
+      "step": 28745
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7398152922975954e-05,
+      "loss": 0.0782,
+      "step": 28750
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7397684121700814e-05,
+      "loss": 0.0866,
+      "step": 28755
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7397215320425674e-05,
+      "loss": 0.151,
+      "step": 28760
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7396746519150534e-05,
+      "loss": 0.1959,
+      "step": 28765
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7396277717875393e-05,
+      "loss": 0.2202,
+      "step": 28770
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7395808916600253e-05,
+      "loss": 0.1692,
+      "step": 28775
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7395340115325113e-05,
+      "loss": 0.4181,
+      "step": 28780
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7394871314049973e-05,
+      "loss": 0.3173,
+      "step": 28785
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7394402512774837e-05,
+      "loss": 0.0864,
+      "step": 28790
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7393933711499697e-05,
+      "loss": 0.0677,
+      "step": 28795
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7393464910224556e-05,
+      "loss": 0.0952,
+      "step": 28800
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.739299610894942e-05,
+      "loss": 0.0573,
+      "step": 28805
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.739252730767428e-05,
+      "loss": 0.2943,
+      "step": 28810
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.739205850639914e-05,
+      "loss": 0.142,
+      "step": 28815
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7391589705124e-05,
+      "loss": 0.0782,
+      "step": 28820
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.739112090384886e-05,
+      "loss": 0.2885,
+      "step": 28825
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.739065210257372e-05,
+      "loss": 0.3619,
+      "step": 28830
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.739018330129858e-05,
+      "loss": 0.3443,
+      "step": 28835
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7389714500023443e-05,
+      "loss": 0.0266,
+      "step": 28840
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7389245698748303e-05,
+      "loss": 0.0487,
+      "step": 28845
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7388776897473163e-05,
+      "loss": 0.0475,
+      "step": 28850
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7388308096198023e-05,
+      "loss": 0.1119,
+      "step": 28855
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7387839294922882e-05,
+      "loss": 0.1051,
+      "step": 28860
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7387370493647742e-05,
+      "loss": 0.1066,
+      "step": 28865
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7386901692372606e-05,
+      "loss": 0.1715,
+      "step": 28870
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7386432891097466e-05,
+      "loss": 0.1286,
+      "step": 28875
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7385964089822326e-05,
+      "loss": 0.2792,
+      "step": 28880
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.738549528854719e-05,
+      "loss": 0.2294,
+      "step": 28885
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.738502648727205e-05,
+      "loss": 0.0443,
+      "step": 28890
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.738455768599691e-05,
+      "loss": 0.0679,
+      "step": 28895
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.738408888472177e-05,
+      "loss": 0.0798,
+      "step": 28900
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.738362008344663e-05,
+      "loss": 0.0693,
+      "step": 28905
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.738315128217149e-05,
+      "loss": 0.129,
+      "step": 28910
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.738268248089635e-05,
+      "loss": 0.1704,
+      "step": 28915
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.738221367962121e-05,
+      "loss": 0.1876,
+      "step": 28920
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.738174487834607e-05,
+      "loss": 0.3406,
+      "step": 28925
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.738127607707093e-05,
+      "loss": 0.3488,
+      "step": 28930
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.738080727579579e-05,
+      "loss": 0.3817,
+      "step": 28935
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.738033847452065e-05,
+      "loss": 0.0798,
+      "step": 28940
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.737986967324551e-05,
+      "loss": 0.1087,
+      "step": 28945
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7379400871970375e-05,
+      "loss": 0.0793,
+      "step": 28950
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7378932070695235e-05,
+      "loss": 0.1104,
+      "step": 28955
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7378463269420095e-05,
+      "loss": 0.1382,
+      "step": 28960
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7377994468144955e-05,
+      "loss": 0.207,
+      "step": 28965
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7377525666869815e-05,
+      "loss": 0.1813,
+      "step": 28970
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7377056865594674e-05,
+      "loss": 0.2456,
+      "step": 28975
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7376588064319538e-05,
+      "loss": 0.2929,
+      "step": 28980
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7376119263044398e-05,
+      "loss": 0.2645,
+      "step": 28985
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7375650461769258e-05,
+      "loss": 0.0423,
+      "step": 28990
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7375181660494118e-05,
+      "loss": 0.1462,
+      "step": 28995
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7374712859218978e-05,
+      "loss": 0.0766,
+      "step": 29000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7374244057943837e-05,
+      "loss": 0.0594,
+      "step": 29005
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7373775256668697e-05,
+      "loss": 0.1614,
+      "step": 29010
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.737330645539356e-05,
+      "loss": 0.0778,
+      "step": 29015
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.737283765411842e-05,
+      "loss": 0.1516,
+      "step": 29020
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.737236885284328e-05,
+      "loss": 0.3276,
+      "step": 29025
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7371900051568144e-05,
+      "loss": 0.3269,
+      "step": 29030
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7371431250293004e-05,
+      "loss": 0.3945,
+      "step": 29035
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7370962449017864e-05,
+      "loss": 0.0705,
+      "step": 29040
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7370493647742724e-05,
+      "loss": 0.1158,
+      "step": 29045
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7370024846467584e-05,
+      "loss": 0.0786,
+      "step": 29050
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7369556045192444e-05,
+      "loss": 0.1491,
+      "step": 29055
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7369087243917303e-05,
+      "loss": 0.1254,
+      "step": 29060
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7368618442642163e-05,
+      "loss": 0.0802,
+      "step": 29065
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7368149641367027e-05,
+      "loss": 0.2187,
+      "step": 29070
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7367680840091887e-05,
+      "loss": 0.1845,
+      "step": 29075
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7367212038816747e-05,
+      "loss": 0.4131,
+      "step": 29080
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7366743237541607e-05,
+      "loss": 0.2237,
+      "step": 29085
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7366274436266466e-05,
+      "loss": 0.0269,
+      "step": 29090
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.736580563499133e-05,
+      "loss": 0.1229,
+      "step": 29095
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.736533683371619e-05,
+      "loss": 0.0902,
+      "step": 29100
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.736486803244105e-05,
+      "loss": 0.148,
+      "step": 29105
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.736439923116591e-05,
+      "loss": 0.109,
+      "step": 29110
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7363930429890773e-05,
+      "loss": 0.1508,
+      "step": 29115
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7363461628615633e-05,
+      "loss": 0.1622,
+      "step": 29120
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7362992827340493e-05,
+      "loss": 0.2647,
+      "step": 29125
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7362524026065353e-05,
+      "loss": 0.2541,
+      "step": 29130
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7362055224790213e-05,
+      "loss": 0.2231,
+      "step": 29135
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7361586423515073e-05,
+      "loss": 0.0197,
+      "step": 29140
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7361117622239933e-05,
+      "loss": 0.0373,
+      "step": 29145
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7360648820964792e-05,
+      "loss": 0.0942,
+      "step": 29150
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7360180019689656e-05,
+      "loss": 0.0481,
+      "step": 29155
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7359711218414516e-05,
+      "loss": 0.1787,
+      "step": 29160
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7359242417139376e-05,
+      "loss": 0.1285,
+      "step": 29165
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7358773615864236e-05,
+      "loss": 0.2024,
+      "step": 29170
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.73583048145891e-05,
+      "loss": 0.2614,
+      "step": 29175
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.735783601331396e-05,
+      "loss": 0.2649,
+      "step": 29180
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.735736721203882e-05,
+      "loss": 0.2944,
+      "step": 29185
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.735689841076368e-05,
+      "loss": 0.0346,
+      "step": 29190
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.735642960948854e-05,
+      "loss": 0.1004,
+      "step": 29195
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.73559608082134e-05,
+      "loss": 0.1003,
+      "step": 29200
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.735549200693826e-05,
+      "loss": 0.1607,
+      "step": 29205
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7355023205663122e-05,
+      "loss": 0.153,
+      "step": 29210
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7354554404387982e-05,
+      "loss": 0.122,
+      "step": 29215
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7354085603112842e-05,
+      "loss": 0.2213,
+      "step": 29220
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.73536168018377e-05,
+      "loss": 0.2523,
+      "step": 29225
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.735314800056256e-05,
+      "loss": 0.4636,
+      "step": 29230
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7352679199287425e-05,
+      "loss": 0.3613,
+      "step": 29235
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7352210398012285e-05,
+      "loss": 0.0464,
+      "step": 29240
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7351741596737145e-05,
+      "loss": 0.0687,
+      "step": 29245
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7351272795462005e-05,
+      "loss": 0.0908,
+      "step": 29250
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7350803994186868e-05,
+      "loss": 0.1691,
+      "step": 29255
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7350335192911728e-05,
+      "loss": 0.1535,
+      "step": 29260
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7349866391636588e-05,
+      "loss": 0.205,
+      "step": 29265
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7349397590361448e-05,
+      "loss": 0.1682,
+      "step": 29270
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7348928789086308e-05,
+      "loss": 0.3615,
+      "step": 29275
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7348459987811168e-05,
+      "loss": 0.3168,
+      "step": 29280
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7347991186536028e-05,
+      "loss": 0.3735,
+      "step": 29285
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7347522385260888e-05,
+      "loss": 0.0129,
+      "step": 29290
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7347053583985747e-05,
+      "loss": 0.0673,
+      "step": 29295
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.734658478271061e-05,
+      "loss": 0.0334,
+      "step": 29300
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.734611598143547e-05,
+      "loss": 0.1941,
+      "step": 29305
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.734564718016033e-05,
+      "loss": 0.1482,
+      "step": 29310
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7345178378885194e-05,
+      "loss": 0.1165,
+      "step": 29315
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7344709577610054e-05,
+      "loss": 0.245,
+      "step": 29320
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7344240776334914e-05,
+      "loss": 0.2832,
+      "step": 29325
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7343771975059774e-05,
+      "loss": 0.6608,
+      "step": 29330
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7343303173784634e-05,
+      "loss": 0.5553,
+      "step": 29335
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7342834372509494e-05,
+      "loss": 0.0925,
+      "step": 29340
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7342365571234354e-05,
+      "loss": 0.1114,
+      "step": 29345
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7341896769959217e-05,
+      "loss": 0.0405,
+      "step": 29350
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7341427968684077e-05,
+      "loss": 0.1301,
+      "step": 29355
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7340959167408937e-05,
+      "loss": 0.1224,
+      "step": 29360
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7340490366133797e-05,
+      "loss": 0.1806,
+      "step": 29365
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7340021564858657e-05,
+      "loss": 0.1208,
+      "step": 29370
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7339552763583517e-05,
+      "loss": 0.2126,
+      "step": 29375
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.733908396230838e-05,
+      "loss": 0.2936,
+      "step": 29380
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.733861516103324e-05,
+      "loss": 0.3515,
+      "step": 29385
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.73381463597581e-05,
+      "loss": 0.1816,
+      "step": 29390
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7337677558482963e-05,
+      "loss": 0.065,
+      "step": 29395
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7337208757207823e-05,
+      "loss": 0.089,
+      "step": 29400
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7336739955932683e-05,
+      "loss": 0.0681,
+      "step": 29405
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7336271154657543e-05,
+      "loss": 0.1273,
+      "step": 29410
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7335802353382403e-05,
+      "loss": 0.1598,
+      "step": 29415
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7335333552107263e-05,
+      "loss": 0.1456,
+      "step": 29420
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7334864750832123e-05,
+      "loss": 0.275,
+      "step": 29425
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7334395949556983e-05,
+      "loss": 0.314,
+      "step": 29430
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7333927148281843e-05,
+      "loss": 0.298,
+      "step": 29435
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7333458347006706e-05,
+      "loss": 0.0133,
+      "step": 29440
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7332989545731566e-05,
+      "loss": 0.0531,
+      "step": 29445
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7332520744456426e-05,
+      "loss": 0.0883,
+      "step": 29450
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7332051943181286e-05,
+      "loss": 0.0872,
+      "step": 29455
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.733158314190615e-05,
+      "loss": 0.085,
+      "step": 29460
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.733111434063101e-05,
+      "loss": 0.1929,
+      "step": 29465
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.733064553935587e-05,
+      "loss": 0.1555,
+      "step": 29470
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.733017673808073e-05,
+      "loss": 0.1887,
+      "step": 29475
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.732970793680559e-05,
+      "loss": 0.241,
+      "step": 29480
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.732923913553045e-05,
+      "loss": 0.2993,
+      "step": 29485
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7328770334255312e-05,
+      "loss": 0.0399,
+      "step": 29490
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7328301532980172e-05,
+      "loss": 0.0674,
+      "step": 29495
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7327832731705032e-05,
+      "loss": 0.068,
+      "step": 29500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7327363930429892e-05,
+      "loss": 0.1605,
+      "step": 29505
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7326895129154752e-05,
+      "loss": 0.095,
+      "step": 29510
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.732642632787961e-05,
+      "loss": 0.1541,
+      "step": 29515
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.732595752660447e-05,
+      "loss": 0.197,
+      "step": 29520
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7325488725329335e-05,
+      "loss": 0.1531,
+      "step": 29525
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7325019924054195e-05,
+      "loss": 0.206,
+      "step": 29530
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7324551122779055e-05,
+      "loss": 0.2874,
+      "step": 29535
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7324082321503918e-05,
+      "loss": 0.0179,
+      "step": 29540
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7323613520228778e-05,
+      "loss": 0.0721,
+      "step": 29545
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7323144718953638e-05,
+      "loss": 0.0472,
+      "step": 29550
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7322675917678498e-05,
+      "loss": 0.0831,
+      "step": 29555
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7322207116403358e-05,
+      "loss": 0.0901,
+      "step": 29560
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7321738315128218e-05,
+      "loss": 0.1381,
+      "step": 29565
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7321269513853078e-05,
+      "loss": 0.2179,
+      "step": 29570
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7320800712577938e-05,
+      "loss": 0.298,
+      "step": 29575
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.73203319113028e-05,
+      "loss": 0.3747,
+      "step": 29580
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.731986311002766e-05,
+      "loss": 0.2943,
+      "step": 29585
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.731939430875252e-05,
+      "loss": 0.0136,
+      "step": 29590
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.731892550747738e-05,
+      "loss": 0.1105,
+      "step": 29595
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.731845670620224e-05,
+      "loss": 0.104,
+      "step": 29600
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7317987904927104e-05,
+      "loss": 0.0937,
+      "step": 29605
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7317519103651964e-05,
+      "loss": 0.0857,
+      "step": 29610
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7317050302376824e-05,
+      "loss": 0.2497,
+      "step": 29615
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7316581501101684e-05,
+      "loss": 0.2124,
+      "step": 29620
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7316112699826544e-05,
+      "loss": 0.2004,
+      "step": 29625
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7315643898551407e-05,
+      "loss": 0.5103,
+      "step": 29630
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7315175097276267e-05,
+      "loss": 0.2454,
+      "step": 29635
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7314706296001127e-05,
+      "loss": 0.0501,
+      "step": 29640
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7314237494725987e-05,
+      "loss": 0.0609,
+      "step": 29645
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7313768693450847e-05,
+      "loss": 0.1009,
+      "step": 29650
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7313299892175707e-05,
+      "loss": 0.1467,
+      "step": 29655
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7312831090900567e-05,
+      "loss": 0.1255,
+      "step": 29660
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.731236228962543e-05,
+      "loss": 0.2066,
+      "step": 29665
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.731189348835029e-05,
+      "loss": 0.1712,
+      "step": 29670
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.731142468707515e-05,
+      "loss": 0.2046,
+      "step": 29675
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7310955885800013e-05,
+      "loss": 0.2975,
+      "step": 29680
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.7310487084524873e-05,
+      "loss": 0.2379,
+      "step": 29685
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.7310018283249733e-05,
+      "loss": 0.0619,
+      "step": 29690
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.7309549481974593e-05,
+      "loss": 0.0921,
+      "step": 29695
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.7309080680699453e-05,
+      "loss": 0.0904,
+      "step": 29700
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.7308611879424313e-05,
+      "loss": 0.1603,
+      "step": 29705
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.7308143078149173e-05,
+      "loss": 0.1683,
+      "step": 29710
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.7307674276874033e-05,
+      "loss": 0.2013,
+      "step": 29715
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.7307205475598896e-05,
+      "loss": 0.193,
+      "step": 29720
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.7306736674323756e-05,
+      "loss": 0.2606,
+      "step": 29725
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.7306267873048616e-05,
+      "loss": 0.4187,
+      "step": 29730
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.7305799071773476e-05,
+      "loss": 0.2754,
+      "step": 29735
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.7305330270498336e-05,
+      "loss": 0.0698,
+      "step": 29740
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.73048614692232e-05,
+      "loss": 0.053,
+      "step": 29745
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.730439266794806e-05,
+      "loss": 0.0753,
+      "step": 29750
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.730392386667292e-05,
+      "loss": 0.1207,
+      "step": 29755
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.730345506539778e-05,
+      "loss": 0.1284,
+      "step": 29760
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.7302986264122642e-05,
+      "loss": 0.1345,
+      "step": 29765
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.7302517462847502e-05,
+      "loss": 0.1827,
+      "step": 29770
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.7302048661572362e-05,
+      "loss": 0.1537,
+      "step": 29775
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.7301579860297222e-05,
+      "loss": 0.3358,
+      "step": 29780
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.7301111059022082e-05,
+      "loss": 0.2873,
+      "step": 29785
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.7300642257746942e-05,
+      "loss": 0.0583,
+      "step": 29790
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.7300173456471802e-05,
+      "loss": 0.0595,
+      "step": 29795
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.7299704655196662e-05,
+      "loss": 0.0834,
+      "step": 29800
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.7299235853921522e-05,
+      "loss": 0.0797,
+      "step": 29805
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.7298767052646385e-05,
+      "loss": 0.0914,
+      "step": 29810
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.7298298251371245e-05,
+      "loss": 0.1655,
+      "step": 29815
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.7297829450096105e-05,
+      "loss": 0.383,
+      "step": 29820
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.7297360648820968e-05,
+      "loss": 0.2981,
+      "step": 29825
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.7296891847545828e-05,
+      "loss": 0.3188,
+      "step": 29830
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.7296423046270688e-05,
+      "loss": 0.3305,
+      "step": 29835
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.7295954244995548e-05,
+      "loss": 0.0652,
+      "step": 29840
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.7295485443720408e-05,
+      "loss": 0.0954,
+      "step": 29845
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.7295016642445268e-05,
+      "loss": 0.1109,
+      "step": 29850
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.7294547841170128e-05,
+      "loss": 0.091,
+      "step": 29855
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.729407903989499e-05,
+      "loss": 0.1538,
+      "step": 29860
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.729361023861985e-05,
+      "loss": 0.1593,
+      "step": 29865
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.729314143734471e-05,
+      "loss": 0.1569,
+      "step": 29870
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.729267263606957e-05,
+      "loss": 0.2395,
+      "step": 29875
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.729220383479443e-05,
+      "loss": 0.3388,
+      "step": 29880
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.729173503351929e-05,
+      "loss": 0.3125,
+      "step": 29885
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.7291266232244154e-05,
+      "loss": 0.0445,
+      "step": 29890
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.7290797430969014e-05,
+      "loss": 0.0567,
+      "step": 29895
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.7290328629693874e-05,
+      "loss": 0.0666,
+      "step": 29900
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.7289859828418737e-05,
+      "loss": 0.1269,
+      "step": 29905
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.7289391027143597e-05,
+      "loss": 0.0666,
+      "step": 29910
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.7288922225868457e-05,
+      "loss": 0.1662,
+      "step": 29915
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.7288453424593317e-05,
+      "loss": 0.2673,
+      "step": 29920
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.7287984623318177e-05,
+      "loss": 0.2463,
+      "step": 29925
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.7287515822043037e-05,
+      "loss": 0.4108,
+      "step": 29930
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.7287047020767897e-05,
+      "loss": 0.3232,
+      "step": 29935
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.7286578219492757e-05,
+      "loss": 0.057,
+      "step": 29940
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.7286109418217617e-05,
+      "loss": 0.0386,
+      "step": 29945
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.728564061694248e-05,
+      "loss": 0.0733,
+      "step": 29950
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.728517181566734e-05,
+      "loss": 0.0618,
+      "step": 29955
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.72847030143922e-05,
+      "loss": 0.1045,
+      "step": 29960
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.728423421311706e-05,
+      "loss": 0.1213,
+      "step": 29965
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.7283765411841923e-05,
+      "loss": 0.1967,
+      "step": 29970
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.7283296610566783e-05,
+      "loss": 0.1985,
+      "step": 29975
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.7282827809291643e-05,
+      "loss": 0.2068,
+      "step": 29980
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.7282359008016503e-05,
+      "loss": 0.3445,
+      "step": 29985
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.7281890206741363e-05,
+      "loss": 0.0561,
+      "step": 29990
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.7281421405466223e-05,
+      "loss": 0.049,
+      "step": 29995
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.7280952604191086e-05,
+      "loss": 0.0819,
+      "step": 30000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.7280483802915946e-05,
+      "loss": 0.0887,
+      "step": 30005
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.7280015001640806e-05,
+      "loss": 0.0983,
+      "step": 30010
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.7279546200365666e-05,
+      "loss": 0.1115,
+      "step": 30015
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.7279077399090526e-05,
+      "loss": 0.1199,
+      "step": 30020
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.7278608597815386e-05,
+      "loss": 0.277,
+      "step": 30025
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.727813979654025e-05,
+      "loss": 0.2758,
+      "step": 30030
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.727767099526511e-05,
+      "loss": 0.3149,
+      "step": 30035
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.727720219398997e-05,
+      "loss": 0.0286,
+      "step": 30040
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.727673339271483e-05,
+      "loss": 0.0506,
+      "step": 30045
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.7276264591439692e-05,
+      "loss": 0.0629,
+      "step": 30050
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.7275795790164552e-05,
+      "loss": 0.1089,
+      "step": 30055
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.7275326988889412e-05,
+      "loss": 0.0838,
+      "step": 30060
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.7274858187614272e-05,
+      "loss": 0.1634,
+      "step": 30065
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.7274389386339132e-05,
+      "loss": 0.2009,
+      "step": 30070
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.7273920585063992e-05,
+      "loss": 0.1953,
+      "step": 30075
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.7273451783788852e-05,
+      "loss": 0.3996,
+      "step": 30080
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.7272982982513712e-05,
+      "loss": 0.4205,
+      "step": 30085
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.7272514181238575e-05,
+      "loss": 0.0856,
+      "step": 30090
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.7272045379963435e-05,
+      "loss": 0.1023,
+      "step": 30095
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.7271576578688295e-05,
+      "loss": 0.0996,
+      "step": 30100
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.7271107777413155e-05,
+      "loss": 0.131,
+      "step": 30105
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.727063897613802e-05,
+      "loss": 0.1035,
+      "step": 30110
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.7270170174862878e-05,
+      "loss": 0.15,
+      "step": 30115
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.7269701373587738e-05,
+      "loss": 0.144,
+      "step": 30120
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.7269232572312598e-05,
+      "loss": 0.1548,
+      "step": 30125
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.7268763771037458e-05,
+      "loss": 0.3646,
+      "step": 30130
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.7268294969762318e-05,
+      "loss": 0.4213,
+      "step": 30135
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.726782616848718e-05,
+      "loss": 0.0461,
+      "step": 30140
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.726735736721204e-05,
+      "loss": 0.0972,
+      "step": 30145
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.72668885659369e-05,
+      "loss": 0.0614,
+      "step": 30150
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.726641976466176e-05,
+      "loss": 0.1055,
+      "step": 30155
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.726595096338662e-05,
+      "loss": 0.1407,
+      "step": 30160
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.726548216211148e-05,
+      "loss": 0.1042,
+      "step": 30165
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.726501336083634e-05,
+      "loss": 0.23,
+      "step": 30170
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.7264544559561204e-05,
+      "loss": 0.2761,
+      "step": 30175
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.7264075758286064e-05,
+      "loss": 0.2038,
+      "step": 30180
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.7263606957010924e-05,
+      "loss": 0.3806,
+      "step": 30185
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.7263138155735787e-05,
+      "loss": 0.0514,
+      "step": 30190
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.7262669354460647e-05,
+      "loss": 0.0675,
+      "step": 30195
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.7262200553185507e-05,
+      "loss": 0.0407,
+      "step": 30200
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.7261731751910367e-05,
+      "loss": 0.1628,
+      "step": 30205
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.7261262950635227e-05,
+      "loss": 0.1318,
+      "step": 30210
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.7260794149360087e-05,
+      "loss": 0.1692,
+      "step": 30215
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.7260325348084947e-05,
+      "loss": 0.1106,
+      "step": 30220
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.7259856546809807e-05,
+      "loss": 0.2392,
+      "step": 30225
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.725938774553467e-05,
+      "loss": 0.3086,
+      "step": 30230
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.725891894425953e-05,
+      "loss": 0.3793,
+      "step": 30235
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.725845014298439e-05,
+      "loss": 0.0984,
+      "step": 30240
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.725798134170925e-05,
+      "loss": 0.0687,
+      "step": 30245
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.725751254043411e-05,
+      "loss": 0.0722,
+      "step": 30250
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.7257043739158973e-05,
+      "loss": 0.1206,
+      "step": 30255
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.7256574937883833e-05,
+      "loss": 0.0961,
+      "step": 30260
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.7256106136608693e-05,
+      "loss": 0.1549,
+      "step": 30265
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.7255637335333553e-05,
+      "loss": 0.1804,
+      "step": 30270
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.7255168534058413e-05,
+      "loss": 0.2152,
+      "step": 30275
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.7254699732783276e-05,
+      "loss": 0.3638,
+      "step": 30280
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.7254230931508136e-05,
+      "loss": 0.2409,
+      "step": 30285
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.7253762130232996e-05,
+      "loss": 0.0768,
+      "step": 30290
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.7253293328957856e-05,
+      "loss": 0.0842,
+      "step": 30295
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.7252824527682716e-05,
+      "loss": 0.0726,
+      "step": 30300
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.7252355726407576e-05,
+      "loss": 0.0589,
+      "step": 30305
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.7251886925132436e-05,
+      "loss": 0.2003,
+      "step": 30310
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.7251418123857296e-05,
+      "loss": 0.1153,
+      "step": 30315
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.725094932258216e-05,
+      "loss": 0.1932,
+      "step": 30320
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.725048052130702e-05,
+      "loss": 0.2228,
+      "step": 30325
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.725001172003188e-05,
+      "loss": 0.4502,
+      "step": 30330
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7249542918756742e-05,
+      "loss": 0.2931,
+      "step": 30335
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7249074117481602e-05,
+      "loss": 0.0577,
+      "step": 30340
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7248605316206462e-05,
+      "loss": 0.0496,
+      "step": 30345
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7248136514931322e-05,
+      "loss": 0.1063,
+      "step": 30350
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7247667713656182e-05,
+      "loss": 0.0715,
+      "step": 30355
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7247198912381042e-05,
+      "loss": 0.1267,
+      "step": 30360
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7246730111105902e-05,
+      "loss": 0.1406,
+      "step": 30365
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7246261309830765e-05,
+      "loss": 0.1855,
+      "step": 30370
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7245792508555625e-05,
+      "loss": 0.2098,
+      "step": 30375
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7245323707280485e-05,
+      "loss": 0.294,
+      "step": 30380
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7244854906005345e-05,
+      "loss": 0.3826,
+      "step": 30385
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7244386104730205e-05,
+      "loss": 0.0421,
+      "step": 30390
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7243917303455065e-05,
+      "loss": 0.0705,
+      "step": 30395
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.724344850217993e-05,
+      "loss": 0.0582,
+      "step": 30400
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7242979700904788e-05,
+      "loss": 0.0843,
+      "step": 30405
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7242510899629648e-05,
+      "loss": 0.0893,
+      "step": 30410
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.724204209835451e-05,
+      "loss": 0.0917,
+      "step": 30415
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.724157329707937e-05,
+      "loss": 0.1858,
+      "step": 30420
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.724110449580423e-05,
+      "loss": 0.1572,
+      "step": 30425
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.724063569452909e-05,
+      "loss": 0.3922,
+      "step": 30430
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.724016689325395e-05,
+      "loss": 0.2831,
+      "step": 30435
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.723969809197881e-05,
+      "loss": 0.0452,
+      "step": 30440
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.723922929070367e-05,
+      "loss": 0.0772,
+      "step": 30445
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.723876048942853e-05,
+      "loss": 0.0977,
+      "step": 30450
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.723829168815339e-05,
+      "loss": 0.1266,
+      "step": 30455
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7237822886878254e-05,
+      "loss": 0.1233,
+      "step": 30460
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7237354085603114e-05,
+      "loss": 0.1393,
+      "step": 30465
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7236885284327974e-05,
+      "loss": 0.2311,
+      "step": 30470
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7236416483052834e-05,
+      "loss": 0.2408,
+      "step": 30475
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7235947681777697e-05,
+      "loss": 0.3132,
+      "step": 30480
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7235478880502557e-05,
+      "loss": 0.3054,
+      "step": 30485
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7235010079227417e-05,
+      "loss": 0.0488,
+      "step": 30490
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7234541277952277e-05,
+      "loss": 0.0335,
+      "step": 30495
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7234072476677137e-05,
+      "loss": 0.1066,
+      "step": 30500
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7233603675401997e-05,
+      "loss": 0.1301,
+      "step": 30505
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.723313487412686e-05,
+      "loss": 0.1551,
+      "step": 30510
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.723266607285172e-05,
+      "loss": 0.2577,
+      "step": 30515
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.723219727157658e-05,
+      "loss": 0.2499,
+      "step": 30520
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.723172847030144e-05,
+      "loss": 0.156,
+      "step": 30525
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.72312596690263e-05,
+      "loss": 0.4003,
+      "step": 30530
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.723079086775116e-05,
+      "loss": 0.2101,
+      "step": 30535
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7230322066476023e-05,
+      "loss": 0.0443,
+      "step": 30540
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7229853265200883e-05,
+      "loss": 0.0775,
+      "step": 30545
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7229384463925743e-05,
+      "loss": 0.0966,
+      "step": 30550
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7228915662650603e-05,
+      "loss": 0.1195,
+      "step": 30555
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7228446861375467e-05,
+      "loss": 0.1571,
+      "step": 30560
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7227978060100326e-05,
+      "loss": 0.0814,
+      "step": 30565
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7227509258825186e-05,
+      "loss": 0.2036,
+      "step": 30570
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7227040457550046e-05,
+      "loss": 0.2727,
+      "step": 30575
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7226571656274906e-05,
+      "loss": 0.4466,
+      "step": 30580
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7226102854999766e-05,
+      "loss": 0.209,
+      "step": 30585
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7225634053724626e-05,
+      "loss": 0.0356,
+      "step": 30590
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7225165252449486e-05,
+      "loss": 0.0369,
+      "step": 30595
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7224696451174346e-05,
+      "loss": 0.1753,
+      "step": 30600
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.722422764989921e-05,
+      "loss": 0.1393,
+      "step": 30605
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.722375884862407e-05,
+      "loss": 0.1501,
+      "step": 30610
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.722329004734893e-05,
+      "loss": 0.0778,
+      "step": 30615
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7222821246073793e-05,
+      "loss": 0.2176,
+      "step": 30620
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7222352444798652e-05,
+      "loss": 0.2829,
+      "step": 30625
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7221883643523512e-05,
+      "loss": 0.4749,
+      "step": 30630
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7221414842248372e-05,
+      "loss": 0.1959,
+      "step": 30635
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7220946040973232e-05,
+      "loss": 0.074,
+      "step": 30640
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7220477239698092e-05,
+      "loss": 0.1144,
+      "step": 30645
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7220008438422956e-05,
+      "loss": 0.1019,
+      "step": 30650
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7219539637147815e-05,
+      "loss": 0.0822,
+      "step": 30655
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7219070835872675e-05,
+      "loss": 0.1619,
+      "step": 30660
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7218602034597535e-05,
+      "loss": 0.1428,
+      "step": 30665
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7218133233322395e-05,
+      "loss": 0.1916,
+      "step": 30670
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7217664432047255e-05,
+      "loss": 0.1891,
+      "step": 30675
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7217195630772115e-05,
+      "loss": 0.2762,
+      "step": 30680
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.721672682949698e-05,
+      "loss": 0.1885,
+      "step": 30685
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.721625802822184e-05,
+      "loss": 0.06,
+      "step": 30690
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.72157892269467e-05,
+      "loss": 0.098,
+      "step": 30695
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.721532042567156e-05,
+      "loss": 0.0611,
+      "step": 30700
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.721485162439642e-05,
+      "loss": 0.0604,
+      "step": 30705
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.721438282312128e-05,
+      "loss": 0.1263,
+      "step": 30710
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.721391402184614e-05,
+      "loss": 0.1096,
+      "step": 30715
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7213445220571e-05,
+      "loss": 0.1963,
+      "step": 30720
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.721297641929586e-05,
+      "loss": 0.2184,
+      "step": 30725
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.721250761802072e-05,
+      "loss": 0.3873,
+      "step": 30730
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.721203881674558e-05,
+      "loss": 0.2638,
+      "step": 30735
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7211570015470444e-05,
+      "loss": 0.0434,
+      "step": 30740
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7211101214195304e-05,
+      "loss": 0.0737,
+      "step": 30745
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7210632412920164e-05,
+      "loss": 0.0951,
+      "step": 30750
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7210163611645024e-05,
+      "loss": 0.1319,
+      "step": 30755
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7209694810369884e-05,
+      "loss": 0.1413,
+      "step": 30760
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7209226009094748e-05,
+      "loss": 0.112,
+      "step": 30765
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7208757207819607e-05,
+      "loss": 0.1886,
+      "step": 30770
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7208288406544467e-05,
+      "loss": 0.3033,
+      "step": 30775
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7207819605269327e-05,
+      "loss": 0.3538,
+      "step": 30780
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7207350803994187e-05,
+      "loss": 0.3352,
+      "step": 30785
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.720688200271905e-05,
+      "loss": 0.0869,
+      "step": 30790
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.720641320144391e-05,
+      "loss": 0.034,
+      "step": 30795
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.720594440016877e-05,
+      "loss": 0.1027,
+      "step": 30800
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.720547559889363e-05,
+      "loss": 0.1315,
+      "step": 30805
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.720500679761849e-05,
+      "loss": 0.1198,
+      "step": 30810
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.720453799634335e-05,
+      "loss": 0.1869,
+      "step": 30815
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.720406919506821e-05,
+      "loss": 0.1343,
+      "step": 30820
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.720360039379307e-05,
+      "loss": 0.2136,
+      "step": 30825
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7203131592517933e-05,
+      "loss": 0.3957,
+      "step": 30830
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7202662791242793e-05,
+      "loss": 0.291,
+      "step": 30835
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7202193989967653e-05,
+      "loss": 0.0897,
+      "step": 30840
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7201725188692517e-05,
+      "loss": 0.0911,
+      "step": 30845
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7201256387417377e-05,
+      "loss": 0.1066,
+      "step": 30850
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7200787586142237e-05,
+      "loss": 0.1329,
+      "step": 30855
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7200318784867096e-05,
+      "loss": 0.0502,
+      "step": 30860
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7199849983591956e-05,
+      "loss": 0.0901,
+      "step": 30865
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7199381182316816e-05,
+      "loss": 0.145,
+      "step": 30870
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7198912381041676e-05,
+      "loss": 0.2188,
+      "step": 30875
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.719844357976654e-05,
+      "loss": 0.3924,
+      "step": 30880
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.71979747784914e-05,
+      "loss": 0.2868,
+      "step": 30885
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.719750597721626e-05,
+      "loss": 0.0285,
+      "step": 30890
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.719703717594112e-05,
+      "loss": 0.0456,
+      "step": 30895
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.719656837466598e-05,
+      "loss": 0.0869,
+      "step": 30900
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.719609957339084e-05,
+      "loss": 0.1951,
+      "step": 30905
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7195630772115703e-05,
+      "loss": 0.1306,
+      "step": 30910
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7195161970840562e-05,
+      "loss": 0.1019,
+      "step": 30915
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7194693169565422e-05,
+      "loss": 0.1409,
+      "step": 30920
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7194224368290282e-05,
+      "loss": 0.3232,
+      "step": 30925
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7193755567015146e-05,
+      "loss": 0.2785,
+      "step": 30930
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7193286765740006e-05,
+      "loss": 0.1941,
+      "step": 30935
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7192817964464866e-05,
+      "loss": 0.0544,
+      "step": 30940
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7192349163189725e-05,
+      "loss": 0.0817,
+      "step": 30945
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7191880361914585e-05,
+      "loss": 0.1332,
+      "step": 30950
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7191411560639445e-05,
+      "loss": 0.0841,
+      "step": 30955
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7190942759364305e-05,
+      "loss": 0.148,
+      "step": 30960
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7190473958089165e-05,
+      "loss": 0.2526,
+      "step": 30965
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.719000515681403e-05,
+      "loss": 0.1986,
+      "step": 30970
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.718953635553889e-05,
+      "loss": 0.1943,
+      "step": 30975
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.718906755426375e-05,
+      "loss": 0.361,
+      "step": 30980
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.718859875298861e-05,
+      "loss": 0.1899,
+      "step": 30985
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.718812995171347e-05,
+      "loss": 0.0256,
+      "step": 30990
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.718766115043833e-05,
+      "loss": 0.0823,
+      "step": 30995
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.718719234916319e-05,
+      "loss": 0.1232,
+      "step": 31000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.718672354788805e-05,
+      "loss": 0.0927,
+      "step": 31005
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.718625474661291e-05,
+      "loss": 0.112,
+      "step": 31010
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.718578594533777e-05,
+      "loss": 0.1358,
+      "step": 31015
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7185317144062635e-05,
+      "loss": 0.1677,
+      "step": 31020
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7184848342787495e-05,
+      "loss": 0.2751,
+      "step": 31025
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7184379541512355e-05,
+      "loss": 0.2994,
+      "step": 31030
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7183910740237214e-05,
+      "loss": 0.2568,
+      "step": 31035
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7183441938962074e-05,
+      "loss": 0.0416,
+      "step": 31040
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7182973137686934e-05,
+      "loss": 0.087,
+      "step": 31045
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7182504336411798e-05,
+      "loss": 0.084,
+      "step": 31050
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7182035535136658e-05,
+      "loss": 0.1638,
+      "step": 31055
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7181566733861518e-05,
+      "loss": 0.1059,
+      "step": 31060
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.718109793258638e-05,
+      "loss": 0.0665,
+      "step": 31065
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.718062913131124e-05,
+      "loss": 0.0996,
+      "step": 31070
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.71801603300361e-05,
+      "loss": 0.1804,
+      "step": 31075
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.717969152876096e-05,
+      "loss": 0.3874,
+      "step": 31080
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.717922272748582e-05,
+      "loss": 0.2361,
+      "step": 31085
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.717875392621068e-05,
+      "loss": 0.0242,
+      "step": 31090
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.717828512493554e-05,
+      "loss": 0.0469,
+      "step": 31095
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.71778163236604e-05,
+      "loss": 0.0775,
+      "step": 31100
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.717734752238526e-05,
+      "loss": 0.1437,
+      "step": 31105
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.717687872111012e-05,
+      "loss": 0.1462,
+      "step": 31110
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7176409919834984e-05,
+      "loss": 0.1569,
+      "step": 31115
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7175941118559843e-05,
+      "loss": 0.1341,
+      "step": 31120
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7175472317284703e-05,
+      "loss": 0.2074,
+      "step": 31125
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7175003516009567e-05,
+      "loss": 0.3704,
+      "step": 31130
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7174534714734427e-05,
+      "loss": 0.3183,
+      "step": 31135
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7174065913459287e-05,
+      "loss": 0.0238,
+      "step": 31140
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7173597112184147e-05,
+      "loss": 0.0825,
+      "step": 31145
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7173128310909006e-05,
+      "loss": 0.0863,
+      "step": 31150
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7172659509633866e-05,
+      "loss": 0.1149,
+      "step": 31155
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.717219070835873e-05,
+      "loss": 0.1763,
+      "step": 31160
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.717172190708359e-05,
+      "loss": 0.1381,
+      "step": 31165
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.717125310580845e-05,
+      "loss": 0.1784,
+      "step": 31170
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.717078430453331e-05,
+      "loss": 0.272,
+      "step": 31175
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.717031550325817e-05,
+      "loss": 0.4498,
+      "step": 31180
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.716984670198303e-05,
+      "loss": 0.4483,
+      "step": 31185
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.716937790070789e-05,
+      "loss": 0.0412,
+      "step": 31190
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7168909099432753e-05,
+      "loss": 0.0857,
+      "step": 31195
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7168440298157613e-05,
+      "loss": 0.1485,
+      "step": 31200
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7167971496882473e-05,
+      "loss": 0.1242,
+      "step": 31205
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7167502695607336e-05,
+      "loss": 0.0397,
+      "step": 31210
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7167033894332196e-05,
+      "loss": 0.1162,
+      "step": 31215
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7166565093057056e-05,
+      "loss": 0.0797,
+      "step": 31220
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7166096291781916e-05,
+      "loss": 0.2146,
+      "step": 31225
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7165627490506776e-05,
+      "loss": 0.4251,
+      "step": 31230
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7165158689231636e-05,
+      "loss": 0.2648,
+      "step": 31235
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7164689887956495e-05,
+      "loss": 0.0691,
+      "step": 31240
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7164221086681355e-05,
+      "loss": 0.0339,
+      "step": 31245
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7163752285406215e-05,
+      "loss": 0.0822,
+      "step": 31250
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.716328348413108e-05,
+      "loss": 0.1465,
+      "step": 31255
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.716281468285594e-05,
+      "loss": 0.0895,
+      "step": 31260
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.71623458815808e-05,
+      "loss": 0.093,
+      "step": 31265
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.716187708030566e-05,
+      "loss": 0.1726,
+      "step": 31270
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7161408279030522e-05,
+      "loss": 0.2715,
+      "step": 31275
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7160939477755382e-05,
+      "loss": 0.2339,
+      "step": 31280
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.716047067648024e-05,
+      "loss": 0.424,
+      "step": 31285
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.71600018752051e-05,
+      "loss": 0.0541,
+      "step": 31290
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.715953307392996e-05,
+      "loss": 0.076,
+      "step": 31295
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7159064272654825e-05,
+      "loss": 0.0524,
+      "step": 31300
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7158595471379685e-05,
+      "loss": 0.142,
+      "step": 31305
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7158126670104545e-05,
+      "loss": 0.1337,
+      "step": 31310
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7157657868829405e-05,
+      "loss": 0.171,
+      "step": 31315
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7157189067554265e-05,
+      "loss": 0.1233,
+      "step": 31320
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7156720266279124e-05,
+      "loss": 0.2044,
+      "step": 31325
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7156251465003984e-05,
+      "loss": 0.3685,
+      "step": 31330
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7155782663728844e-05,
+      "loss": 0.2695,
+      "step": 31335
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7155313862453708e-05,
+      "loss": 0.0894,
+      "step": 31340
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7154845061178568e-05,
+      "loss": 0.0491,
+      "step": 31345
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7154376259903428e-05,
+      "loss": 0.0855,
+      "step": 31350
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.715390745862829e-05,
+      "loss": 0.0858,
+      "step": 31355
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.715343865735315e-05,
+      "loss": 0.1007,
+      "step": 31360
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.715296985607801e-05,
+      "loss": 0.247,
+      "step": 31365
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.715250105480287e-05,
+      "loss": 0.1082,
+      "step": 31370
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.715203225352773e-05,
+      "loss": 0.2143,
+      "step": 31375
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.715156345225259e-05,
+      "loss": 0.4156,
+      "step": 31380
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.715109465097745e-05,
+      "loss": 0.3131,
+      "step": 31385
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7150625849702314e-05,
+      "loss": 0.038,
+      "step": 31390
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7150157048427174e-05,
+      "loss": 0.0726,
+      "step": 31395
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7149688247152034e-05,
+      "loss": 0.0399,
+      "step": 31400
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7149219445876894e-05,
+      "loss": 0.1019,
+      "step": 31405
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7148750644601754e-05,
+      "loss": 0.0969,
+      "step": 31410
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7148281843326617e-05,
+      "loss": 0.1317,
+      "step": 31415
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7147813042051477e-05,
+      "loss": 0.1599,
+      "step": 31420
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7147344240776337e-05,
+      "loss": 0.1803,
+      "step": 31425
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7146875439501197e-05,
+      "loss": 0.2713,
+      "step": 31430
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7146406638226057e-05,
+      "loss": 0.348,
+      "step": 31435
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.714593783695092e-05,
+      "loss": 0.0418,
+      "step": 31440
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.714546903567578e-05,
+      "loss": 0.0582,
+      "step": 31445
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.714500023440064e-05,
+      "loss": 0.0857,
+      "step": 31450
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.71445314331255e-05,
+      "loss": 0.1093,
+      "step": 31455
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.714406263185036e-05,
+      "loss": 0.143,
+      "step": 31460
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.714359383057522e-05,
+      "loss": 0.2097,
+      "step": 31465
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.714312502930008e-05,
+      "loss": 0.2391,
+      "step": 31470
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.714265622802494e-05,
+      "loss": 0.2575,
+      "step": 31475
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7142187426749803e-05,
+      "loss": 0.3507,
+      "step": 31480
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7141718625474663e-05,
+      "loss": 0.2108,
+      "step": 31485
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7141249824199523e-05,
+      "loss": 0.1031,
+      "step": 31490
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7140781022924386e-05,
+      "loss": 0.0743,
+      "step": 31495
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7140312221649246e-05,
+      "loss": 0.1272,
+      "step": 31500
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7139843420374106e-05,
+      "loss": 0.1309,
+      "step": 31505
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7139374619098966e-05,
+      "loss": 0.1227,
+      "step": 31510
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7138905817823826e-05,
+      "loss": 0.125,
+      "step": 31515
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7138437016548686e-05,
+      "loss": 0.1504,
+      "step": 31520
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7137968215273546e-05,
+      "loss": 0.1449,
+      "step": 31525
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.713749941399841e-05,
+      "loss": 0.338,
+      "step": 31530
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.713703061272327e-05,
+      "loss": 0.2274,
+      "step": 31535
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.713656181144813e-05,
+      "loss": 0.3505,
+      "step": 31540
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.713609301017299e-05,
+      "loss": 0.0287,
+      "step": 31545
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.713562420889785e-05,
+      "loss": 0.0951,
+      "step": 31550
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.713515540762271e-05,
+      "loss": 0.0786,
+      "step": 31555
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7134686606347572e-05,
+      "loss": 0.0759,
+      "step": 31560
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7134217805072432e-05,
+      "loss": 0.1229,
+      "step": 31565
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7133749003797292e-05,
+      "loss": 0.2114,
+      "step": 31570
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.713328020252215e-05,
+      "loss": 0.1955,
+      "step": 31575
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7132811401247015e-05,
+      "loss": 0.2346,
+      "step": 31580
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7132342599971875e-05,
+      "loss": 0.1724,
+      "step": 31585
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7131873798696735e-05,
+      "loss": 0.0742,
+      "step": 31590
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7131404997421595e-05,
+      "loss": 0.0828,
+      "step": 31595
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7130936196146455e-05,
+      "loss": 0.0835,
+      "step": 31600
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7130467394871315e-05,
+      "loss": 0.1199,
+      "step": 31605
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7129998593596175e-05,
+      "loss": 0.1261,
+      "step": 31610
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7129529792321035e-05,
+      "loss": 0.096,
+      "step": 31615
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7129060991045894e-05,
+      "loss": 0.2419,
+      "step": 31620
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7128592189770758e-05,
+      "loss": 0.225,
+      "step": 31625
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7128123388495618e-05,
+      "loss": 0.286,
+      "step": 31630
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7127654587220478e-05,
+      "loss": 0.2153,
+      "step": 31635
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.712718578594534e-05,
+      "loss": 0.0429,
+      "step": 31640
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.71267169846702e-05,
+      "loss": 0.0757,
+      "step": 31645
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.712624818339506e-05,
+      "loss": 0.0863,
+      "step": 31650
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.712577938211992e-05,
+      "loss": 0.0917,
+      "step": 31655
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.712531058084478e-05,
+      "loss": 0.104,
+      "step": 31660
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.712484177956964e-05,
+      "loss": 0.1161,
+      "step": 31665
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7124372978294504e-05,
+      "loss": 0.1551,
+      "step": 31670
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7123904177019364e-05,
+      "loss": 0.2739,
+      "step": 31675
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7123435375744224e-05,
+      "loss": 0.4039,
+      "step": 31680
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7122966574469084e-05,
+      "loss": 0.2652,
+      "step": 31685
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7122497773193944e-05,
+      "loss": 0.0578,
+      "step": 31690
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7122028971918804e-05,
+      "loss": 0.0959,
+      "step": 31695
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7121560170643664e-05,
+      "loss": 0.1124,
+      "step": 31700
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7121091369368527e-05,
+      "loss": 0.0831,
+      "step": 31705
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7120622568093387e-05,
+      "loss": 0.1715,
+      "step": 31710
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7120153766818247e-05,
+      "loss": 0.1528,
+      "step": 31715
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.711968496554311e-05,
+      "loss": 0.1222,
+      "step": 31720
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.711921616426797e-05,
+      "loss": 0.3348,
+      "step": 31725
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.711874736299283e-05,
+      "loss": 0.2315,
+      "step": 31730
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.711827856171769e-05,
+      "loss": 0.3509,
+      "step": 31735
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.711780976044255e-05,
+      "loss": 0.025,
+      "step": 31740
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.711734095916741e-05,
+      "loss": 0.0601,
+      "step": 31745
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.711687215789227e-05,
+      "loss": 0.0986,
+      "step": 31750
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.711640335661713e-05,
+      "loss": 0.117,
+      "step": 31755
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.711593455534199e-05,
+      "loss": 0.0785,
+      "step": 31760
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7115465754066853e-05,
+      "loss": 0.1231,
+      "step": 31765
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7114996952791713e-05,
+      "loss": 0.1392,
+      "step": 31770
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7114528151516573e-05,
+      "loss": 0.1812,
+      "step": 31775
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7114059350241433e-05,
+      "loss": 0.2136,
+      "step": 31780
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7113590548966296e-05,
+      "loss": 0.257,
+      "step": 31785
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7113121747691156e-05,
+      "loss": 0.025,
+      "step": 31790
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7112652946416016e-05,
+      "loss": 0.0484,
+      "step": 31795
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7112184145140876e-05,
+      "loss": 0.0609,
+      "step": 31800
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7111715343865736e-05,
+      "loss": 0.1174,
+      "step": 31805
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.71112465425906e-05,
+      "loss": 0.0714,
+      "step": 31810
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.711077774131546e-05,
+      "loss": 0.111,
+      "step": 31815
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.711030894004032e-05,
+      "loss": 0.1114,
+      "step": 31820
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.710984013876518e-05,
+      "loss": 0.207,
+      "step": 31825
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.710937133749004e-05,
+      "loss": 0.299,
+      "step": 31830
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.71089025362149e-05,
+      "loss": 0.4295,
+      "step": 31835
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.710843373493976e-05,
+      "loss": 0.1151,
+      "step": 31840
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7107964933664622e-05,
+      "loss": 0.1369,
+      "step": 31845
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7107496132389482e-05,
+      "loss": 0.0604,
+      "step": 31850
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7107027331114342e-05,
+      "loss": 0.1176,
+      "step": 31855
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7106558529839202e-05,
+      "loss": 0.0861,
+      "step": 31860
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7106089728564065e-05,
+      "loss": 0.1066,
+      "step": 31865
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7105620927288925e-05,
+      "loss": 0.1939,
+      "step": 31870
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7105152126013785e-05,
+      "loss": 0.1461,
+      "step": 31875
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7104683324738645e-05,
+      "loss": 0.2631,
+      "step": 31880
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7104214523463505e-05,
+      "loss": 0.2348,
+      "step": 31885
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7103745722188365e-05,
+      "loss": 0.0548,
+      "step": 31890
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7103276920913225e-05,
+      "loss": 0.085,
+      "step": 31895
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7102808119638085e-05,
+      "loss": 0.0418,
+      "step": 31900
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7102339318362948e-05,
+      "loss": 0.1059,
+      "step": 31905
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7101870517087808e-05,
+      "loss": 0.1142,
+      "step": 31910
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7101401715812668e-05,
+      "loss": 0.1028,
+      "step": 31915
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7100932914537528e-05,
+      "loss": 0.1925,
+      "step": 31920
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.710046411326239e-05,
+      "loss": 0.1953,
+      "step": 31925
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.709999531198725e-05,
+      "loss": 0.3952,
+      "step": 31930
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.709952651071211e-05,
+      "loss": 0.316,
+      "step": 31935
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.709905770943697e-05,
+      "loss": 0.0727,
+      "step": 31940
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.709858890816183e-05,
+      "loss": 0.0474,
+      "step": 31945
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7098120106886694e-05,
+      "loss": 0.0985,
+      "step": 31950
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7097651305611554e-05,
+      "loss": 0.1191,
+      "step": 31955
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7097182504336414e-05,
+      "loss": 0.1053,
+      "step": 31960
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7096713703061274e-05,
+      "loss": 0.1443,
+      "step": 31965
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7096244901786134e-05,
+      "loss": 0.1016,
+      "step": 31970
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7095776100510994e-05,
+      "loss": 0.2535,
+      "step": 31975
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7095307299235854e-05,
+      "loss": 0.2324,
+      "step": 31980
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7094838497960714e-05,
+      "loss": 0.3434,
+      "step": 31985
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7094369696685577e-05,
+      "loss": 0.057,
+      "step": 31990
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7093900895410437e-05,
+      "loss": 0.1534,
+      "step": 31995
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7093432094135297e-05,
+      "loss": 0.0607,
+      "step": 32000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.709296329286016e-05,
+      "loss": 0.0599,
+      "step": 32005
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.709249449158502e-05,
+      "loss": 0.1019,
+      "step": 32010
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.709202569030988e-05,
+      "loss": 0.0972,
+      "step": 32015
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.709155688903474e-05,
+      "loss": 0.1218,
+      "step": 32020
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.70910880877596e-05,
+      "loss": 0.234,
+      "step": 32025
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.709061928648446e-05,
+      "loss": 0.3776,
+      "step": 32030
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.709015048520932e-05,
+      "loss": 0.2142,
+      "step": 32035
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7089681683934183e-05,
+      "loss": 0.0744,
+      "step": 32040
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7089212882659043e-05,
+      "loss": 0.0301,
+      "step": 32045
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7088744081383903e-05,
+      "loss": 0.1003,
+      "step": 32050
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7088275280108763e-05,
+      "loss": 0.0953,
+      "step": 32055
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7087806478833623e-05,
+      "loss": 0.1135,
+      "step": 32060
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7087337677558483e-05,
+      "loss": 0.097,
+      "step": 32065
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7086868876283346e-05,
+      "loss": 0.2052,
+      "step": 32070
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7086400075008206e-05,
+      "loss": 0.1643,
+      "step": 32075
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7085931273733066e-05,
+      "loss": 0.301,
+      "step": 32080
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7085462472457926e-05,
+      "loss": 0.389,
+      "step": 32085
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.708499367118279e-05,
+      "loss": 0.0549,
+      "step": 32090
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.708452486990765e-05,
+      "loss": 0.0714,
+      "step": 32095
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.708405606863251e-05,
+      "loss": 0.1293,
+      "step": 32100
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.708358726735737e-05,
+      "loss": 0.0897,
+      "step": 32105
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.708311846608223e-05,
+      "loss": 0.162,
+      "step": 32110
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.708264966480709e-05,
+      "loss": 0.1221,
+      "step": 32115
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.708218086353195e-05,
+      "loss": 0.1646,
+      "step": 32120
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.708171206225681e-05,
+      "loss": 0.271,
+      "step": 32125
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.708124326098167e-05,
+      "loss": 0.3324,
+      "step": 32130
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7080774459706532e-05,
+      "loss": 0.295,
+      "step": 32135
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7080305658431392e-05,
+      "loss": 0.0784,
+      "step": 32140
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7079836857156252e-05,
+      "loss": 0.0895,
+      "step": 32145
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7079368055881115e-05,
+      "loss": 0.0987,
+      "step": 32150
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7078899254605975e-05,
+      "loss": 0.1247,
+      "step": 32155
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7078430453330835e-05,
+      "loss": 0.145,
+      "step": 32160
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7077961652055695e-05,
+      "loss": 0.0755,
+      "step": 32165
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7077492850780555e-05,
+      "loss": 0.1894,
+      "step": 32170
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7077024049505415e-05,
+      "loss": 0.2969,
+      "step": 32175
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7076555248230278e-05,
+      "loss": 0.3833,
+      "step": 32180
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7076086446955138e-05,
+      "loss": 0.2779,
+      "step": 32185
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7075617645679998e-05,
+      "loss": 0.0559,
+      "step": 32190
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7075148844404858e-05,
+      "loss": 0.0812,
+      "step": 32195
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7074680043129718e-05,
+      "loss": 0.0472,
+      "step": 32200
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7074211241854578e-05,
+      "loss": 0.0631,
+      "step": 32205
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7073742440579438e-05,
+      "loss": 0.1842,
+      "step": 32210
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.70732736393043e-05,
+      "loss": 0.1205,
+      "step": 32215
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.707280483802916e-05,
+      "loss": 0.1232,
+      "step": 32220
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.707233603675402e-05,
+      "loss": 0.1249,
+      "step": 32225
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7071867235478884e-05,
+      "loss": 0.3041,
+      "step": 32230
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7071398434203744e-05,
+      "loss": 0.3015,
+      "step": 32235
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7070929632928604e-05,
+      "loss": 0.0641,
+      "step": 32240
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7070460831653464e-05,
+      "loss": 0.0933,
+      "step": 32245
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7069992030378324e-05,
+      "loss": 0.0691,
+      "step": 32250
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7069523229103184e-05,
+      "loss": 0.1205,
+      "step": 32255
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7069054427828044e-05,
+      "loss": 0.1456,
+      "step": 32260
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7068585626552904e-05,
+      "loss": 0.0755,
+      "step": 32265
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7068116825277764e-05,
+      "loss": 0.1583,
+      "step": 32270
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7067648024002627e-05,
+      "loss": 0.2089,
+      "step": 32275
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7067179222727487e-05,
+      "loss": 0.3421,
+      "step": 32280
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7066710421452347e-05,
+      "loss": 0.1861,
+      "step": 32285
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7066241620177207e-05,
+      "loss": 0.0438,
+      "step": 32290
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.706577281890207e-05,
+      "loss": 0.1262,
+      "step": 32295
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.706530401762693e-05,
+      "loss": 0.0893,
+      "step": 32300
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.706483521635179e-05,
+      "loss": 0.0865,
+      "step": 32305
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.706436641507665e-05,
+      "loss": 0.1191,
+      "step": 32310
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.706389761380151e-05,
+      "loss": 0.1458,
+      "step": 32315
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7063428812526373e-05,
+      "loss": 0.1978,
+      "step": 32320
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7062960011251233e-05,
+      "loss": 0.197,
+      "step": 32325
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7062491209976093e-05,
+      "loss": 0.3146,
+      "step": 32330
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7062022408700953e-05,
+      "loss": 0.2121,
+      "step": 32335
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7061553607425813e-05,
+      "loss": 0.07,
+      "step": 32340
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7061084806150673e-05,
+      "loss": 0.1082,
+      "step": 32345
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7060616004875533e-05,
+      "loss": 0.1035,
+      "step": 32350
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7060147203600396e-05,
+      "loss": 0.1162,
+      "step": 32355
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7059678402325256e-05,
+      "loss": 0.1757,
+      "step": 32360
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7059209601050116e-05,
+      "loss": 0.1391,
+      "step": 32365
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7058740799774976e-05,
+      "loss": 0.1575,
+      "step": 32370
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.705827199849984e-05,
+      "loss": 0.1514,
+      "step": 32375
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.70578031972247e-05,
+      "loss": 0.2948,
+      "step": 32380
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.705733439594956e-05,
+      "loss": 0.2625,
+      "step": 32385
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.705686559467442e-05,
+      "loss": 0.0237,
+      "step": 32390
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.705639679339928e-05,
+      "loss": 0.0852,
+      "step": 32395
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.705592799212414e-05,
+      "loss": 0.1271,
+      "step": 32400
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7055459190849e-05,
+      "loss": 0.0789,
+      "step": 32405
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.705499038957386e-05,
+      "loss": 0.1637,
+      "step": 32410
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7054521588298722e-05,
+      "loss": 0.1025,
+      "step": 32415
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7054052787023582e-05,
+      "loss": 0.1824,
+      "step": 32420
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7053583985748442e-05,
+      "loss": 0.2877,
+      "step": 32425
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7053115184473302e-05,
+      "loss": 0.3047,
+      "step": 32430
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7052646383198165e-05,
+      "loss": 0.2598,
+      "step": 32435
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7052177581923025e-05,
+      "loss": 0.0308,
+      "step": 32440
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7051708780647885e-05,
+      "loss": 0.0843,
+      "step": 32445
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7051239979372745e-05,
+      "loss": 0.0413,
+      "step": 32450
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7050771178097605e-05,
+      "loss": 0.1075,
+      "step": 32455
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.705030237682247e-05,
+      "loss": 0.0626,
+      "step": 32460
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7049833575547328e-05,
+      "loss": 0.134,
+      "step": 32465
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7049364774272188e-05,
+      "loss": 0.2199,
+      "step": 32470
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7048895972997048e-05,
+      "loss": 0.1289,
+      "step": 32475
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7048427171721908e-05,
+      "loss": 0.3547,
+      "step": 32480
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7047958370446768e-05,
+      "loss": 0.21,
+      "step": 32485
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7047489569171628e-05,
+      "loss": 0.0306,
+      "step": 32490
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7047020767896488e-05,
+      "loss": 0.0337,
+      "step": 32495
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.704655196662135e-05,
+      "loss": 0.0887,
+      "step": 32500
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.704608316534621e-05,
+      "loss": 0.1361,
+      "step": 32505
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.704561436407107e-05,
+      "loss": 0.1427,
+      "step": 32510
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7045145562795934e-05,
+      "loss": 0.1748,
+      "step": 32515
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7044676761520794e-05,
+      "loss": 0.2038,
+      "step": 32520
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7044207960245654e-05,
+      "loss": 0.1809,
+      "step": 32525
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7043739158970514e-05,
+      "loss": 0.2253,
+      "step": 32530
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7043270357695374e-05,
+      "loss": 0.3482,
+      "step": 32535
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7042801556420234e-05,
+      "loss": 0.1008,
+      "step": 32540
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7042332755145094e-05,
+      "loss": 0.0441,
+      "step": 32545
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7041863953869954e-05,
+      "loss": 0.1013,
+      "step": 32550
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7041395152594817e-05,
+      "loss": 0.0886,
+      "step": 32555
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7040926351319677e-05,
+      "loss": 0.074,
+      "step": 32560
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7040457550044537e-05,
+      "loss": 0.1257,
+      "step": 32565
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7039988748769397e-05,
+      "loss": 0.1518,
+      "step": 32570
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7039519947494257e-05,
+      "loss": 0.2869,
+      "step": 32575
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.703905114621912e-05,
+      "loss": 0.295,
+      "step": 32580
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.703858234494398e-05,
+      "loss": 0.2024,
+      "step": 32585
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.703811354366884e-05,
+      "loss": 0.1001,
+      "step": 32590
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.70376447423937e-05,
+      "loss": 0.0773,
+      "step": 32595
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7037175941118563e-05,
+      "loss": 0.0731,
+      "step": 32600
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7036707139843423e-05,
+      "loss": 0.0798,
+      "step": 32605
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7036238338568283e-05,
+      "loss": 0.1625,
+      "step": 32610
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7035769537293143e-05,
+      "loss": 0.2427,
+      "step": 32615
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7035300736018003e-05,
+      "loss": 0.1587,
+      "step": 32620
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7034831934742863e-05,
+      "loss": 0.1713,
+      "step": 32625
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7034363133467723e-05,
+      "loss": 0.3759,
+      "step": 32630
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7033894332192583e-05,
+      "loss": 0.3358,
+      "step": 32635
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7033425530917443e-05,
+      "loss": 0.0724,
+      "step": 32640
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7032956729642306e-05,
+      "loss": 0.0727,
+      "step": 32645
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7032487928367166e-05,
+      "loss": 0.0715,
+      "step": 32650
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7032019127092026e-05,
+      "loss": 0.0836,
+      "step": 32655
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.703155032581689e-05,
+      "loss": 0.1122,
+      "step": 32660
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.703108152454175e-05,
+      "loss": 0.2039,
+      "step": 32665
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.703061272326661e-05,
+      "loss": 0.2377,
+      "step": 32670
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.703014392199147e-05,
+      "loss": 0.198,
+      "step": 32675
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.702967512071633e-05,
+      "loss": 0.2264,
+      "step": 32680
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.702920631944119e-05,
+      "loss": 0.2453,
+      "step": 32685
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.702873751816605e-05,
+      "loss": 0.0258,
+      "step": 32690
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7028268716890912e-05,
+      "loss": 0.0916,
+      "step": 32695
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7027799915615772e-05,
+      "loss": 0.075,
+      "step": 32700
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7027331114340632e-05,
+      "loss": 0.0646,
+      "step": 32705
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7026862313065492e-05,
+      "loss": 0.0724,
+      "step": 32710
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7026393511790352e-05,
+      "loss": 0.1197,
+      "step": 32715
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7025924710515212e-05,
+      "loss": 0.1027,
+      "step": 32720
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7025455909240075e-05,
+      "loss": 0.1982,
+      "step": 32725
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7024987107964935e-05,
+      "loss": 0.5231,
+      "step": 32730
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7024518306689795e-05,
+      "loss": 0.2235,
+      "step": 32735
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.702404950541466e-05,
+      "loss": 0.0613,
+      "step": 32740
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.702358070413952e-05,
+      "loss": 0.0882,
+      "step": 32745
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.702311190286438e-05,
+      "loss": 0.1131,
+      "step": 32750
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7022643101589238e-05,
+      "loss": 0.0765,
+      "step": 32755
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7022174300314098e-05,
+      "loss": 0.1578,
+      "step": 32760
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7021705499038958e-05,
+      "loss": 0.1737,
+      "step": 32765
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7021236697763818e-05,
+      "loss": 0.1597,
+      "step": 32770
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7020767896488678e-05,
+      "loss": 0.1821,
+      "step": 32775
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7020299095213538e-05,
+      "loss": 0.3385,
+      "step": 32780
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.70198302939384e-05,
+      "loss": 0.2524,
+      "step": 32785
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.701936149266326e-05,
+      "loss": 0.0534,
+      "step": 32790
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.701889269138812e-05,
+      "loss": 0.0347,
+      "step": 32795
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7018423890112984e-05,
+      "loss": 0.0499,
+      "step": 32800
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7017955088837844e-05,
+      "loss": 0.1067,
+      "step": 32805
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7017486287562704e-05,
+      "loss": 0.1506,
+      "step": 32810
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7017017486287564e-05,
+      "loss": 0.1958,
+      "step": 32815
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7016548685012424e-05,
+      "loss": 0.1141,
+      "step": 32820
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7016079883737284e-05,
+      "loss": 0.3248,
+      "step": 32825
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7015611082462147e-05,
+      "loss": 0.3599,
+      "step": 32830
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7015142281187007e-05,
+      "loss": 0.2578,
+      "step": 32835
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7014673479911867e-05,
+      "loss": 0.0727,
+      "step": 32840
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7014204678636727e-05,
+      "loss": 0.057,
+      "step": 32845
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7013735877361587e-05,
+      "loss": 0.0768,
+      "step": 32850
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7013267076086447e-05,
+      "loss": 0.1104,
+      "step": 32855
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7012798274811307e-05,
+      "loss": 0.18,
+      "step": 32860
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.701232947353617e-05,
+      "loss": 0.1028,
+      "step": 32865
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.701186067226103e-05,
+      "loss": 0.298,
+      "step": 32870
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.701139187098589e-05,
+      "loss": 0.1878,
+      "step": 32875
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7010923069710754e-05,
+      "loss": 0.3387,
+      "step": 32880
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7010454268435614e-05,
+      "loss": 0.2769,
+      "step": 32885
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7009985467160473e-05,
+      "loss": 0.0933,
+      "step": 32890
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7009516665885333e-05,
+      "loss": 0.0418,
+      "step": 32895
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.7009047864610193e-05,
+      "loss": 0.0895,
+      "step": 32900
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.7008579063335053e-05,
+      "loss": 0.1456,
+      "step": 32905
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.7008110262059913e-05,
+      "loss": 0.0858,
+      "step": 32910
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.7007641460784773e-05,
+      "loss": 0.0988,
+      "step": 32915
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.7007172659509633e-05,
+      "loss": 0.1303,
+      "step": 32920
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.7006703858234496e-05,
+      "loss": 0.193,
+      "step": 32925
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.7006235056959356e-05,
+      "loss": 0.3254,
+      "step": 32930
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.7005766255684216e-05,
+      "loss": 0.1753,
+      "step": 32935
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.7005297454409076e-05,
+      "loss": 0.0474,
+      "step": 32940
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.700482865313394e-05,
+      "loss": 0.0911,
+      "step": 32945
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.70043598518588e-05,
+      "loss": 0.1147,
+      "step": 32950
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.700389105058366e-05,
+      "loss": 0.1001,
+      "step": 32955
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.700342224930852e-05,
+      "loss": 0.1293,
+      "step": 32960
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.700295344803338e-05,
+      "loss": 0.1598,
+      "step": 32965
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.7002484646758243e-05,
+      "loss": 0.1911,
+      "step": 32970
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.7002015845483102e-05,
+      "loss": 0.2595,
+      "step": 32975
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.7001547044207962e-05,
+      "loss": 0.2554,
+      "step": 32980
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.7001078242932822e-05,
+      "loss": 0.2857,
+      "step": 32985
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.7000609441657682e-05,
+      "loss": 0.0408,
+      "step": 32990
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.7000140640382542e-05,
+      "loss": 0.0558,
+      "step": 32995
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6999671839107402e-05,
+      "loss": 0.0425,
+      "step": 33000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6999203037832262e-05,
+      "loss": 0.1221,
+      "step": 33005
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6998734236557125e-05,
+      "loss": 0.0814,
+      "step": 33010
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6998265435281985e-05,
+      "loss": 0.1134,
+      "step": 33015
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6997796634006845e-05,
+      "loss": 0.1492,
+      "step": 33020
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.699732783273171e-05,
+      "loss": 0.2222,
+      "step": 33025
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.699685903145657e-05,
+      "loss": 0.2727,
+      "step": 33030
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.699639023018143e-05,
+      "loss": 0.2581,
+      "step": 33035
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.699592142890629e-05,
+      "loss": 0.0504,
+      "step": 33040
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.699545262763115e-05,
+      "loss": 0.0664,
+      "step": 33045
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6994983826356008e-05,
+      "loss": 0.0636,
+      "step": 33050
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6994515025080868e-05,
+      "loss": 0.1054,
+      "step": 33055
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6994046223805728e-05,
+      "loss": 0.0915,
+      "step": 33060
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.699357742253059e-05,
+      "loss": 0.1315,
+      "step": 33065
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.699310862125545e-05,
+      "loss": 0.1195,
+      "step": 33070
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.699263981998031e-05,
+      "loss": 0.2035,
+      "step": 33075
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.699217101870517e-05,
+      "loss": 0.3399,
+      "step": 33080
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.699170221743003e-05,
+      "loss": 0.4258,
+      "step": 33085
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6991233416154895e-05,
+      "loss": 0.0815,
+      "step": 33090
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6990764614879754e-05,
+      "loss": 0.0727,
+      "step": 33095
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6990295813604614e-05,
+      "loss": 0.0544,
+      "step": 33100
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6989827012329474e-05,
+      "loss": 0.0978,
+      "step": 33105
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6989358211054338e-05,
+      "loss": 0.0592,
+      "step": 33110
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6988889409779198e-05,
+      "loss": 0.075,
+      "step": 33115
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6988420608504058e-05,
+      "loss": 0.2028,
+      "step": 33120
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6987951807228917e-05,
+      "loss": 0.1716,
+      "step": 33125
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6987483005953777e-05,
+      "loss": 0.4326,
+      "step": 33130
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6987014204678637e-05,
+      "loss": 0.1933,
+      "step": 33135
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6986545403403497e-05,
+      "loss": 0.0459,
+      "step": 33140
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6986076602128357e-05,
+      "loss": 0.0417,
+      "step": 33145
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.698560780085322e-05,
+      "loss": 0.0917,
+      "step": 33150
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.698513899957808e-05,
+      "loss": 0.1252,
+      "step": 33155
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.698467019830294e-05,
+      "loss": 0.132,
+      "step": 33160
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.69842013970278e-05,
+      "loss": 0.1409,
+      "step": 33165
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6983732595752664e-05,
+      "loss": 0.1662,
+      "step": 33170
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6983263794477524e-05,
+      "loss": 0.1353,
+      "step": 33175
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6982794993202383e-05,
+      "loss": 0.3043,
+      "step": 33180
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6982326191927243e-05,
+      "loss": 0.3157,
+      "step": 33185
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6981857390652103e-05,
+      "loss": 0.041,
+      "step": 33190
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6981388589376963e-05,
+      "loss": 0.0717,
+      "step": 33195
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6980919788101823e-05,
+      "loss": 0.1099,
+      "step": 33200
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6980450986826687e-05,
+      "loss": 0.0831,
+      "step": 33205
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6979982185551546e-05,
+      "loss": 0.084,
+      "step": 33210
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6979513384276406e-05,
+      "loss": 0.1968,
+      "step": 33215
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6979044583001266e-05,
+      "loss": 0.1033,
+      "step": 33220
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6978575781726126e-05,
+      "loss": 0.1612,
+      "step": 33225
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.697810698045099e-05,
+      "loss": 0.4094,
+      "step": 33230
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.697763817917585e-05,
+      "loss": 0.4097,
+      "step": 33235
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.697716937790071e-05,
+      "loss": 0.0522,
+      "step": 33240
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.697670057662557e-05,
+      "loss": 0.0752,
+      "step": 33245
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6976231775350433e-05,
+      "loss": 0.0983,
+      "step": 33250
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6975762974075293e-05,
+      "loss": 0.0858,
+      "step": 33255
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6975294172800153e-05,
+      "loss": 0.0756,
+      "step": 33260
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6974825371525013e-05,
+      "loss": 0.1873,
+      "step": 33265
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6974356570249872e-05,
+      "loss": 0.2313,
+      "step": 33270
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6973887768974732e-05,
+      "loss": 0.196,
+      "step": 33275
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6973418967699592e-05,
+      "loss": 0.3991,
+      "step": 33280
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6972950166424452e-05,
+      "loss": 0.3086,
+      "step": 33285
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6972481365149312e-05,
+      "loss": 0.0442,
+      "step": 33290
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6972012563874176e-05,
+      "loss": 0.0435,
+      "step": 33295
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6971543762599035e-05,
+      "loss": 0.0705,
+      "step": 33300
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6971074961323895e-05,
+      "loss": 0.1096,
+      "step": 33305
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.697060616004876e-05,
+      "loss": 0.1128,
+      "step": 33310
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.697013735877362e-05,
+      "loss": 0.1482,
+      "step": 33315
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.696966855749848e-05,
+      "loss": 0.2205,
+      "step": 33320
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.696919975622334e-05,
+      "loss": 0.2817,
+      "step": 33325
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.69687309549482e-05,
+      "loss": 0.3504,
+      "step": 33330
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.696826215367306e-05,
+      "loss": 0.3524,
+      "step": 33335
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6967793352397918e-05,
+      "loss": 0.0259,
+      "step": 33340
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.696732455112278e-05,
+      "loss": 0.1418,
+      "step": 33345
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.696685574984764e-05,
+      "loss": 0.0996,
+      "step": 33350
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.69663869485725e-05,
+      "loss": 0.0931,
+      "step": 33355
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.696591814729736e-05,
+      "loss": 0.1358,
+      "step": 33360
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.696544934602222e-05,
+      "loss": 0.162,
+      "step": 33365
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.696498054474708e-05,
+      "loss": 0.1283,
+      "step": 33370
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6964511743471945e-05,
+      "loss": 0.192,
+      "step": 33375
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6964042942196805e-05,
+      "loss": 0.386,
+      "step": 33380
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6963574140921664e-05,
+      "loss": 0.2255,
+      "step": 33385
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6963105339646528e-05,
+      "loss": 0.0262,
+      "step": 33390
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6962636538371388e-05,
+      "loss": 0.0905,
+      "step": 33395
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6962167737096248e-05,
+      "loss": 0.0624,
+      "step": 33400
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6961698935821108e-05,
+      "loss": 0.1754,
+      "step": 33405
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6961230134545968e-05,
+      "loss": 0.0953,
+      "step": 33410
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6960761333270827e-05,
+      "loss": 0.1157,
+      "step": 33415
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6960292531995687e-05,
+      "loss": 0.1889,
+      "step": 33420
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6959823730720547e-05,
+      "loss": 0.2906,
+      "step": 33425
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6959354929445407e-05,
+      "loss": 0.3266,
+      "step": 33430
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.695888612817027e-05,
+      "loss": 0.278,
+      "step": 33435
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.695841732689513e-05,
+      "loss": 0.0655,
+      "step": 33440
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.695794852561999e-05,
+      "loss": 0.138,
+      "step": 33445
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.695747972434485e-05,
+      "loss": 0.0724,
+      "step": 33450
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6957010923069714e-05,
+      "loss": 0.0803,
+      "step": 33455
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6956542121794574e-05,
+      "loss": 0.1474,
+      "step": 33460
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6956073320519434e-05,
+      "loss": 0.1498,
+      "step": 33465
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6955604519244294e-05,
+      "loss": 0.195,
+      "step": 33470
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6955135717969153e-05,
+      "loss": 0.2709,
+      "step": 33475
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6954666916694017e-05,
+      "loss": 0.318,
+      "step": 33480
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6954198115418877e-05,
+      "loss": 0.3973,
+      "step": 33485
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6953729314143737e-05,
+      "loss": 0.0661,
+      "step": 33490
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6953260512868597e-05,
+      "loss": 0.0608,
+      "step": 33495
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6952791711593457e-05,
+      "loss": 0.0867,
+      "step": 33500
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6952322910318316e-05,
+      "loss": 0.1162,
+      "step": 33505
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6951854109043176e-05,
+      "loss": 0.1134,
+      "step": 33510
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6951385307768036e-05,
+      "loss": 0.0616,
+      "step": 33515
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.69509165064929e-05,
+      "loss": 0.1131,
+      "step": 33520
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.695044770521776e-05,
+      "loss": 0.2365,
+      "step": 33525
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.694997890394262e-05,
+      "loss": 0.3303,
+      "step": 33530
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6949510102667483e-05,
+      "loss": 0.2189,
+      "step": 33535
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6949041301392343e-05,
+      "loss": 0.0139,
+      "step": 33540
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6948572500117203e-05,
+      "loss": 0.0497,
+      "step": 33545
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6948103698842063e-05,
+      "loss": 0.1044,
+      "step": 33550
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6947634897566923e-05,
+      "loss": 0.0744,
+      "step": 33555
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6947166096291782e-05,
+      "loss": 0.1061,
+      "step": 33560
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6946697295016642e-05,
+      "loss": 0.1483,
+      "step": 33565
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6946228493741502e-05,
+      "loss": 0.1121,
+      "step": 33570
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6945759692466366e-05,
+      "loss": 0.2839,
+      "step": 33575
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6945290891191226e-05,
+      "loss": 0.3162,
+      "step": 33580
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6944822089916086e-05,
+      "loss": 0.3926,
+      "step": 33585
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6944353288640945e-05,
+      "loss": 0.0653,
+      "step": 33590
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6943884487365805e-05,
+      "loss": 0.0839,
+      "step": 33595
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.694341568609067e-05,
+      "loss": 0.0872,
+      "step": 33600
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.694294688481553e-05,
+      "loss": 0.0573,
+      "step": 33605
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.694247808354039e-05,
+      "loss": 0.1543,
+      "step": 33610
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.694200928226525e-05,
+      "loss": 0.1087,
+      "step": 33615
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6941540480990112e-05,
+      "loss": 0.2351,
+      "step": 33620
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6941071679714972e-05,
+      "loss": 0.2309,
+      "step": 33625
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6940602878439832e-05,
+      "loss": 0.431,
+      "step": 33630
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.694013407716469e-05,
+      "loss": 0.2245,
+      "step": 33635
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.693966527588955e-05,
+      "loss": 0.0425,
+      "step": 33640
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.693919647461441e-05,
+      "loss": 0.0515,
+      "step": 33645
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.693872767333927e-05,
+      "loss": 0.0823,
+      "step": 33650
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.693825887206413e-05,
+      "loss": 0.1146,
+      "step": 33655
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6937790070788995e-05,
+      "loss": 0.0919,
+      "step": 33660
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6937321269513855e-05,
+      "loss": 0.1207,
+      "step": 33665
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6936852468238715e-05,
+      "loss": 0.1433,
+      "step": 33670
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6936383666963575e-05,
+      "loss": 0.1465,
+      "step": 33675
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6935914865688438e-05,
+      "loss": 0.3602,
+      "step": 33680
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6935446064413298e-05,
+      "loss": 0.1597,
+      "step": 33685
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6934977263138158e-05,
+      "loss": 0.0452,
+      "step": 33690
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6934508461863018e-05,
+      "loss": 0.0767,
+      "step": 33695
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6934039660587878e-05,
+      "loss": 0.1225,
+      "step": 33700
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6933570859312738e-05,
+      "loss": 0.0508,
+      "step": 33705
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6933102058037597e-05,
+      "loss": 0.0468,
+      "step": 33710
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.693263325676246e-05,
+      "loss": 0.1234,
+      "step": 33715
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.693216445548732e-05,
+      "loss": 0.1179,
+      "step": 33720
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.693169565421218e-05,
+      "loss": 0.2601,
+      "step": 33725
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.693122685293704e-05,
+      "loss": 0.3565,
+      "step": 33730
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.69307580516619e-05,
+      "loss": 0.2142,
+      "step": 33735
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6930289250386764e-05,
+      "loss": 0.0426,
+      "step": 33740
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6929820449111624e-05,
+      "loss": 0.0552,
+      "step": 33745
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6929351647836484e-05,
+      "loss": 0.1314,
+      "step": 33750
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6928882846561344e-05,
+      "loss": 0.1151,
+      "step": 33755
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6928414045286207e-05,
+      "loss": 0.0642,
+      "step": 33760
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6927945244011067e-05,
+      "loss": 0.1305,
+      "step": 33765
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6927476442735927e-05,
+      "loss": 0.1309,
+      "step": 33770
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6927007641460787e-05,
+      "loss": 0.1447,
+      "step": 33775
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6926538840185647e-05,
+      "loss": 0.3888,
+      "step": 33780
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6926070038910507e-05,
+      "loss": 0.2636,
+      "step": 33785
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6925601237635367e-05,
+      "loss": 0.1061,
+      "step": 33790
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6925132436360226e-05,
+      "loss": 0.0731,
+      "step": 33795
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6924663635085086e-05,
+      "loss": 0.0947,
+      "step": 33800
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.692419483380995e-05,
+      "loss": 0.1717,
+      "step": 33805
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.692372603253481e-05,
+      "loss": 0.1695,
+      "step": 33810
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.692325723125967e-05,
+      "loss": 0.1291,
+      "step": 33815
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6922788429984533e-05,
+      "loss": 0.1663,
+      "step": 33820
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6922319628709393e-05,
+      "loss": 0.1583,
+      "step": 33825
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6921850827434253e-05,
+      "loss": 0.2274,
+      "step": 33830
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6921382026159113e-05,
+      "loss": 0.3158,
+      "step": 33835
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6920913224883973e-05,
+      "loss": 0.0246,
+      "step": 33840
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6920444423608833e-05,
+      "loss": 0.0657,
+      "step": 33845
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6919975622333693e-05,
+      "loss": 0.0545,
+      "step": 33850
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6919506821058556e-05,
+      "loss": 0.0805,
+      "step": 33855
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6919038019783416e-05,
+      "loss": 0.1578,
+      "step": 33860
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6918569218508276e-05,
+      "loss": 0.1488,
+      "step": 33865
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6918100417233136e-05,
+      "loss": 0.1877,
+      "step": 33870
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6917631615957996e-05,
+      "loss": 0.2024,
+      "step": 33875
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6917162814682856e-05,
+      "loss": 0.3046,
+      "step": 33880
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.691669401340772e-05,
+      "loss": 0.2374,
+      "step": 33885
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.691622521213258e-05,
+      "loss": 0.0777,
+      "step": 33890
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.691575641085744e-05,
+      "loss": 0.0611,
+      "step": 33895
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6915287609582302e-05,
+      "loss": 0.0871,
+      "step": 33900
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6914818808307162e-05,
+      "loss": 0.0591,
+      "step": 33905
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6914350007032022e-05,
+      "loss": 0.1289,
+      "step": 33910
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6913881205756882e-05,
+      "loss": 0.2123,
+      "step": 33915
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6913412404481742e-05,
+      "loss": 0.1573,
+      "step": 33920
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.69129436032066e-05,
+      "loss": 0.2144,
+      "step": 33925
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.691247480193146e-05,
+      "loss": 0.3514,
+      "step": 33930
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.691200600065632e-05,
+      "loss": 0.2415,
+      "step": 33935
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.691153719938118e-05,
+      "loss": 0.0814,
+      "step": 33940
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6911068398106045e-05,
+      "loss": 0.1162,
+      "step": 33945
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6910599596830905e-05,
+      "loss": 0.0979,
+      "step": 33950
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6910130795555765e-05,
+      "loss": 0.1339,
+      "step": 33955
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6909661994280625e-05,
+      "loss": 0.1206,
+      "step": 33960
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6909193193005488e-05,
+      "loss": 0.1281,
+      "step": 33965
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.6908724391730348e-05,
+      "loss": 0.2954,
+      "step": 33970
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.6908255590455208e-05,
+      "loss": 0.3219,
+      "step": 33975
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.6907786789180068e-05,
+      "loss": 0.4502,
+      "step": 33980
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.6907317987904928e-05,
+      "loss": 0.2787,
+      "step": 33985
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.6906849186629788e-05,
+      "loss": 0.0437,
+      "step": 33990
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.690638038535465e-05,
+      "loss": 0.0631,
+      "step": 33995
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.690591158407951e-05,
+      "loss": 0.1087,
+      "step": 34000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.690544278280437e-05,
+      "loss": 0.1206,
+      "step": 34005
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.690497398152923e-05,
+      "loss": 0.1209,
+      "step": 34010
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.690450518025409e-05,
+      "loss": 0.0778,
+      "step": 34015
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.690403637897895e-05,
+      "loss": 0.2117,
+      "step": 34020
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.690356757770381e-05,
+      "loss": 0.2959,
+      "step": 34025
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.6903098776428674e-05,
+      "loss": 0.2687,
+      "step": 34030
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.6902629975153534e-05,
+      "loss": 0.2491,
+      "step": 34035
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.6902161173878394e-05,
+      "loss": 0.0504,
+      "step": 34040
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.6901692372603257e-05,
+      "loss": 0.039,
+      "step": 34045
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.6901223571328117e-05,
+      "loss": 0.0585,
+      "step": 34050
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.6900754770052977e-05,
+      "loss": 0.095,
+      "step": 34055
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.6900285968777837e-05,
+      "loss": 0.0796,
+      "step": 34060
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.6899817167502697e-05,
+      "loss": 0.1026,
+      "step": 34065
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.6899348366227557e-05,
+      "loss": 0.2815,
+      "step": 34070
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.6898879564952417e-05,
+      "loss": 0.1817,
+      "step": 34075
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.6898410763677277e-05,
+      "loss": 0.3487,
+      "step": 34080
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.689794196240214e-05,
+      "loss": 0.4519,
+      "step": 34085
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.6897473161127e-05,
+      "loss": 0.1144,
+      "step": 34090
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.689700435985186e-05,
+      "loss": 0.0615,
+      "step": 34095
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.689653555857672e-05,
+      "loss": 0.1148,
+      "step": 34100
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.689606675730158e-05,
+      "loss": 0.14,
+      "step": 34105
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.6895597956026443e-05,
+      "loss": 0.1061,
+      "step": 34110
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.6895129154751303e-05,
+      "loss": 0.1066,
+      "step": 34115
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.6894660353476163e-05,
+      "loss": 0.1549,
+      "step": 34120
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.6894191552201023e-05,
+      "loss": 0.1757,
+      "step": 34125
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.6893722750925886e-05,
+      "loss": 0.226,
+      "step": 34130
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.6893253949650746e-05,
+      "loss": 0.1872,
+      "step": 34135
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.6892785148375606e-05,
+      "loss": 0.0721,
+      "step": 34140
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.6892316347100466e-05,
+      "loss": 0.0644,
+      "step": 34145
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.6891847545825326e-05,
+      "loss": 0.082,
+      "step": 34150
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.6891378744550186e-05,
+      "loss": 0.1829,
+      "step": 34155
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.6890909943275046e-05,
+      "loss": 0.1974,
+      "step": 34160
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.6890441141999906e-05,
+      "loss": 0.1326,
+      "step": 34165
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.688997234072477e-05,
+      "loss": 0.197,
+      "step": 34170
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.688950353944963e-05,
+      "loss": 0.272,
+      "step": 34175
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.688903473817449e-05,
+      "loss": 0.3754,
+      "step": 34180
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.688856593689935e-05,
+      "loss": 0.21,
+      "step": 34185
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.6888097135624212e-05,
+      "loss": 0.072,
+      "step": 34190
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.6887628334349072e-05,
+      "loss": 0.0515,
+      "step": 34195
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.6887159533073932e-05,
+      "loss": 0.0671,
+      "step": 34200
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.6886690731798792e-05,
+      "loss": 0.3223,
+      "step": 34205
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.6886221930523652e-05,
+      "loss": 0.065,
+      "step": 34210
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.6885753129248512e-05,
+      "loss": 0.1724,
+      "step": 34215
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.688528432797337e-05,
+      "loss": 0.2077,
+      "step": 34220
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.6884815526698235e-05,
+      "loss": 0.2935,
+      "step": 34225
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.6884346725423095e-05,
+      "loss": 0.4017,
+      "step": 34230
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.6883877924147955e-05,
+      "loss": 0.1805,
+      "step": 34235
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.6883409122872815e-05,
+      "loss": 0.0684,
+      "step": 34240
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.6882940321597675e-05,
+      "loss": 0.1078,
+      "step": 34245
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.6882471520322538e-05,
+      "loss": 0.0505,
+      "step": 34250
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.6882002719047398e-05,
+      "loss": 0.0993,
+      "step": 34255
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.6881533917772258e-05,
+      "loss": 0.1775,
+      "step": 34260
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.6881065116497118e-05,
+      "loss": 0.0837,
+      "step": 34265
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.688059631522198e-05,
+      "loss": 0.1667,
+      "step": 34270
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.688012751394684e-05,
+      "loss": 0.233,
+      "step": 34275
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.68796587126717e-05,
+      "loss": 0.2761,
+      "step": 34280
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.687918991139656e-05,
+      "loss": 0.371,
+      "step": 34285
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.687872111012142e-05,
+      "loss": 0.0497,
+      "step": 34290
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.687825230884628e-05,
+      "loss": 0.0667,
+      "step": 34295
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.687778350757114e-05,
+      "loss": 0.096,
+      "step": 34300
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.6877314706296e-05,
+      "loss": 0.0963,
+      "step": 34305
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.687684590502086e-05,
+      "loss": 0.2588,
+      "step": 34310
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.6876377103745724e-05,
+      "loss": 0.0869,
+      "step": 34315
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.6875908302470584e-05,
+      "loss": 0.1955,
+      "step": 34320
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.6875439501195444e-05,
+      "loss": 0.2895,
+      "step": 34325
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.6874970699920307e-05,
+      "loss": 0.3213,
+      "step": 34330
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.6874501898645167e-05,
+      "loss": 0.3409,
+      "step": 34335
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.6874033097370027e-05,
+      "loss": 0.0989,
+      "step": 34340
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.6873564296094887e-05,
+      "loss": 0.0809,
+      "step": 34345
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.6873095494819747e-05,
+      "loss": 0.0576,
+      "step": 34350
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.6872626693544607e-05,
+      "loss": 0.1124,
+      "step": 34355
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.6872157892269467e-05,
+      "loss": 0.1191,
+      "step": 34360
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.687168909099433e-05,
+      "loss": 0.1092,
+      "step": 34365
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.687122028971919e-05,
+      "loss": 0.1064,
+      "step": 34370
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.687075148844405e-05,
+      "loss": 0.2331,
+      "step": 34375
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.687028268716891e-05,
+      "loss": 0.3882,
+      "step": 34380
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.686981388589377e-05,
+      "loss": 0.2774,
+      "step": 34385
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.686934508461863e-05,
+      "loss": 0.0547,
+      "step": 34390
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.6868876283343493e-05,
+      "loss": 0.0845,
+      "step": 34395
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.6868407482068353e-05,
+      "loss": 0.1467,
+      "step": 34400
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.6867938680793213e-05,
+      "loss": 0.1287,
+      "step": 34405
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.6867469879518076e-05,
+      "loss": 0.1478,
+      "step": 34410
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.6867001078242936e-05,
+      "loss": 0.2076,
+      "step": 34415
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.6866532276967796e-05,
+      "loss": 0.1496,
+      "step": 34420
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.6866063475692656e-05,
+      "loss": 0.2264,
+      "step": 34425
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.6865594674417516e-05,
+      "loss": 0.3986,
+      "step": 34430
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.6865125873142376e-05,
+      "loss": 0.3061,
+      "step": 34435
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.6864657071867236e-05,
+      "loss": 0.0217,
+      "step": 34440
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.6864188270592096e-05,
+      "loss": 0.0725,
+      "step": 34445
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.6863719469316956e-05,
+      "loss": 0.0708,
+      "step": 34450
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.686325066804182e-05,
+      "loss": 0.1335,
+      "step": 34455
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.686278186676668e-05,
+      "loss": 0.1592,
+      "step": 34460
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.686231306549154e-05,
+      "loss": 0.2168,
+      "step": 34465
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.68618442642164e-05,
+      "loss": 0.1201,
+      "step": 34470
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.6861375462941262e-05,
+      "loss": 0.241,
+      "step": 34475
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.6860906661666122e-05,
+      "loss": 0.4339,
+      "step": 34480
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.6860437860390982e-05,
+      "loss": 0.332,
+      "step": 34485
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.6859969059115842e-05,
+      "loss": 0.0336,
+      "step": 34490
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.6859500257840702e-05,
+      "loss": 0.0586,
+      "step": 34495
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.6859031456565562e-05,
+      "loss": 0.0608,
+      "step": 34500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.6858562655290425e-05,
+      "loss": 0.1441,
+      "step": 34505
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.6858093854015285e-05,
+      "loss": 0.0687,
+      "step": 34510
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.6857625052740145e-05,
+      "loss": 0.1453,
+      "step": 34515
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.6857156251465005e-05,
+      "loss": 0.1763,
+      "step": 34520
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.6856687450189865e-05,
+      "loss": 0.2107,
+      "step": 34525
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.6856218648914725e-05,
+      "loss": 0.3186,
+      "step": 34530
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.6855749847639585e-05,
+      "loss": 0.3441,
+      "step": 34535
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.6855281046364448e-05,
+      "loss": 0.0226,
+      "step": 34540
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.6854812245089308e-05,
+      "loss": 0.075,
+      "step": 34545
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.6854343443814168e-05,
+      "loss": 0.1509,
+      "step": 34550
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.685387464253903e-05,
+      "loss": 0.0642,
+      "step": 34555
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.685340584126389e-05,
+      "loss": 0.1539,
+      "step": 34560
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.685293703998875e-05,
+      "loss": 0.1392,
+      "step": 34565
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.685246823871361e-05,
+      "loss": 0.0896,
+      "step": 34570
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.685199943743847e-05,
+      "loss": 0.3167,
+      "step": 34575
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.685153063616333e-05,
+      "loss": 0.3073,
+      "step": 34580
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.685106183488819e-05,
+      "loss": 0.2712,
+      "step": 34585
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.685059303361305e-05,
+      "loss": 0.0389,
+      "step": 34590
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.6850124232337914e-05,
+      "loss": 0.0515,
+      "step": 34595
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.6849655431062774e-05,
+      "loss": 0.1278,
+      "step": 34600
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.6849186629787634e-05,
+      "loss": 0.0901,
+      "step": 34605
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.6848717828512494e-05,
+      "loss": 0.1417,
+      "step": 34610
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.6848249027237357e-05,
+      "loss": 0.1445,
+      "step": 34615
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.6847780225962217e-05,
+      "loss": 0.1738,
+      "step": 34620
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.6847311424687077e-05,
+      "loss": 0.23,
+      "step": 34625
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.6846842623411937e-05,
+      "loss": 0.3401,
+      "step": 34630
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.6846373822136797e-05,
+      "loss": 0.2777,
+      "step": 34635
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.6845905020861657e-05,
+      "loss": 0.0903,
+      "step": 34640
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.684543621958652e-05,
+      "loss": 0.0529,
+      "step": 34645
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.684496741831138e-05,
+      "loss": 0.1014,
+      "step": 34650
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.684449861703624e-05,
+      "loss": 0.1217,
+      "step": 34655
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.68440298157611e-05,
+      "loss": 0.1712,
+      "step": 34660
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.684356101448596e-05,
+      "loss": 0.1573,
+      "step": 34665
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.684309221321082e-05,
+      "loss": 0.2182,
+      "step": 34670
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.684262341193568e-05,
+      "loss": 0.1673,
+      "step": 34675
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.6842154610660543e-05,
+      "loss": 0.1951,
+      "step": 34680
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.6841685809385403e-05,
+      "loss": 0.2755,
+      "step": 34685
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.6841217008110263e-05,
+      "loss": 0.0314,
+      "step": 34690
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.6840748206835126e-05,
+      "loss": 0.0377,
+      "step": 34695
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.6840279405559986e-05,
+      "loss": 0.0633,
+      "step": 34700
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.6839810604284846e-05,
+      "loss": 0.0738,
+      "step": 34705
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.6839341803009706e-05,
+      "loss": 0.147,
+      "step": 34710
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.6838873001734566e-05,
+      "loss": 0.0918,
+      "step": 34715
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.6838404200459426e-05,
+      "loss": 0.113,
+      "step": 34720
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.6837935399184286e-05,
+      "loss": 0.1378,
+      "step": 34725
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.6837466597909146e-05,
+      "loss": 0.2208,
+      "step": 34730
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.683699779663401e-05,
+      "loss": 0.3092,
+      "step": 34735
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.683652899535887e-05,
+      "loss": 0.0954,
+      "step": 34740
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.683606019408373e-05,
+      "loss": 0.0398,
+      "step": 34745
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.683559139280859e-05,
+      "loss": 0.0969,
+      "step": 34750
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.683512259153345e-05,
+      "loss": 0.075,
+      "step": 34755
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.6834653790258312e-05,
+      "loss": 0.1861,
+      "step": 34760
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.6834184988983172e-05,
+      "loss": 0.1478,
+      "step": 34765
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.6833716187708032e-05,
+      "loss": 0.1653,
+      "step": 34770
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.6833247386432892e-05,
+      "loss": 0.2705,
+      "step": 34775
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.6832778585157755e-05,
+      "loss": 0.2997,
+      "step": 34780
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.6832309783882615e-05,
+      "loss": 0.2416,
+      "step": 34785
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.6831840982607475e-05,
+      "loss": 0.0703,
+      "step": 34790
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.6831372181332335e-05,
+      "loss": 0.0518,
+      "step": 34795
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.6830903380057195e-05,
+      "loss": 0.1567,
+      "step": 34800
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.6830434578782055e-05,
+      "loss": 0.0769,
+      "step": 34805
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.6829965777506915e-05,
+      "loss": 0.0968,
+      "step": 34810
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.6829496976231775e-05,
+      "loss": 0.0712,
+      "step": 34815
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.6829028174956635e-05,
+      "loss": 0.2025,
+      "step": 34820
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.6828559373681498e-05,
+      "loss": 0.2455,
+      "step": 34825
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6828090572406358e-05,
+      "loss": 0.1866,
+      "step": 34830
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6827621771131218e-05,
+      "loss": 0.3097,
+      "step": 34835
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.682715296985608e-05,
+      "loss": 0.0356,
+      "step": 34840
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.682668416858094e-05,
+      "loss": 0.0695,
+      "step": 34845
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.68262153673058e-05,
+      "loss": 0.0611,
+      "step": 34850
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.682574656603066e-05,
+      "loss": 0.0836,
+      "step": 34855
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.682527776475552e-05,
+      "loss": 0.1121,
+      "step": 34860
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.682480896348038e-05,
+      "loss": 0.1199,
+      "step": 34865
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.682434016220524e-05,
+      "loss": 0.1647,
+      "step": 34870
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6823871360930104e-05,
+      "loss": 0.1451,
+      "step": 34875
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6823402559654964e-05,
+      "loss": 0.2457,
+      "step": 34880
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6822933758379824e-05,
+      "loss": 0.2741,
+      "step": 34885
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6822464957104684e-05,
+      "loss": 0.0439,
+      "step": 34890
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6821996155829544e-05,
+      "loss": 0.0378,
+      "step": 34895
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6821527354554404e-05,
+      "loss": 0.134,
+      "step": 34900
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6821058553279267e-05,
+      "loss": 0.0693,
+      "step": 34905
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6820589752004127e-05,
+      "loss": 0.1013,
+      "step": 34910
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6820120950728987e-05,
+      "loss": 0.1133,
+      "step": 34915
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.681965214945385e-05,
+      "loss": 0.1309,
+      "step": 34920
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.681918334817871e-05,
+      "loss": 0.1584,
+      "step": 34925
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.681871454690357e-05,
+      "loss": 0.2859,
+      "step": 34930
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.681824574562843e-05,
+      "loss": 0.2421,
+      "step": 34935
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.681777694435329e-05,
+      "loss": 0.0499,
+      "step": 34940
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.681730814307815e-05,
+      "loss": 0.0865,
+      "step": 34945
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.681683934180301e-05,
+      "loss": 0.0344,
+      "step": 34950
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.681637054052787e-05,
+      "loss": 0.1128,
+      "step": 34955
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.681590173925273e-05,
+      "loss": 0.0706,
+      "step": 34960
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6815432937977593e-05,
+      "loss": 0.1731,
+      "step": 34965
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6814964136702453e-05,
+      "loss": 0.1794,
+      "step": 34970
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6814495335427313e-05,
+      "loss": 0.2741,
+      "step": 34975
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6814026534152173e-05,
+      "loss": 0.4299,
+      "step": 34980
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6813557732877036e-05,
+      "loss": 0.2528,
+      "step": 34985
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6813088931601896e-05,
+      "loss": 0.0214,
+      "step": 34990
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6812620130326756e-05,
+      "loss": 0.0732,
+      "step": 34995
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6812151329051616e-05,
+      "loss": 0.0879,
+      "step": 35000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6811682527776476e-05,
+      "loss": 0.0779,
+      "step": 35005
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6811213726501336e-05,
+      "loss": 0.1751,
+      "step": 35010
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.68107449252262e-05,
+      "loss": 0.1612,
+      "step": 35015
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.681027612395106e-05,
+      "loss": 0.1184,
+      "step": 35020
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.680980732267592e-05,
+      "loss": 0.2088,
+      "step": 35025
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.680933852140078e-05,
+      "loss": 0.2776,
+      "step": 35030
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.680886972012564e-05,
+      "loss": 0.3245,
+      "step": 35035
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.68084009188505e-05,
+      "loss": 0.0458,
+      "step": 35040
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6807932117575362e-05,
+      "loss": 0.0426,
+      "step": 35045
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6807463316300222e-05,
+      "loss": 0.0691,
+      "step": 35050
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6806994515025082e-05,
+      "loss": 0.1131,
+      "step": 35055
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6806525713749942e-05,
+      "loss": 0.1071,
+      "step": 35060
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6806056912474805e-05,
+      "loss": 0.1247,
+      "step": 35065
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6805588111199665e-05,
+      "loss": 0.207,
+      "step": 35070
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6805119309924525e-05,
+      "loss": 0.3076,
+      "step": 35075
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6804650508649385e-05,
+      "loss": 0.353,
+      "step": 35080
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6804181707374245e-05,
+      "loss": 0.2484,
+      "step": 35085
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6803712906099105e-05,
+      "loss": 0.0702,
+      "step": 35090
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6803244104823965e-05,
+      "loss": 0.0584,
+      "step": 35095
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6802775303548825e-05,
+      "loss": 0.0575,
+      "step": 35100
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.680230650227369e-05,
+      "loss": 0.0894,
+      "step": 35105
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6801837700998548e-05,
+      "loss": 0.1053,
+      "step": 35110
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6801368899723408e-05,
+      "loss": 0.0946,
+      "step": 35115
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6800900098448268e-05,
+      "loss": 0.1376,
+      "step": 35120
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.680043129717313e-05,
+      "loss": 0.2744,
+      "step": 35125
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.679996249589799e-05,
+      "loss": 0.4252,
+      "step": 35130
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.679949369462285e-05,
+      "loss": 0.235,
+      "step": 35135
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.679902489334771e-05,
+      "loss": 0.0659,
+      "step": 35140
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.679855609207257e-05,
+      "loss": 0.0471,
+      "step": 35145
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.679808729079743e-05,
+      "loss": 0.0852,
+      "step": 35150
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6797618489522294e-05,
+      "loss": 0.1383,
+      "step": 35155
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6797149688247154e-05,
+      "loss": 0.0855,
+      "step": 35160
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6796680886972014e-05,
+      "loss": 0.1548,
+      "step": 35165
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6796212085696874e-05,
+      "loss": 0.1816,
+      "step": 35170
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6795743284421734e-05,
+      "loss": 0.118,
+      "step": 35175
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6795274483146594e-05,
+      "loss": 0.414,
+      "step": 35180
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6794805681871454e-05,
+      "loss": 0.2278,
+      "step": 35185
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6794336880596317e-05,
+      "loss": 0.074,
+      "step": 35190
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6793868079321177e-05,
+      "loss": 0.051,
+      "step": 35195
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6793399278046037e-05,
+      "loss": 0.0984,
+      "step": 35200
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.67929304767709e-05,
+      "loss": 0.0757,
+      "step": 35205
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.679246167549576e-05,
+      "loss": 0.1527,
+      "step": 35210
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.679199287422062e-05,
+      "loss": 0.1969,
+      "step": 35215
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.679152407294548e-05,
+      "loss": 0.0733,
+      "step": 35220
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.679105527167034e-05,
+      "loss": 0.2065,
+      "step": 35225
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.67905864703952e-05,
+      "loss": 0.2298,
+      "step": 35230
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.679011766912006e-05,
+      "loss": 0.2383,
+      "step": 35235
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.678964886784492e-05,
+      "loss": 0.0261,
+      "step": 35240
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6789180066569783e-05,
+      "loss": 0.0558,
+      "step": 35245
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6788711265294643e-05,
+      "loss": 0.1007,
+      "step": 35250
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6788242464019503e-05,
+      "loss": 0.1138,
+      "step": 35255
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6787773662744363e-05,
+      "loss": 0.1242,
+      "step": 35260
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6787304861469223e-05,
+      "loss": 0.2085,
+      "step": 35265
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6786836060194086e-05,
+      "loss": 0.171,
+      "step": 35270
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6786367258918946e-05,
+      "loss": 0.1888,
+      "step": 35275
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6785898457643806e-05,
+      "loss": 0.2637,
+      "step": 35280
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6785429656368666e-05,
+      "loss": 0.3491,
+      "step": 35285
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6784960855093526e-05,
+      "loss": 0.0144,
+      "step": 35290
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.678449205381839e-05,
+      "loss": 0.0612,
+      "step": 35295
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.678402325254325e-05,
+      "loss": 0.0288,
+      "step": 35300
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.678355445126811e-05,
+      "loss": 0.1251,
+      "step": 35305
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.678308564999297e-05,
+      "loss": 0.1111,
+      "step": 35310
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.678261684871783e-05,
+      "loss": 0.1563,
+      "step": 35315
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.678214804744269e-05,
+      "loss": 0.1506,
+      "step": 35320
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.678167924616755e-05,
+      "loss": 0.2064,
+      "step": 35325
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.678121044489241e-05,
+      "loss": 0.4103,
+      "step": 35330
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6780741643617272e-05,
+      "loss": 0.253,
+      "step": 35335
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6780272842342132e-05,
+      "loss": 0.0347,
+      "step": 35340
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6779804041066992e-05,
+      "loss": 0.0921,
+      "step": 35345
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6779335239791856e-05,
+      "loss": 0.0761,
+      "step": 35350
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6778866438516716e-05,
+      "loss": 0.053,
+      "step": 35355
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6778397637241575e-05,
+      "loss": 0.0907,
+      "step": 35360
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6777928835966435e-05,
+      "loss": 0.1111,
+      "step": 35365
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6777460034691295e-05,
+      "loss": 0.171,
+      "step": 35370
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6776991233416155e-05,
+      "loss": 0.1687,
+      "step": 35375
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6776522432141015e-05,
+      "loss": 0.3957,
+      "step": 35380
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.677605363086588e-05,
+      "loss": 0.2045,
+      "step": 35385
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.677558482959074e-05,
+      "loss": 0.0608,
+      "step": 35390
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.67751160283156e-05,
+      "loss": 0.0678,
+      "step": 35395
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6774647227040458e-05,
+      "loss": 0.1181,
+      "step": 35400
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6774178425765318e-05,
+      "loss": 0.0646,
+      "step": 35405
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6773709624490178e-05,
+      "loss": 0.1283,
+      "step": 35410
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.677324082321504e-05,
+      "loss": 0.1436,
+      "step": 35415
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.67727720219399e-05,
+      "loss": 0.1407,
+      "step": 35420
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.677230322066476e-05,
+      "loss": 0.1615,
+      "step": 35425
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6771834419389625e-05,
+      "loss": 0.3031,
+      "step": 35430
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6771365618114485e-05,
+      "loss": 0.2634,
+      "step": 35435
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6770896816839345e-05,
+      "loss": 0.0649,
+      "step": 35440
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6770428015564204e-05,
+      "loss": 0.0962,
+      "step": 35445
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6769959214289064e-05,
+      "loss": 0.0512,
+      "step": 35450
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6769490413013924e-05,
+      "loss": 0.1274,
+      "step": 35455
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6769021611738784e-05,
+      "loss": 0.0368,
+      "step": 35460
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6768552810463644e-05,
+      "loss": 0.1301,
+      "step": 35465
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.6768084009188504e-05,
+      "loss": 0.1465,
+      "step": 35470
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.6767615207913367e-05,
+      "loss": 0.2885,
+      "step": 35475
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.6767146406638227e-05,
+      "loss": 0.4663,
+      "step": 35480
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.6766677605363087e-05,
+      "loss": 0.4114,
+      "step": 35485
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.6766208804087947e-05,
+      "loss": 0.0577,
+      "step": 35490
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.676574000281281e-05,
+      "loss": 0.0584,
+      "step": 35495
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.676527120153767e-05,
+      "loss": 0.0834,
+      "step": 35500
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.676480240026253e-05,
+      "loss": 0.0556,
+      "step": 35505
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.676433359898739e-05,
+      "loss": 0.2583,
+      "step": 35510
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.676386479771225e-05,
+      "loss": 0.1383,
+      "step": 35515
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.676339599643711e-05,
+      "loss": 0.2096,
+      "step": 35520
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.6762927195161974e-05,
+      "loss": 0.2905,
+      "step": 35525
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.6762458393886834e-05,
+      "loss": 0.4707,
+      "step": 35530
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.6761989592611693e-05,
+      "loss": 0.3049,
+      "step": 35535
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.6761520791336553e-05,
+      "loss": 0.0263,
+      "step": 35540
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.6761051990061413e-05,
+      "loss": 0.1083,
+      "step": 35545
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.6760583188786273e-05,
+      "loss": 0.1045,
+      "step": 35550
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.6760114387511137e-05,
+      "loss": 0.0981,
+      "step": 35555
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.6759645586235997e-05,
+      "loss": 0.1292,
+      "step": 35560
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.6759176784960856e-05,
+      "loss": 0.1232,
+      "step": 35565
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.6758707983685716e-05,
+      "loss": 0.1619,
+      "step": 35570
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.675823918241058e-05,
+      "loss": 0.1738,
+      "step": 35575
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.675777038113544e-05,
+      "loss": 0.3627,
+      "step": 35580
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.67573015798603e-05,
+      "loss": 0.2957,
+      "step": 35585
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.675683277858516e-05,
+      "loss": 0.0471,
+      "step": 35590
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.675636397731002e-05,
+      "loss": 0.0628,
+      "step": 35595
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.675589517603488e-05,
+      "loss": 0.0737,
+      "step": 35600
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.675542637475974e-05,
+      "loss": 0.115,
+      "step": 35605
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.67549575734846e-05,
+      "loss": 0.12,
+      "step": 35610
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.675448877220946e-05,
+      "loss": 0.1704,
+      "step": 35615
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.6754019970934322e-05,
+      "loss": 0.1348,
+      "step": 35620
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.6753551169659182e-05,
+      "loss": 0.1868,
+      "step": 35625
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.6753082368384042e-05,
+      "loss": 0.4204,
+      "step": 35630
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.6752613567108906e-05,
+      "loss": 0.3445,
+      "step": 35635
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.6752144765833766e-05,
+      "loss": 0.0448,
+      "step": 35640
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.6751675964558626e-05,
+      "loss": 0.0615,
+      "step": 35645
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.6751207163283485e-05,
+      "loss": 0.0439,
+      "step": 35650
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.6750738362008345e-05,
+      "loss": 0.1228,
+      "step": 35655
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.6750269560733205e-05,
+      "loss": 0.1486,
+      "step": 35660
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.674980075945807e-05,
+      "loss": 0.2386,
+      "step": 35665
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.674933195818293e-05,
+      "loss": 0.1217,
+      "step": 35670
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.674886315690779e-05,
+      "loss": 0.2752,
+      "step": 35675
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.674839435563265e-05,
+      "loss": 0.4395,
+      "step": 35680
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.674792555435751e-05,
+      "loss": 0.3155,
+      "step": 35685
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.674745675308237e-05,
+      "loss": 0.0589,
+      "step": 35690
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6746987951807228e-05,
+      "loss": 0.0838,
+      "step": 35695
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.674651915053209e-05,
+      "loss": 0.0867,
+      "step": 35700
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.674605034925695e-05,
+      "loss": 0.1114,
+      "step": 35705
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.674558154798181e-05,
+      "loss": 0.1468,
+      "step": 35710
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6745112746706675e-05,
+      "loss": 0.1577,
+      "step": 35715
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6744643945431535e-05,
+      "loss": 0.1506,
+      "step": 35720
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6744175144156395e-05,
+      "loss": 0.2544,
+      "step": 35725
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6743706342881255e-05,
+      "loss": 0.2944,
+      "step": 35730
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6743237541606115e-05,
+      "loss": 0.2873,
+      "step": 35735
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6742768740330974e-05,
+      "loss": 0.0319,
+      "step": 35740
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6742299939055834e-05,
+      "loss": 0.0863,
+      "step": 35745
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6741831137780694e-05,
+      "loss": 0.1371,
+      "step": 35750
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6741362336505558e-05,
+      "loss": 0.0674,
+      "step": 35755
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6740893535230418e-05,
+      "loss": 0.1225,
+      "step": 35760
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6740424733955278e-05,
+      "loss": 0.2157,
+      "step": 35765
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6739955932680137e-05,
+      "loss": 0.1537,
+      "step": 35770
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6739487131404997e-05,
+      "loss": 0.1671,
+      "step": 35775
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.673901833012986e-05,
+      "loss": 0.2707,
+      "step": 35780
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.673854952885472e-05,
+      "loss": 0.3311,
+      "step": 35785
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.673808072757958e-05,
+      "loss": 0.0228,
+      "step": 35790
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.673761192630444e-05,
+      "loss": 0.1209,
+      "step": 35795
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.67371431250293e-05,
+      "loss": 0.0254,
+      "step": 35800
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6736674323754164e-05,
+      "loss": 0.1099,
+      "step": 35805
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6736205522479024e-05,
+      "loss": 0.2051,
+      "step": 35810
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6735736721203884e-05,
+      "loss": 0.2122,
+      "step": 35815
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6735267919928744e-05,
+      "loss": 0.2434,
+      "step": 35820
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6734799118653603e-05,
+      "loss": 0.2191,
+      "step": 35825
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6734330317378463e-05,
+      "loss": 0.3036,
+      "step": 35830
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6733861516103323e-05,
+      "loss": 0.2985,
+      "step": 35835
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6733392714828183e-05,
+      "loss": 0.0408,
+      "step": 35840
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6732923913553047e-05,
+      "loss": 0.0536,
+      "step": 35845
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6732455112277907e-05,
+      "loss": 0.0663,
+      "step": 35850
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6731986311002766e-05,
+      "loss": 0.1087,
+      "step": 35855
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.673151750972763e-05,
+      "loss": 0.1669,
+      "step": 35860
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.673104870845249e-05,
+      "loss": 0.1232,
+      "step": 35865
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.673057990717735e-05,
+      "loss": 0.154,
+      "step": 35870
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.673011110590221e-05,
+      "loss": 0.2032,
+      "step": 35875
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.672964230462707e-05,
+      "loss": 0.3006,
+      "step": 35880
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.672917350335193e-05,
+      "loss": 0.2368,
+      "step": 35885
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.672870470207679e-05,
+      "loss": 0.0608,
+      "step": 35890
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6728235900801653e-05,
+      "loss": 0.0982,
+      "step": 35895
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.6727767099526513e-05,
+      "loss": 0.082,
+      "step": 35900
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.6727298298251373e-05,
+      "loss": 0.1158,
+      "step": 35905
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.6726829496976233e-05,
+      "loss": 0.0961,
+      "step": 35910
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.6726360695701092e-05,
+      "loss": 0.157,
+      "step": 35915
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.6725891894425952e-05,
+      "loss": 0.181,
+      "step": 35920
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.6725423093150816e-05,
+      "loss": 0.2806,
+      "step": 35925
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.6724954291875676e-05,
+      "loss": 0.3032,
+      "step": 35930
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.6724485490600536e-05,
+      "loss": 0.3249,
+      "step": 35935
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.6724016689325396e-05,
+      "loss": 0.055,
+      "step": 35940
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.672354788805026e-05,
+      "loss": 0.0191,
+      "step": 35945
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.672307908677512e-05,
+      "loss": 0.0989,
+      "step": 35950
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.672261028549998e-05,
+      "loss": 0.0961,
+      "step": 35955
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.672214148422484e-05,
+      "loss": 0.1469,
+      "step": 35960
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.67216726829497e-05,
+      "loss": 0.1081,
+      "step": 35965
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.672120388167456e-05,
+      "loss": 0.1817,
+      "step": 35970
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.672073508039942e-05,
+      "loss": 0.2109,
+      "step": 35975
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.672026627912428e-05,
+      "loss": 0.4137,
+      "step": 35980
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.671979747784914e-05,
+      "loss": 0.2394,
+      "step": 35985
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.6719328676574e-05,
+      "loss": 0.0622,
+      "step": 35990
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.671885987529886e-05,
+      "loss": 0.0821,
+      "step": 35995
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.6718391074023725e-05,
+      "loss": 0.0751,
+      "step": 36000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.6717922272748585e-05,
+      "loss": 0.1294,
+      "step": 36005
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.6717453471473445e-05,
+      "loss": 0.1102,
+      "step": 36010
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.6716984670198305e-05,
+      "loss": 0.0996,
+      "step": 36015
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.6716515868923165e-05,
+      "loss": 0.2002,
+      "step": 36020
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.6716047067648025e-05,
+      "loss": 0.2179,
+      "step": 36025
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.6715578266372884e-05,
+      "loss": 0.3336,
+      "step": 36030
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.6715109465097748e-05,
+      "loss": 0.264,
+      "step": 36035
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.6714640663822608e-05,
+      "loss": 0.0501,
+      "step": 36040
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.6714171862547468e-05,
+      "loss": 0.0216,
+      "step": 36045
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.6713703061272328e-05,
+      "loss": 0.0658,
+      "step": 36050
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.6713234259997188e-05,
+      "loss": 0.1062,
+      "step": 36055
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.6712765458722047e-05,
+      "loss": 0.1623,
+      "step": 36060
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.671229665744691e-05,
+      "loss": 0.1972,
+      "step": 36065
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.671182785617177e-05,
+      "loss": 0.1783,
+      "step": 36070
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.671135905489663e-05,
+      "loss": 0.2079,
+      "step": 36075
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.6710890253621494e-05,
+      "loss": 0.4228,
+      "step": 36080
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.6710421452346354e-05,
+      "loss": 0.2925,
+      "step": 36085
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.6709952651071214e-05,
+      "loss": 0.0565,
+      "step": 36090
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.6709483849796074e-05,
+      "loss": 0.0666,
+      "step": 36095
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.6709015048520934e-05,
+      "loss": 0.5187,
+      "step": 36100
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.6708546247245794e-05,
+      "loss": 0.1156,
+      "step": 36105
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.6708077445970654e-05,
+      "loss": 0.1737,
+      "step": 36110
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.6707608644695514e-05,
+      "loss": 0.2334,
+      "step": 36115
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.6707139843420373e-05,
+      "loss": 0.1251,
+      "step": 36120
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.6706671042145233e-05,
+      "loss": 0.2372,
+      "step": 36125
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.6706202240870097e-05,
+      "loss": 0.3235,
+      "step": 36130
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.6705733439594957e-05,
+      "loss": 0.1987,
+      "step": 36135
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.6705264638319817e-05,
+      "loss": 0.0099,
+      "step": 36140
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.670479583704468e-05,
+      "loss": 0.0683,
+      "step": 36145
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.670432703576954e-05,
+      "loss": 0.0952,
+      "step": 36150
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.67038582344944e-05,
+      "loss": 0.0887,
+      "step": 36155
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.670338943321926e-05,
+      "loss": 0.0684,
+      "step": 36160
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.670292063194412e-05,
+      "loss": 0.1626,
+      "step": 36165
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.670245183066898e-05,
+      "loss": 0.2195,
+      "step": 36170
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.6701983029393843e-05,
+      "loss": 0.2328,
+      "step": 36175
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.6701514228118703e-05,
+      "loss": 0.3627,
+      "step": 36180
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.6701045426843563e-05,
+      "loss": 0.2381,
+      "step": 36185
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.6700576625568423e-05,
+      "loss": 0.0505,
+      "step": 36190
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.6700107824293283e-05,
+      "loss": 0.0658,
+      "step": 36195
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.6699639023018143e-05,
+      "loss": 0.0611,
+      "step": 36200
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.6699170221743002e-05,
+      "loss": 0.0871,
+      "step": 36205
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.6698701420467866e-05,
+      "loss": 0.1666,
+      "step": 36210
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.6698232619192726e-05,
+      "loss": 0.0819,
+      "step": 36215
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.6697763817917586e-05,
+      "loss": 0.1677,
+      "step": 36220
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.669729501664245e-05,
+      "loss": 0.2659,
+      "step": 36225
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.669682621536731e-05,
+      "loss": 0.2803,
+      "step": 36230
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.669635741409217e-05,
+      "loss": 0.2431,
+      "step": 36235
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.669588861281703e-05,
+      "loss": 0.04,
+      "step": 36240
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.669541981154189e-05,
+      "loss": 0.0686,
+      "step": 36245
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.669495101026675e-05,
+      "loss": 0.0622,
+      "step": 36250
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.669448220899161e-05,
+      "loss": 0.0969,
+      "step": 36255
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.669401340771647e-05,
+      "loss": 0.0752,
+      "step": 36260
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.669354460644133e-05,
+      "loss": 0.0922,
+      "step": 36265
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.6693075805166192e-05,
+      "loss": 0.1751,
+      "step": 36270
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.6692607003891052e-05,
+      "loss": 0.2008,
+      "step": 36275
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.669213820261591e-05,
+      "loss": 0.38,
+      "step": 36280
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.669166940134077e-05,
+      "loss": 0.3769,
+      "step": 36285
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.6691200600065635e-05,
+      "loss": 0.0415,
+      "step": 36290
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.6690731798790495e-05,
+      "loss": 0.057,
+      "step": 36295
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.6690262997515355e-05,
+      "loss": 0.073,
+      "step": 36300
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.6689794196240215e-05,
+      "loss": 0.0942,
+      "step": 36305
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.6689325394965075e-05,
+      "loss": 0.1002,
+      "step": 36310
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.6688856593689938e-05,
+      "loss": 0.1361,
+      "step": 36315
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.6688387792414798e-05,
+      "loss": 0.142,
+      "step": 36320
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.6687918991139658e-05,
+      "loss": 0.319,
+      "step": 36325
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6687450189864518e-05,
+      "loss": 0.3598,
+      "step": 36330
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6686981388589378e-05,
+      "loss": 0.2306,
+      "step": 36335
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6686512587314238e-05,
+      "loss": 0.0275,
+      "step": 36340
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6686043786039098e-05,
+      "loss": 0.0459,
+      "step": 36345
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.668557498476396e-05,
+      "loss": 0.0861,
+      "step": 36350
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.668510618348882e-05,
+      "loss": 0.113,
+      "step": 36355
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.668463738221368e-05,
+      "loss": 0.127,
+      "step": 36360
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.668416858093854e-05,
+      "loss": 0.2109,
+      "step": 36365
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6683699779663404e-05,
+      "loss": 0.1536,
+      "step": 36370
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6683230978388264e-05,
+      "loss": 0.2749,
+      "step": 36375
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6682762177113124e-05,
+      "loss": 0.4515,
+      "step": 36380
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6682293375837984e-05,
+      "loss": 0.2774,
+      "step": 36385
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6681824574562844e-05,
+      "loss": 0.0556,
+      "step": 36390
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6681355773287704e-05,
+      "loss": 0.0561,
+      "step": 36395
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6680886972012564e-05,
+      "loss": 0.0449,
+      "step": 36400
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6680418170737427e-05,
+      "loss": 0.0686,
+      "step": 36405
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6679949369462287e-05,
+      "loss": 0.1487,
+      "step": 36410
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6679480568187147e-05,
+      "loss": 0.1531,
+      "step": 36415
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6679011766912007e-05,
+      "loss": 0.1708,
+      "step": 36420
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6678542965636867e-05,
+      "loss": 0.3106,
+      "step": 36425
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.667807416436173e-05,
+      "loss": 0.2997,
+      "step": 36430
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.667760536308659e-05,
+      "loss": 0.3525,
+      "step": 36435
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.667713656181145e-05,
+      "loss": 0.0381,
+      "step": 36440
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.667666776053631e-05,
+      "loss": 0.058,
+      "step": 36445
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.667619895926117e-05,
+      "loss": 0.0493,
+      "step": 36450
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6675730157986033e-05,
+      "loss": 0.0938,
+      "step": 36455
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6675261356710893e-05,
+      "loss": 0.0731,
+      "step": 36460
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6674792555435753e-05,
+      "loss": 0.1313,
+      "step": 36465
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6674323754160613e-05,
+      "loss": 0.1673,
+      "step": 36470
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6673854952885473e-05,
+      "loss": 0.2234,
+      "step": 36475
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6673386151610333e-05,
+      "loss": 0.3119,
+      "step": 36480
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6672917350335193e-05,
+      "loss": 0.3143,
+      "step": 36485
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6672448549060053e-05,
+      "loss": 0.061,
+      "step": 36490
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6671979747784916e-05,
+      "loss": 0.0817,
+      "step": 36495
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6671510946509776e-05,
+      "loss": 0.0985,
+      "step": 36500
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6671042145234636e-05,
+      "loss": 0.055,
+      "step": 36505
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.66705733439595e-05,
+      "loss": 0.0985,
+      "step": 36510
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.667010454268436e-05,
+      "loss": 0.1559,
+      "step": 36515
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.666963574140922e-05,
+      "loss": 0.1947,
+      "step": 36520
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.666916694013408e-05,
+      "loss": 0.2521,
+      "step": 36525
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.666869813885894e-05,
+      "loss": 0.3802,
+      "step": 36530
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.66682293375838e-05,
+      "loss": 0.4049,
+      "step": 36535
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.666776053630866e-05,
+      "loss": 0.018,
+      "step": 36540
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6667291735033522e-05,
+      "loss": 0.0357,
+      "step": 36545
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6666822933758382e-05,
+      "loss": 0.112,
+      "step": 36550
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6666354132483242e-05,
+      "loss": 0.0682,
+      "step": 36555
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6665885331208102e-05,
+      "loss": 0.1314,
+      "step": 36560
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6665416529932962e-05,
+      "loss": 0.1949,
+      "step": 36565
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.666494772865782e-05,
+      "loss": 0.1298,
+      "step": 36570
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6664478927382685e-05,
+      "loss": 0.1495,
+      "step": 36575
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6664010126107545e-05,
+      "loss": 0.2125,
+      "step": 36580
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6663541324832405e-05,
+      "loss": 0.193,
+      "step": 36585
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6663072523557265e-05,
+      "loss": 0.0462,
+      "step": 36590
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6662603722282128e-05,
+      "loss": 0.087,
+      "step": 36595
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6662134921006988e-05,
+      "loss": 0.0735,
+      "step": 36600
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6661666119731848e-05,
+      "loss": 0.1022,
+      "step": 36605
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6661197318456708e-05,
+      "loss": 0.1351,
+      "step": 36610
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6660728517181568e-05,
+      "loss": 0.1015,
+      "step": 36615
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6660259715906428e-05,
+      "loss": 0.1699,
+      "step": 36620
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6659790914631288e-05,
+      "loss": 0.1358,
+      "step": 36625
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6659322113356148e-05,
+      "loss": 0.3579,
+      "step": 36630
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6658853312081008e-05,
+      "loss": 0.3272,
+      "step": 36635
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.665838451080587e-05,
+      "loss": 0.0505,
+      "step": 36640
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.665791570953073e-05,
+      "loss": 0.1199,
+      "step": 36645
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.665744690825559e-05,
+      "loss": 0.0736,
+      "step": 36650
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6656978106980454e-05,
+      "loss": 0.0974,
+      "step": 36655
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6656509305705314e-05,
+      "loss": 0.0455,
+      "step": 36660
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6656040504430174e-05,
+      "loss": 0.1603,
+      "step": 36665
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6655571703155034e-05,
+      "loss": 0.1358,
+      "step": 36670
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6655102901879894e-05,
+      "loss": 0.1231,
+      "step": 36675
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6654634100604754e-05,
+      "loss": 0.439,
+      "step": 36680
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6654165299329617e-05,
+      "loss": 0.2816,
+      "step": 36685
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6653696498054477e-05,
+      "loss": 0.0421,
+      "step": 36690
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6653227696779337e-05,
+      "loss": 0.0871,
+      "step": 36695
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6652758895504197e-05,
+      "loss": 0.0918,
+      "step": 36700
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6652290094229057e-05,
+      "loss": 0.0698,
+      "step": 36705
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6651821292953917e-05,
+      "loss": 0.1282,
+      "step": 36710
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6651352491678777e-05,
+      "loss": 0.172,
+      "step": 36715
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.665088369040364e-05,
+      "loss": 0.1262,
+      "step": 36720
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.66504148891285e-05,
+      "loss": 0.2513,
+      "step": 36725
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.664994608785336e-05,
+      "loss": 0.4673,
+      "step": 36730
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6649477286578223e-05,
+      "loss": 0.2621,
+      "step": 36735
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6649008485303083e-05,
+      "loss": 0.055,
+      "step": 36740
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6648539684027943e-05,
+      "loss": 0.0606,
+      "step": 36745
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6648070882752803e-05,
+      "loss": 0.0547,
+      "step": 36750
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.6647602081477663e-05,
+      "loss": 0.0559,
+      "step": 36755
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.6647133280202523e-05,
+      "loss": 0.0763,
+      "step": 36760
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.6646664478927383e-05,
+      "loss": 0.0889,
+      "step": 36765
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.6646195677652243e-05,
+      "loss": 0.1193,
+      "step": 36770
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.6645726876377103e-05,
+      "loss": 0.2715,
+      "step": 36775
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.6645258075101966e-05,
+      "loss": 0.2703,
+      "step": 36780
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.6644789273826826e-05,
+      "loss": 0.3361,
+      "step": 36785
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.6644320472551686e-05,
+      "loss": 0.0797,
+      "step": 36790
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.6643851671276546e-05,
+      "loss": 0.0685,
+      "step": 36795
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.664338287000141e-05,
+      "loss": 0.0875,
+      "step": 36800
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.664291406872627e-05,
+      "loss": 0.1153,
+      "step": 36805
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.664244526745113e-05,
+      "loss": 0.0888,
+      "step": 36810
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.664197646617599e-05,
+      "loss": 0.1436,
+      "step": 36815
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.664150766490085e-05,
+      "loss": 0.1164,
+      "step": 36820
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.6641038863625712e-05,
+      "loss": 0.2245,
+      "step": 36825
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.6640570062350572e-05,
+      "loss": 0.2687,
+      "step": 36830
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.6640101261075432e-05,
+      "loss": 0.2908,
+      "step": 36835
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.6639632459800292e-05,
+      "loss": 0.0288,
+      "step": 36840
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.6639163658525152e-05,
+      "loss": 0.0551,
+      "step": 36845
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.6638694857250012e-05,
+      "loss": 0.0861,
+      "step": 36850
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.6638226055974872e-05,
+      "loss": 0.0831,
+      "step": 36855
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.6637757254699735e-05,
+      "loss": 0.1346,
+      "step": 36860
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.6637288453424595e-05,
+      "loss": 0.1029,
+      "step": 36865
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.6636819652149455e-05,
+      "loss": 0.18,
+      "step": 36870
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.6636350850874315e-05,
+      "loss": 0.1931,
+      "step": 36875
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.6635882049599178e-05,
+      "loss": 0.3946,
+      "step": 36880
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.6635413248324038e-05,
+      "loss": 0.3143,
+      "step": 36885
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.6634944447048898e-05,
+      "loss": 0.0555,
+      "step": 36890
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.6634475645773758e-05,
+      "loss": 0.0364,
+      "step": 36895
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.6634006844498618e-05,
+      "loss": 0.0298,
+      "step": 36900
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.6633538043223478e-05,
+      "loss": 0.0705,
+      "step": 36905
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.6633069241948338e-05,
+      "loss": 0.1701,
+      "step": 36910
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.6632600440673198e-05,
+      "loss": 0.1294,
+      "step": 36915
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.663213163939806e-05,
+      "loss": 0.2156,
+      "step": 36920
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.663166283812292e-05,
+      "loss": 0.2613,
+      "step": 36925
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.663119403684778e-05,
+      "loss": 0.3791,
+      "step": 36930
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.663072523557264e-05,
+      "loss": 0.286,
+      "step": 36935
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.6630256434297504e-05,
+      "loss": 0.0744,
+      "step": 36940
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.6629787633022364e-05,
+      "loss": 0.0724,
+      "step": 36945
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.6629318831747224e-05,
+      "loss": 0.076,
+      "step": 36950
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.6628850030472084e-05,
+      "loss": 0.0432,
+      "step": 36955
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.6628381229196944e-05,
+      "loss": 0.1873,
+      "step": 36960
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.6627912427921807e-05,
+      "loss": 0.1363,
+      "step": 36965
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.6627443626646667e-05,
+      "loss": 0.2091,
+      "step": 36970
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.6626974825371527e-05,
+      "loss": 0.2916,
+      "step": 36975
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.6626506024096387e-05,
+      "loss": 0.3069,
+      "step": 36980
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.6626037222821247e-05,
+      "loss": 0.3146,
+      "step": 36985
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.6625568421546107e-05,
+      "loss": 0.0509,
+      "step": 36990
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.6625099620270967e-05,
+      "loss": 0.1014,
+      "step": 36995
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.6624630818995827e-05,
+      "loss": 0.1048,
+      "step": 37000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.662416201772069e-05,
+      "loss": 0.0465,
+      "step": 37005
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.662369321644555e-05,
+      "loss": 0.0972,
+      "step": 37010
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.662322441517041e-05,
+      "loss": 0.1586,
+      "step": 37015
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.6622755613895273e-05,
+      "loss": 0.1031,
+      "step": 37020
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.6622286812620133e-05,
+      "loss": 0.226,
+      "step": 37025
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.6621818011344993e-05,
+      "loss": 0.2767,
+      "step": 37030
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.6621349210069853e-05,
+      "loss": 0.1909,
+      "step": 37035
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.6620880408794713e-05,
+      "loss": 0.0913,
+      "step": 37040
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.6620411607519573e-05,
+      "loss": 0.1058,
+      "step": 37045
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.6619942806244433e-05,
+      "loss": 0.0695,
+      "step": 37050
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.6619474004969296e-05,
+      "loss": 0.0649,
+      "step": 37055
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.6619005203694156e-05,
+      "loss": 0.1028,
+      "step": 37060
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.6618536402419016e-05,
+      "loss": 0.125,
+      "step": 37065
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.6618067601143876e-05,
+      "loss": 0.1258,
+      "step": 37070
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.6617598799868736e-05,
+      "loss": 0.2151,
+      "step": 37075
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.6617129998593596e-05,
+      "loss": 0.2019,
+      "step": 37080
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.661666119731846e-05,
+      "loss": 0.2783,
+      "step": 37085
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.661619239604332e-05,
+      "loss": 0.038,
+      "step": 37090
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.661572359476818e-05,
+      "loss": 0.0266,
+      "step": 37095
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.661525479349304e-05,
+      "loss": 0.0531,
+      "step": 37100
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.6614785992217902e-05,
+      "loss": 0.0735,
+      "step": 37105
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.6614317190942762e-05,
+      "loss": 0.1017,
+      "step": 37110
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.6613848389667622e-05,
+      "loss": 0.0881,
+      "step": 37115
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.6613379588392482e-05,
+      "loss": 0.1619,
+      "step": 37120
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.6612910787117342e-05,
+      "loss": 0.128,
+      "step": 37125
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.6612441985842202e-05,
+      "loss": 0.269,
+      "step": 37130
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.6611973184567062e-05,
+      "loss": 0.2594,
+      "step": 37135
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.6611504383291922e-05,
+      "loss": 0.0342,
+      "step": 37140
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.6611035582016782e-05,
+      "loss": 0.0537,
+      "step": 37145
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.6610566780741645e-05,
+      "loss": 0.1418,
+      "step": 37150
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.6610097979466505e-05,
+      "loss": 0.0692,
+      "step": 37155
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.6609629178191365e-05,
+      "loss": 0.2006,
+      "step": 37160
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.660916037691623e-05,
+      "loss": 0.1502,
+      "step": 37165
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.6608691575641088e-05,
+      "loss": 0.2243,
+      "step": 37170
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.6608222774365948e-05,
+      "loss": 0.1569,
+      "step": 37175
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.6607753973090808e-05,
+      "loss": 0.1983,
+      "step": 37180
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.6607285171815668e-05,
+      "loss": 0.2452,
+      "step": 37185
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.6606816370540528e-05,
+      "loss": 0.0124,
+      "step": 37190
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.660634756926539e-05,
+      "loss": 0.0762,
+      "step": 37195
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.660587876799025e-05,
+      "loss": 0.0947,
+      "step": 37200
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.660540996671511e-05,
+      "loss": 0.07,
+      "step": 37205
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.660494116543997e-05,
+      "loss": 0.0849,
+      "step": 37210
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.660447236416483e-05,
+      "loss": 0.0985,
+      "step": 37215
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.660400356288969e-05,
+      "loss": 0.0841,
+      "step": 37220
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.660353476161455e-05,
+      "loss": 0.1331,
+      "step": 37225
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.6603065960339414e-05,
+      "loss": 0.2944,
+      "step": 37230
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.6602597159064274e-05,
+      "loss": 0.2725,
+      "step": 37235
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.6602128357789134e-05,
+      "loss": 0.0535,
+      "step": 37240
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.6601659556513997e-05,
+      "loss": 0.0732,
+      "step": 37245
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.6601190755238857e-05,
+      "loss": 0.1134,
+      "step": 37250
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.6600721953963717e-05,
+      "loss": 0.1084,
+      "step": 37255
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.6600253152688577e-05,
+      "loss": 0.2014,
+      "step": 37260
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.6599784351413437e-05,
+      "loss": 0.1195,
+      "step": 37265
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.6599315550138297e-05,
+      "loss": 0.1519,
+      "step": 37270
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.6598846748863157e-05,
+      "loss": 0.3407,
+      "step": 37275
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.6598377947588017e-05,
+      "loss": 0.2364,
+      "step": 37280
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.6597909146312877e-05,
+      "loss": 0.251,
+      "step": 37285
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.659744034503774e-05,
+      "loss": 0.0888,
+      "step": 37290
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.65969715437626e-05,
+      "loss": 0.0529,
+      "step": 37295
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.659650274248746e-05,
+      "loss": 0.0577,
+      "step": 37300
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.659603394121232e-05,
+      "loss": 0.0705,
+      "step": 37305
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.6595565139937183e-05,
+      "loss": 0.0848,
+      "step": 37310
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.6595096338662043e-05,
+      "loss": 0.0932,
+      "step": 37315
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.6594627537386903e-05,
+      "loss": 0.2257,
+      "step": 37320
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.6594158736111763e-05,
+      "loss": 0.1595,
+      "step": 37325
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.6593689934836623e-05,
+      "loss": 0.282,
+      "step": 37330
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.6593221133561486e-05,
+      "loss": 0.2518,
+      "step": 37335
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.6592752332286346e-05,
+      "loss": 0.061,
+      "step": 37340
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.6592283531011206e-05,
+      "loss": 0.0853,
+      "step": 37345
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.6591814729736066e-05,
+      "loss": 0.0483,
+      "step": 37350
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.6591345928460926e-05,
+      "loss": 0.0589,
+      "step": 37355
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.6590877127185786e-05,
+      "loss": 0.1197,
+      "step": 37360
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.6590408325910646e-05,
+      "loss": 0.1924,
+      "step": 37365
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.658993952463551e-05,
+      "loss": 0.1366,
+      "step": 37370
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.658947072336037e-05,
+      "loss": 0.2581,
+      "step": 37375
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.658900192208523e-05,
+      "loss": 0.3968,
+      "step": 37380
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.658853312081009e-05,
+      "loss": 0.2865,
+      "step": 37385
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.6588064319534952e-05,
+      "loss": 0.0164,
+      "step": 37390
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.6587595518259812e-05,
+      "loss": 0.0344,
+      "step": 37395
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.6587126716984672e-05,
+      "loss": 0.0603,
+      "step": 37400
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.6586657915709532e-05,
+      "loss": 0.1074,
+      "step": 37405
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.6586189114434392e-05,
+      "loss": 0.1031,
+      "step": 37410
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.6585720313159252e-05,
+      "loss": 0.1761,
+      "step": 37415
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.6585251511884112e-05,
+      "loss": 0.0686,
+      "step": 37420
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.6584782710608972e-05,
+      "loss": 0.1569,
+      "step": 37425
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.6584313909333835e-05,
+      "loss": 0.3677,
+      "step": 37430
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.6583845108058695e-05,
+      "loss": 0.3944,
+      "step": 37435
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.6583376306783555e-05,
+      "loss": 0.0395,
+      "step": 37440
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.6582907505508415e-05,
+      "loss": 0.0894,
+      "step": 37445
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.658243870423328e-05,
+      "loss": 0.0371,
+      "step": 37450
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.658196990295814e-05,
+      "loss": 0.1232,
+      "step": 37455
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.6581501101682998e-05,
+      "loss": 0.1115,
+      "step": 37460
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.6581032300407858e-05,
+      "loss": 0.1376,
+      "step": 37465
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.6580563499132718e-05,
+      "loss": 0.138,
+      "step": 37470
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.658009469785758e-05,
+      "loss": 0.2424,
+      "step": 37475
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.657962589658244e-05,
+      "loss": 0.3563,
+      "step": 37480
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.65791570953073e-05,
+      "loss": 0.2199,
+      "step": 37485
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.657868829403216e-05,
+      "loss": 0.0285,
+      "step": 37490
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.657821949275702e-05,
+      "loss": 0.0596,
+      "step": 37495
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.657775069148188e-05,
+      "loss": 0.0993,
+      "step": 37500
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.657728189020674e-05,
+      "loss": 0.118,
+      "step": 37505
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.65768130889316e-05,
+      "loss": 0.1173,
+      "step": 37510
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.6576344287656464e-05,
+      "loss": 0.1519,
+      "step": 37515
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.6575875486381324e-05,
+      "loss": 0.1847,
+      "step": 37520
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.6575406685106184e-05,
+      "loss": 0.1778,
+      "step": 37525
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.6574937883831048e-05,
+      "loss": 0.399,
+      "step": 37530
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.6574469082555907e-05,
+      "loss": 0.2536,
+      "step": 37535
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.6574000281280767e-05,
+      "loss": 0.0367,
+      "step": 37540
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.6573531480005627e-05,
+      "loss": 0.0508,
+      "step": 37545
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.6573062678730487e-05,
+      "loss": 0.0708,
+      "step": 37550
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.6572593877455347e-05,
+      "loss": 0.1328,
+      "step": 37555
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.6572125076180207e-05,
+      "loss": 0.1137,
+      "step": 37560
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.6571656274905067e-05,
+      "loss": 0.1144,
+      "step": 37565
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.657118747362993e-05,
+      "loss": 0.103,
+      "step": 37570
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.657071867235479e-05,
+      "loss": 0.2853,
+      "step": 37575
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.657024987107965e-05,
+      "loss": 0.3472,
+      "step": 37580
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.656978106980451e-05,
+      "loss": 0.2283,
+      "step": 37585
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.656931226852937e-05,
+      "loss": 0.0364,
+      "step": 37590
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.6568843467254233e-05,
+      "loss": 0.0524,
+      "step": 37595
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.6568374665979093e-05,
+      "loss": 0.1779,
+      "step": 37600
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.6567905864703953e-05,
+      "loss": 0.1489,
+      "step": 37605
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.6567437063428813e-05,
+      "loss": 0.1019,
+      "step": 37610
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.6566968262153677e-05,
+      "loss": 0.2143,
+      "step": 37615
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.6566499460878536e-05,
+      "loss": 0.21,
+      "step": 37620
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.6566030659603396e-05,
+      "loss": 0.1246,
+      "step": 37625
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.6565561858328256e-05,
+      "loss": 0.2592,
+      "step": 37630
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.6565093057053116e-05,
+      "loss": 0.194,
+      "step": 37635
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.6564624255777976e-05,
+      "loss": 0.0456,
+      "step": 37640
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.6564155454502836e-05,
+      "loss": 0.1152,
+      "step": 37645
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.6563686653227696e-05,
+      "loss": 0.0478,
+      "step": 37650
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.6563217851952556e-05,
+      "loss": 0.0457,
+      "step": 37655
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.656274905067742e-05,
+      "loss": 0.1135,
+      "step": 37660
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.656228024940228e-05,
+      "loss": 0.1155,
+      "step": 37665
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.656181144812714e-05,
+      "loss": 0.1544,
+      "step": 37670
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.6561342646852003e-05,
+      "loss": 0.2425,
+      "step": 37675
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.6560873845576862e-05,
+      "loss": 0.3254,
+      "step": 37680
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.6560405044301722e-05,
+      "loss": 0.3002,
+      "step": 37685
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.6559936243026582e-05,
+      "loss": 0.0239,
+      "step": 37690
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.6559467441751442e-05,
+      "loss": 0.0624,
+      "step": 37695
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.6558998640476302e-05,
+      "loss": 0.059,
+      "step": 37700
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.6558529839201166e-05,
+      "loss": 0.0739,
+      "step": 37705
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.6558061037926025e-05,
+      "loss": 0.1165,
+      "step": 37710
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.6557592236650885e-05,
+      "loss": 0.2043,
+      "step": 37715
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.6557123435375745e-05,
+      "loss": 0.1547,
+      "step": 37720
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.6556654634100605e-05,
+      "loss": 0.2518,
+      "step": 37725
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.6556185832825465e-05,
+      "loss": 0.2447,
+      "step": 37730
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.655571703155033e-05,
+      "loss": 0.2753,
+      "step": 37735
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.655524823027519e-05,
+      "loss": 0.039,
+      "step": 37740
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.655477942900005e-05,
+      "loss": 0.034,
+      "step": 37745
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.655431062772491e-05,
+      "loss": 0.1092,
+      "step": 37750
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.655384182644977e-05,
+      "loss": 0.1046,
+      "step": 37755
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.655337302517463e-05,
+      "loss": 0.1291,
+      "step": 37760
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.655290422389949e-05,
+      "loss": 0.0632,
+      "step": 37765
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.655243542262435e-05,
+      "loss": 0.1347,
+      "step": 37770
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.655196662134921e-05,
+      "loss": 0.2168,
+      "step": 37775
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.655149782007407e-05,
+      "loss": 0.2543,
+      "step": 37780
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.655102901879893e-05,
+      "loss": 0.2751,
+      "step": 37785
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.655056021752379e-05,
+      "loss": 0.0621,
+      "step": 37790
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.655009141624865e-05,
+      "loss": 0.0669,
+      "step": 37795
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.6549622614973514e-05,
+      "loss": 0.1206,
+      "step": 37800
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.6549153813698374e-05,
+      "loss": 0.0601,
+      "step": 37805
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.6548685012423234e-05,
+      "loss": 0.1387,
+      "step": 37810
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.6548216211148098e-05,
+      "loss": 0.1075,
+      "step": 37815
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.6547747409872958e-05,
+      "loss": 0.1224,
+      "step": 37820
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.6547278608597817e-05,
+      "loss": 0.2139,
+      "step": 37825
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.6546809807322677e-05,
+      "loss": 0.322,
+      "step": 37830
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.6546341006047537e-05,
+      "loss": 0.2943,
+      "step": 37835
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.6545872204772397e-05,
+      "loss": 0.0263,
+      "step": 37840
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.654540340349726e-05,
+      "loss": 0.1074,
+      "step": 37845
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.654493460222212e-05,
+      "loss": 0.0846,
+      "step": 37850
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.654446580094698e-05,
+      "loss": 0.1118,
+      "step": 37855
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.654399699967184e-05,
+      "loss": 0.123,
+      "step": 37860
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.65435281983967e-05,
+      "loss": 0.101,
+      "step": 37865
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.654305939712156e-05,
+      "loss": 0.1544,
+      "step": 37870
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.654259059584642e-05,
+      "loss": 0.1887,
+      "step": 37875
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.6542121794571284e-05,
+      "loss": 0.2745,
+      "step": 37880
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.6541652993296143e-05,
+      "loss": 0.2631,
+      "step": 37885
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.6541184192021003e-05,
+      "loss": 0.0647,
+      "step": 37890
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.6540715390745867e-05,
+      "loss": 0.0606,
+      "step": 37895
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.6540246589470727e-05,
+      "loss": 0.0729,
+      "step": 37900
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.6539777788195587e-05,
+      "loss": 0.0672,
+      "step": 37905
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.6539308986920447e-05,
+      "loss": 0.0966,
+      "step": 37910
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.6538840185645306e-05,
+      "loss": 0.1284,
+      "step": 37915
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.6538371384370166e-05,
+      "loss": 0.1478,
+      "step": 37920
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.6537902583095026e-05,
+      "loss": 0.1679,
+      "step": 37925
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.6537433781819886e-05,
+      "loss": 0.2213,
+      "step": 37930
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.6536964980544746e-05,
+      "loss": 0.3319,
+      "step": 37935
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.653649617926961e-05,
+      "loss": 0.0198,
+      "step": 37940
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.653602737799447e-05,
+      "loss": 0.0484,
+      "step": 37945
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.653555857671933e-05,
+      "loss": 0.0247,
+      "step": 37950
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.653508977544419e-05,
+      "loss": 0.1661,
+      "step": 37955
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.6534620974169053e-05,
+      "loss": 0.1487,
+      "step": 37960
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.6534152172893913e-05,
+      "loss": 0.1593,
+      "step": 37965
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.6533683371618773e-05,
+      "loss": 0.2306,
+      "step": 37970
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.6533214570343632e-05,
+      "loss": 0.1424,
+      "step": 37975
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.6532745769068492e-05,
+      "loss": 0.1848,
+      "step": 37980
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.6532276967793356e-05,
+      "loss": 0.1803,
+      "step": 37985
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.6531808166518216e-05,
+      "loss": 0.0608,
+      "step": 37990
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.6531339365243076e-05,
+      "loss": 0.0602,
+      "step": 37995
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.6530870563967936e-05,
+      "loss": 0.0948,
+      "step": 38000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.6530401762692795e-05,
+      "loss": 0.0966,
+      "step": 38005
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.6529932961417655e-05,
+      "loss": 0.1048,
+      "step": 38010
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.6529464160142515e-05,
+      "loss": 0.2214,
+      "step": 38015
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.6528995358867375e-05,
+      "loss": 0.1509,
+      "step": 38020
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.652852655759224e-05,
+      "loss": 0.1914,
+      "step": 38025
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.65280577563171e-05,
+      "loss": 0.1909,
+      "step": 38030
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.652758895504196e-05,
+      "loss": 0.2226,
+      "step": 38035
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.6527120153766822e-05,
+      "loss": 0.085,
+      "step": 38040
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.652665135249168e-05,
+      "loss": 0.0693,
+      "step": 38045
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.652618255121654e-05,
+      "loss": 0.0931,
+      "step": 38050
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.65257137499414e-05,
+      "loss": 0.0726,
+      "step": 38055
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.652524494866626e-05,
+      "loss": 0.0906,
+      "step": 38060
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.652477614739112e-05,
+      "loss": 0.1186,
+      "step": 38065
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.652430734611598e-05,
+      "loss": 0.2951,
+      "step": 38070
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.652383854484084e-05,
+      "loss": 0.2605,
+      "step": 38075
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.6523369743565705e-05,
+      "loss": 0.3871,
+      "step": 38080
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.6522900942290565e-05,
+      "loss": 0.1769,
+      "step": 38085
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.6522432141015424e-05,
+      "loss": 0.0579,
+      "step": 38090
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.6521963339740284e-05,
+      "loss": 0.0193,
+      "step": 38095
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.6521494538465144e-05,
+      "loss": 0.03,
+      "step": 38100
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.6521025737190008e-05,
+      "loss": 0.0837,
+      "step": 38105
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.6520556935914868e-05,
+      "loss": 0.1056,
+      "step": 38110
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.6520088134639728e-05,
+      "loss": 0.1369,
+      "step": 38115
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.6519619333364587e-05,
+      "loss": 0.0668,
+      "step": 38120
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.651915053208945e-05,
+      "loss": 0.1117,
+      "step": 38125
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.651868173081431e-05,
+      "loss": 0.3179,
+      "step": 38130
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.651821292953917e-05,
+      "loss": 0.3893,
+      "step": 38135
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.651774412826403e-05,
+      "loss": 0.0323,
+      "step": 38140
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.651727532698889e-05,
+      "loss": 0.0336,
+      "step": 38145
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.651680652571375e-05,
+      "loss": 0.0483,
+      "step": 38150
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.651633772443861e-05,
+      "loss": 0.0455,
+      "step": 38155
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.651586892316347e-05,
+      "loss": 0.0886,
+      "step": 38160
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.6515400121888334e-05,
+      "loss": 0.0876,
+      "step": 38165
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.6514931320613194e-05,
+      "loss": 0.2145,
+      "step": 38170
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.6514462519338054e-05,
+      "loss": 0.1848,
+      "step": 38175
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.6513993718062913e-05,
+      "loss": 0.2946,
+      "step": 38180
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.6513524916787777e-05,
+      "loss": 0.3475,
+      "step": 38185
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.6513056115512637e-05,
+      "loss": 0.0299,
+      "step": 38190
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.6512587314237497e-05,
+      "loss": 0.1046,
+      "step": 38195
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.6512118512962357e-05,
+      "loss": 0.1024,
+      "step": 38200
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.6511649711687216e-05,
+      "loss": 0.0565,
+      "step": 38205
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.6511180910412076e-05,
+      "loss": 0.101,
+      "step": 38210
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.6510712109136936e-05,
+      "loss": 0.0764,
+      "step": 38215
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.65102433078618e-05,
+      "loss": 0.131,
+      "step": 38220
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.650977450658666e-05,
+      "loss": 0.2516,
+      "step": 38225
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.650930570531152e-05,
+      "loss": 0.3604,
+      "step": 38230
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.650883690403638e-05,
+      "loss": 0.1973,
+      "step": 38235
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.650836810276124e-05,
+      "loss": 0.0291,
+      "step": 38240
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.6507899301486103e-05,
+      "loss": 0.0749,
+      "step": 38245
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.6507430500210963e-05,
+      "loss": 0.0902,
+      "step": 38250
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6506961698935823e-05,
+      "loss": 0.0639,
+      "step": 38255
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6506492897660683e-05,
+      "loss": 0.0581,
+      "step": 38260
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6506024096385546e-05,
+      "loss": 0.1075,
+      "step": 38265
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6505555295110406e-05,
+      "loss": 0.1366,
+      "step": 38270
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6505086493835266e-05,
+      "loss": 0.1522,
+      "step": 38275
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6504617692560126e-05,
+      "loss": 0.2609,
+      "step": 38280
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6504148891284986e-05,
+      "loss": 0.2683,
+      "step": 38285
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6503680090009846e-05,
+      "loss": 0.0229,
+      "step": 38290
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6503211288734705e-05,
+      "loss": 0.0375,
+      "step": 38295
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6502742487459565e-05,
+      "loss": 0.056,
+      "step": 38300
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6502273686184425e-05,
+      "loss": 0.1326,
+      "step": 38305
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.650180488490929e-05,
+      "loss": 0.0702,
+      "step": 38310
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.650133608363415e-05,
+      "loss": 0.1117,
+      "step": 38315
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.650086728235901e-05,
+      "loss": 0.1461,
+      "step": 38320
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6500398481083872e-05,
+      "loss": 0.2323,
+      "step": 38325
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6499929679808732e-05,
+      "loss": 0.3953,
+      "step": 38330
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6499460878533592e-05,
+      "loss": 0.2693,
+      "step": 38335
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.649899207725845e-05,
+      "loss": 0.0308,
+      "step": 38340
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.649852327598331e-05,
+      "loss": 0.0674,
+      "step": 38345
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.649805447470817e-05,
+      "loss": 0.0875,
+      "step": 38350
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6497585673433035e-05,
+      "loss": 0.0632,
+      "step": 38355
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6497116872157895e-05,
+      "loss": 0.1288,
+      "step": 38360
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6496648070882755e-05,
+      "loss": 0.1426,
+      "step": 38365
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6496179269607615e-05,
+      "loss": 0.1112,
+      "step": 38370
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6495710468332475e-05,
+      "loss": 0.2597,
+      "step": 38375
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6495241667057335e-05,
+      "loss": 0.4737,
+      "step": 38380
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6494772865782194e-05,
+      "loss": 0.2821,
+      "step": 38385
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6494304064507058e-05,
+      "loss": 0.0208,
+      "step": 38390
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6493835263231918e-05,
+      "loss": 0.0651,
+      "step": 38395
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6493366461956778e-05,
+      "loss": 0.1064,
+      "step": 38400
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.649289766068164e-05,
+      "loss": 0.0954,
+      "step": 38405
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.64924288594065e-05,
+      "loss": 0.1097,
+      "step": 38410
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.649196005813136e-05,
+      "loss": 0.0886,
+      "step": 38415
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.649149125685622e-05,
+      "loss": 0.1399,
+      "step": 38420
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.649102245558108e-05,
+      "loss": 0.1572,
+      "step": 38425
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.649055365430594e-05,
+      "loss": 0.2984,
+      "step": 38430
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.64900848530308e-05,
+      "loss": 0.2507,
+      "step": 38435
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.648961605175566e-05,
+      "loss": 0.076,
+      "step": 38440
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.648914725048052e-05,
+      "loss": 0.0529,
+      "step": 38445
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6488678449205384e-05,
+      "loss": 0.071,
+      "step": 38450
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6488209647930244e-05,
+      "loss": 0.0789,
+      "step": 38455
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6487740846655104e-05,
+      "loss": 0.2206,
+      "step": 38460
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6487272045379964e-05,
+      "loss": 0.1089,
+      "step": 38465
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6486803244104827e-05,
+      "loss": 0.1432,
+      "step": 38470
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6486334442829687e-05,
+      "loss": 0.1734,
+      "step": 38475
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6485865641554547e-05,
+      "loss": 0.3328,
+      "step": 38480
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6485396840279407e-05,
+      "loss": 0.2703,
+      "step": 38485
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6484928039004267e-05,
+      "loss": 0.0144,
+      "step": 38490
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.648445923772913e-05,
+      "loss": 0.0543,
+      "step": 38495
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.648399043645399e-05,
+      "loss": 0.053,
+      "step": 38500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.648352163517885e-05,
+      "loss": 0.1421,
+      "step": 38505
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.648305283390371e-05,
+      "loss": 0.1658,
+      "step": 38510
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.648258403262857e-05,
+      "loss": 0.1931,
+      "step": 38515
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.648211523135343e-05,
+      "loss": 0.185,
+      "step": 38520
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.648164643007829e-05,
+      "loss": 0.1703,
+      "step": 38525
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.648117762880315e-05,
+      "loss": 0.3438,
+      "step": 38530
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6480708827528013e-05,
+      "loss": 0.2311,
+      "step": 38535
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6480240026252873e-05,
+      "loss": 0.0695,
+      "step": 38540
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6479771224977733e-05,
+      "loss": 0.0261,
+      "step": 38545
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6479302423702596e-05,
+      "loss": 0.0674,
+      "step": 38550
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6478833622427456e-05,
+      "loss": 0.0389,
+      "step": 38555
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6478364821152316e-05,
+      "loss": 0.0607,
+      "step": 38560
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6477896019877176e-05,
+      "loss": 0.1092,
+      "step": 38565
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6477427218602036e-05,
+      "loss": 0.1172,
+      "step": 38570
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6476958417326896e-05,
+      "loss": 0.2803,
+      "step": 38575
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6476489616051756e-05,
+      "loss": 0.3335,
+      "step": 38580
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6476020814776616e-05,
+      "loss": 0.2211,
+      "step": 38585
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.647555201350148e-05,
+      "loss": 0.017,
+      "step": 38590
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.647508321222634e-05,
+      "loss": 0.0326,
+      "step": 38595
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.64746144109512e-05,
+      "loss": 0.1442,
+      "step": 38600
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.647414560967606e-05,
+      "loss": 0.0696,
+      "step": 38605
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.647367680840092e-05,
+      "loss": 0.0805,
+      "step": 38610
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6473208007125782e-05,
+      "loss": 0.1016,
+      "step": 38615
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6472739205850642e-05,
+      "loss": 0.1932,
+      "step": 38620
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6472270404575502e-05,
+      "loss": 0.2845,
+      "step": 38625
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.647180160330036e-05,
+      "loss": 0.2894,
+      "step": 38630
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6471332802025225e-05,
+      "loss": 0.2238,
+      "step": 38635
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6470864000750085e-05,
+      "loss": 0.0211,
+      "step": 38640
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6470395199474945e-05,
+      "loss": 0.1023,
+      "step": 38645
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6469926398199805e-05,
+      "loss": 0.0599,
+      "step": 38650
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6469457596924665e-05,
+      "loss": 0.0586,
+      "step": 38655
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6468988795649525e-05,
+      "loss": 0.0788,
+      "step": 38660
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6468519994374385e-05,
+      "loss": 0.0547,
+      "step": 38665
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6468051193099245e-05,
+      "loss": 0.17,
+      "step": 38670
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6467582391824108e-05,
+      "loss": 0.1925,
+      "step": 38675
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6467113590548968e-05,
+      "loss": 0.3395,
+      "step": 38680
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.6466644789273828e-05,
+      "loss": 0.2278,
+      "step": 38685
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.6466175987998688e-05,
+      "loss": 0.1107,
+      "step": 38690
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.646570718672355e-05,
+      "loss": 0.098,
+      "step": 38695
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.646523838544841e-05,
+      "loss": 0.0812,
+      "step": 38700
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.646476958417327e-05,
+      "loss": 0.1406,
+      "step": 38705
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.646430078289813e-05,
+      "loss": 0.0664,
+      "step": 38710
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.646383198162299e-05,
+      "loss": 0.1197,
+      "step": 38715
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.646336318034785e-05,
+      "loss": 0.2626,
+      "step": 38720
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.646289437907271e-05,
+      "loss": 0.131,
+      "step": 38725
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.6462425577797574e-05,
+      "loss": 0.3359,
+      "step": 38730
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.6461956776522434e-05,
+      "loss": 0.3169,
+      "step": 38735
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.6461487975247294e-05,
+      "loss": 0.04,
+      "step": 38740
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.6461019173972154e-05,
+      "loss": 0.05,
+      "step": 38745
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.6460550372697014e-05,
+      "loss": 0.1038,
+      "step": 38750
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.6460081571421877e-05,
+      "loss": 0.077,
+      "step": 38755
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.6459612770146737e-05,
+      "loss": 0.0512,
+      "step": 38760
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.6459143968871597e-05,
+      "loss": 0.1305,
+      "step": 38765
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.6458675167596457e-05,
+      "loss": 0.1466,
+      "step": 38770
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.645820636632132e-05,
+      "loss": 0.2115,
+      "step": 38775
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.645773756504618e-05,
+      "loss": 0.2877,
+      "step": 38780
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.645726876377104e-05,
+      "loss": 0.2355,
+      "step": 38785
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.64567999624959e-05,
+      "loss": 0.0291,
+      "step": 38790
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.645633116122076e-05,
+      "loss": 0.0392,
+      "step": 38795
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.645586235994562e-05,
+      "loss": 0.0575,
+      "step": 38800
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.645539355867048e-05,
+      "loss": 0.1003,
+      "step": 38805
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.645492475739534e-05,
+      "loss": 0.1671,
+      "step": 38810
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.64544559561202e-05,
+      "loss": 0.1108,
+      "step": 38815
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.6453987154845063e-05,
+      "loss": 0.1881,
+      "step": 38820
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.6453518353569923e-05,
+      "loss": 0.2588,
+      "step": 38825
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.6453049552294783e-05,
+      "loss": 0.2032,
+      "step": 38830
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.6452580751019646e-05,
+      "loss": 0.2829,
+      "step": 38835
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.6452111949744506e-05,
+      "loss": 0.0422,
+      "step": 38840
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.6451643148469366e-05,
+      "loss": 0.0484,
+      "step": 38845
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.6451174347194226e-05,
+      "loss": 0.048,
+      "step": 38850
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.6450705545919086e-05,
+      "loss": 0.1151,
+      "step": 38855
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.6450236744643946e-05,
+      "loss": 0.126,
+      "step": 38860
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.6449767943368806e-05,
+      "loss": 0.1759,
+      "step": 38865
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.644929914209367e-05,
+      "loss": 0.1153,
+      "step": 38870
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.644883034081853e-05,
+      "loss": 0.2098,
+      "step": 38875
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.644836153954339e-05,
+      "loss": 0.3271,
+      "step": 38880
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.644789273826825e-05,
+      "loss": 0.3566,
+      "step": 38885
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.644742393699311e-05,
+      "loss": 0.0361,
+      "step": 38890
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.644695513571797e-05,
+      "loss": 0.0997,
+      "step": 38895
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6446486334442832e-05,
+      "loss": 0.0565,
+      "step": 38900
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6446017533167692e-05,
+      "loss": 0.1225,
+      "step": 38905
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6445548731892552e-05,
+      "loss": 0.0687,
+      "step": 38910
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6445079930617415e-05,
+      "loss": 0.151,
+      "step": 38915
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6444611129342275e-05,
+      "loss": 0.1709,
+      "step": 38920
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6444142328067135e-05,
+      "loss": 0.1887,
+      "step": 38925
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6443673526791995e-05,
+      "loss": 0.2965,
+      "step": 38930
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6443204725516855e-05,
+      "loss": 0.2268,
+      "step": 38935
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6442735924241715e-05,
+      "loss": 0.036,
+      "step": 38940
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6442267122966575e-05,
+      "loss": 0.0206,
+      "step": 38945
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6441798321691435e-05,
+      "loss": 0.0413,
+      "step": 38950
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6441329520416295e-05,
+      "loss": 0.1107,
+      "step": 38955
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6440860719141158e-05,
+      "loss": 0.2031,
+      "step": 38960
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6440391917866018e-05,
+      "loss": 0.1474,
+      "step": 38965
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6439923116590878e-05,
+      "loss": 0.1769,
+      "step": 38970
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6439454315315738e-05,
+      "loss": 0.1849,
+      "step": 38975
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.64389855140406e-05,
+      "loss": 0.3524,
+      "step": 38980
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.643851671276546e-05,
+      "loss": 0.3994,
+      "step": 38985
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.643804791149032e-05,
+      "loss": 0.0317,
+      "step": 38990
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.643757911021518e-05,
+      "loss": 0.0741,
+      "step": 38995
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.643711030894004e-05,
+      "loss": 0.0658,
+      "step": 39000
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6436641507664904e-05,
+      "loss": 0.1135,
+      "step": 39005
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6436172706389764e-05,
+      "loss": 0.0792,
+      "step": 39010
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6435703905114624e-05,
+      "loss": 0.0805,
+      "step": 39015
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6435235103839484e-05,
+      "loss": 0.1,
+      "step": 39020
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6434766302564344e-05,
+      "loss": 0.1059,
+      "step": 39025
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6434297501289204e-05,
+      "loss": 0.2548,
+      "step": 39030
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6433828700014064e-05,
+      "loss": 0.2284,
+      "step": 39035
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6433359898738924e-05,
+      "loss": 0.0653,
+      "step": 39040
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6432891097463787e-05,
+      "loss": 0.0668,
+      "step": 39045
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6432422296188647e-05,
+      "loss": 0.0843,
+      "step": 39050
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6431953494913507e-05,
+      "loss": 0.0682,
+      "step": 39055
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.643148469363837e-05,
+      "loss": 0.1166,
+      "step": 39060
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.643101589236323e-05,
+      "loss": 0.1199,
+      "step": 39065
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.643054709108809e-05,
+      "loss": 0.1583,
+      "step": 39070
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.643007828981295e-05,
+      "loss": 0.1735,
+      "step": 39075
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.642960948853781e-05,
+      "loss": 0.2407,
+      "step": 39080
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.642914068726267e-05,
+      "loss": 0.2047,
+      "step": 39085
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.642867188598753e-05,
+      "loss": 0.0946,
+      "step": 39090
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.642820308471239e-05,
+      "loss": 0.1094,
+      "step": 39095
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6427734283437253e-05,
+      "loss": 0.092,
+      "step": 39100
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6427265482162113e-05,
+      "loss": 0.069,
+      "step": 39105
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6426796680886973e-05,
+      "loss": 0.0871,
+      "step": 39110
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6426327879611833e-05,
+      "loss": 0.1514,
+      "step": 39115
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6425859078336693e-05,
+      "loss": 0.2458,
+      "step": 39120
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6425390277061556e-05,
+      "loss": 0.2014,
+      "step": 39125
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6424921475786416e-05,
+      "loss": 0.2175,
+      "step": 39130
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6424452674511276e-05,
+      "loss": 0.2471,
+      "step": 39135
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6423983873236136e-05,
+      "loss": 0.022,
+      "step": 39140
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6423515071961e-05,
+      "loss": 0.0456,
+      "step": 39145
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.642304627068586e-05,
+      "loss": 0.1,
+      "step": 39150
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.642257746941072e-05,
+      "loss": 0.0888,
+      "step": 39155
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.642210866813558e-05,
+      "loss": 0.0947,
+      "step": 39160
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.642163986686044e-05,
+      "loss": 0.141,
+      "step": 39165
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.64211710655853e-05,
+      "loss": 0.1317,
+      "step": 39170
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.642070226431016e-05,
+      "loss": 0.1457,
+      "step": 39175
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.642023346303502e-05,
+      "loss": 0.2767,
+      "step": 39180
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6419764661759882e-05,
+      "loss": 0.2418,
+      "step": 39185
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6419295860484742e-05,
+      "loss": 0.0442,
+      "step": 39190
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6418827059209602e-05,
+      "loss": 0.1675,
+      "step": 39195
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6418358257934465e-05,
+      "loss": 0.0598,
+      "step": 39200
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6417889456659325e-05,
+      "loss": 0.0661,
+      "step": 39205
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6417420655384185e-05,
+      "loss": 0.1537,
+      "step": 39210
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6416951854109045e-05,
+      "loss": 0.0756,
+      "step": 39215
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6416483052833905e-05,
+      "loss": 0.1681,
+      "step": 39220
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6416014251558765e-05,
+      "loss": 0.2648,
+      "step": 39225
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6415545450283625e-05,
+      "loss": 0.3191,
+      "step": 39230
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6415076649008485e-05,
+      "loss": 0.199,
+      "step": 39235
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6414607847733348e-05,
+      "loss": 0.0329,
+      "step": 39240
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6414139046458208e-05,
+      "loss": 0.1063,
+      "step": 39245
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6413670245183068e-05,
+      "loss": 0.0787,
+      "step": 39250
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6413201443907928e-05,
+      "loss": 0.0685,
+      "step": 39255
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6412732642632788e-05,
+      "loss": 0.1426,
+      "step": 39260
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.641226384135765e-05,
+      "loss": 0.1309,
+      "step": 39265
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.641179504008251e-05,
+      "loss": 0.1751,
+      "step": 39270
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.641132623880737e-05,
+      "loss": 0.2695,
+      "step": 39275
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.641085743753223e-05,
+      "loss": 0.1881,
+      "step": 39280
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6410388636257094e-05,
+      "loss": 0.2784,
+      "step": 39285
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6409919834981954e-05,
+      "loss": 0.044,
+      "step": 39290
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6409451033706814e-05,
+      "loss": 0.078,
+      "step": 39295
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6408982232431674e-05,
+      "loss": 0.0706,
+      "step": 39300
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6408513431156534e-05,
+      "loss": 0.0571,
+      "step": 39305
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6408044629881394e-05,
+      "loss": 0.1244,
+      "step": 39310
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6407575828606254e-05,
+      "loss": 0.092,
+      "step": 39315
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6407107027331114e-05,
+      "loss": 0.1246,
+      "step": 39320
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6406638226055974e-05,
+      "loss": 0.15,
+      "step": 39325
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6406169424780837e-05,
+      "loss": 0.2858,
+      "step": 39330
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6405700623505697e-05,
+      "loss": 0.2405,
+      "step": 39335
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6405231822230557e-05,
+      "loss": 0.016,
+      "step": 39340
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.640476302095542e-05,
+      "loss": 0.0616,
+      "step": 39345
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.640429421968028e-05,
+      "loss": 0.0454,
+      "step": 39350
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.640382541840514e-05,
+      "loss": 0.0602,
+      "step": 39355
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.640335661713e-05,
+      "loss": 0.1301,
+      "step": 39360
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.640288781585486e-05,
+      "loss": 0.1133,
+      "step": 39365
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.640241901457972e-05,
+      "loss": 0.1934,
+      "step": 39370
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.640195021330458e-05,
+      "loss": 0.1995,
+      "step": 39375
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6401481412029443e-05,
+      "loss": 0.2625,
+      "step": 39380
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6401012610754303e-05,
+      "loss": 0.2129,
+      "step": 39385
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6400543809479163e-05,
+      "loss": 0.0542,
+      "step": 39390
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6400075008204023e-05,
+      "loss": 0.0899,
+      "step": 39395
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6399606206928883e-05,
+      "loss": 0.1162,
+      "step": 39400
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6399137405653743e-05,
+      "loss": 0.057,
+      "step": 39405
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6398668604378606e-05,
+      "loss": 0.1113,
+      "step": 39410
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6398199803103466e-05,
+      "loss": 0.1153,
+      "step": 39415
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6397731001828326e-05,
+      "loss": 0.1444,
+      "step": 39420
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.639726220055319e-05,
+      "loss": 0.232,
+      "step": 39425
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.639679339927805e-05,
+      "loss": 0.3296,
+      "step": 39430
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.639632459800291e-05,
+      "loss": 0.3146,
+      "step": 39435
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.639585579672777e-05,
+      "loss": 0.0637,
+      "step": 39440
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.639538699545263e-05,
+      "loss": 0.074,
+      "step": 39445
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.639491819417749e-05,
+      "loss": 0.0524,
+      "step": 39450
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.639444939290235e-05,
+      "loss": 0.0591,
+      "step": 39455
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.639398059162721e-05,
+      "loss": 0.1292,
+      "step": 39460
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.639351179035207e-05,
+      "loss": 0.1521,
+      "step": 39465
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6393042989076932e-05,
+      "loss": 0.1973,
+      "step": 39470
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6392574187801792e-05,
+      "loss": 0.1846,
+      "step": 39475
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6392105386526652e-05,
+      "loss": 0.5535,
+      "step": 39480
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6391636585251512e-05,
+      "loss": 0.3364,
+      "step": 39485
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6391167783976375e-05,
+      "loss": 0.0231,
+      "step": 39490
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6390698982701235e-05,
+      "loss": 0.0411,
+      "step": 39495
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6390230181426095e-05,
+      "loss": 0.0506,
+      "step": 39500
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6389761380150955e-05,
+      "loss": 0.3548,
+      "step": 39505
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6389292578875815e-05,
+      "loss": 0.0719,
+      "step": 39510
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6388823777600675e-05,
+      "loss": 0.1005,
+      "step": 39515
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6388354976325538e-05,
+      "loss": 0.1035,
+      "step": 39520
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6387886175050398e-05,
+      "loss": 0.2762,
+      "step": 39525
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6387417373775258e-05,
+      "loss": 0.2817,
+      "step": 39530
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6386948572500118e-05,
+      "loss": 0.2149,
+      "step": 39535
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6386479771224978e-05,
+      "loss": 0.0713,
+      "step": 39540
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6386010969949838e-05,
+      "loss": 0.0805,
+      "step": 39545
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.63855421686747e-05,
+      "loss": 0.0525,
+      "step": 39550
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.638507336739956e-05,
+      "loss": 0.0531,
+      "step": 39555
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.638460456612442e-05,
+      "loss": 0.1241,
+      "step": 39560
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.638413576484928e-05,
+      "loss": 0.2059,
+      "step": 39565
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6383666963574144e-05,
+      "loss": 0.2234,
+      "step": 39570
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6383198162299004e-05,
+      "loss": 0.1771,
+      "step": 39575
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6382729361023864e-05,
+      "loss": 0.3261,
+      "step": 39580
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6382260559748724e-05,
+      "loss": 0.2079,
+      "step": 39585
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6381791758473584e-05,
+      "loss": 0.0562,
+      "step": 39590
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6381322957198444e-05,
+      "loss": 0.062,
+      "step": 39595
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6380854155923304e-05,
+      "loss": 0.11,
+      "step": 39600
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6380385354648164e-05,
+      "loss": 0.0713,
+      "step": 39605
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6379916553373027e-05,
+      "loss": 0.1697,
+      "step": 39610
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6379447752097887e-05,
+      "loss": 0.0959,
+      "step": 39615
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6378978950822747e-05,
+      "loss": 0.1684,
+      "step": 39620
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6378510149547607e-05,
+      "loss": 0.1653,
+      "step": 39625
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.637804134827247e-05,
+      "loss": 0.3247,
+      "step": 39630
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.637757254699733e-05,
+      "loss": 0.2619,
+      "step": 39635
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.637710374572219e-05,
+      "loss": 0.0579,
+      "step": 39640
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.637663494444705e-05,
+      "loss": 0.0433,
+      "step": 39645
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.637616614317191e-05,
+      "loss": 0.046,
+      "step": 39650
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6375697341896773e-05,
+      "loss": 0.1144,
+      "step": 39655
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6375228540621633e-05,
+      "loss": 0.0582,
+      "step": 39660
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6374759739346493e-05,
+      "loss": 0.0816,
+      "step": 39665
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6374290938071353e-05,
+      "loss": 0.306,
+      "step": 39670
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6373822136796213e-05,
+      "loss": 0.1379,
+      "step": 39675
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6373353335521073e-05,
+      "loss": 0.2536,
+      "step": 39680
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6372884534245933e-05,
+      "loss": 0.3197,
+      "step": 39685
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6372415732970793e-05,
+      "loss": 0.0456,
+      "step": 39690
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6371946931695656e-05,
+      "loss": 0.0858,
+      "step": 39695
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6371478130420516e-05,
+      "loss": 0.0263,
+      "step": 39700
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6371009329145376e-05,
+      "loss": 0.0828,
+      "step": 39705
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.637054052787024e-05,
+      "loss": 0.0959,
+      "step": 39710
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.63700717265951e-05,
+      "loss": 0.1384,
+      "step": 39715
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.636960292531996e-05,
+      "loss": 0.1486,
+      "step": 39720
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.636913412404482e-05,
+      "loss": 0.207,
+      "step": 39725
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.636866532276968e-05,
+      "loss": 0.3915,
+      "step": 39730
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.636819652149454e-05,
+      "loss": 0.2934,
+      "step": 39735
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.63677277202194e-05,
+      "loss": 0.0228,
+      "step": 39740
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.636725891894426e-05,
+      "loss": 0.0695,
+      "step": 39745
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6366790117669122e-05,
+      "loss": 0.0624,
+      "step": 39750
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6366321316393982e-05,
+      "loss": 0.0658,
+      "step": 39755
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6365852515118842e-05,
+      "loss": 0.0543,
+      "step": 39760
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6365383713843702e-05,
+      "loss": 0.1347,
+      "step": 39765
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6364914912568562e-05,
+      "loss": 0.4042,
+      "step": 39770
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6364446111293425e-05,
+      "loss": 0.2546,
+      "step": 39775
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6363977310018285e-05,
+      "loss": 0.3395,
+      "step": 39780
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6363508508743145e-05,
+      "loss": 0.253,
+      "step": 39785
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6363039707468005e-05,
+      "loss": 0.0427,
+      "step": 39790
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.636257090619287e-05,
+      "loss": 0.0368,
+      "step": 39795
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.636210210491773e-05,
+      "loss": 0.0467,
+      "step": 39800
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.636163330364259e-05,
+      "loss": 0.1055,
+      "step": 39805
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.636116450236745e-05,
+      "loss": 0.0447,
+      "step": 39810
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6360695701092308e-05,
+      "loss": 0.108,
+      "step": 39815
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6360226899817168e-05,
+      "loss": 0.1594,
+      "step": 39820
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6359758098542028e-05,
+      "loss": 0.294,
+      "step": 39825
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6359289297266888e-05,
+      "loss": 0.3903,
+      "step": 39830
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6358820495991748e-05,
+      "loss": 0.2344,
+      "step": 39835
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.635835169471661e-05,
+      "loss": 0.0489,
+      "step": 39840
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.635788289344147e-05,
+      "loss": 0.0766,
+      "step": 39845
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.635741409216633e-05,
+      "loss": 0.0605,
+      "step": 39850
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6356945290891195e-05,
+      "loss": 0.1088,
+      "step": 39855
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6356476489616054e-05,
+      "loss": 0.0864,
+      "step": 39860
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6356007688340914e-05,
+      "loss": 0.0917,
+      "step": 39865
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6355538887065774e-05,
+      "loss": 0.2046,
+      "step": 39870
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6355070085790634e-05,
+      "loss": 0.2991,
+      "step": 39875
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6354601284515494e-05,
+      "loss": 0.2481,
+      "step": 39880
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6354132483240354e-05,
+      "loss": 0.2429,
+      "step": 39885
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6353663681965217e-05,
+      "loss": 0.0508,
+      "step": 39890
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6353194880690077e-05,
+      "loss": 0.0608,
+      "step": 39895
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6352726079414937e-05,
+      "loss": 0.0733,
+      "step": 39900
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6352257278139797e-05,
+      "loss": 0.0919,
+      "step": 39905
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6351788476864657e-05,
+      "loss": 0.1093,
+      "step": 39910
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6351319675589517e-05,
+      "loss": 0.2387,
+      "step": 39915
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.635085087431438e-05,
+      "loss": 0.1955,
+      "step": 39920
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.635038207303924e-05,
+      "loss": 0.1817,
+      "step": 39925
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.63499132717641e-05,
+      "loss": 0.3849,
+      "step": 39930
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6349444470488964e-05,
+      "loss": 0.2834,
+      "step": 39935
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6348975669213824e-05,
+      "loss": 0.0689,
+      "step": 39940
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6348506867938683e-05,
+      "loss": 0.0568,
+      "step": 39945
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6348038066663543e-05,
+      "loss": 0.0516,
+      "step": 39950
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6347569265388403e-05,
+      "loss": 0.0815,
+      "step": 39955
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6347100464113263e-05,
+      "loss": 0.1282,
+      "step": 39960
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6346631662838123e-05,
+      "loss": 0.1942,
+      "step": 39965
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.6346162861562983e-05,
+      "loss": 0.21,
+      "step": 39970
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.6345694060287843e-05,
+      "loss": 0.2174,
+      "step": 39975
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.6345225259012706e-05,
+      "loss": 0.2777,
+      "step": 39980
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.6344756457737566e-05,
+      "loss": 0.2424,
+      "step": 39985
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.6344287656462426e-05,
+      "loss": 0.0368,
+      "step": 39990
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.6343818855187286e-05,
+      "loss": 0.026,
+      "step": 39995
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.634335005391215e-05,
+      "loss": 0.0642,
+      "step": 40000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.634288125263701e-05,
+      "loss": 0.0958,
+      "step": 40005
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.634241245136187e-05,
+      "loss": 0.1143,
+      "step": 40010
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.634194365008673e-05,
+      "loss": 0.144,
+      "step": 40015
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.634147484881159e-05,
+      "loss": 0.1679,
+      "step": 40020
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.634100604753645e-05,
+      "loss": 0.2176,
+      "step": 40025
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.6340537246261313e-05,
+      "loss": 0.3805,
+      "step": 40030
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.6340068444986172e-05,
+      "loss": 0.2025,
+      "step": 40035
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.6339599643711032e-05,
+      "loss": 0.0308,
+      "step": 40040
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.6339130842435892e-05,
+      "loss": 0.0844,
+      "step": 40045
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.6338662041160752e-05,
+      "loss": 0.0482,
+      "step": 40050
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.6338193239885612e-05,
+      "loss": 0.0883,
+      "step": 40055
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.6337724438610475e-05,
+      "loss": 0.1405,
+      "step": 40060
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.6337255637335335e-05,
+      "loss": 0.0711,
+      "step": 40065
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.6336786836060195e-05,
+      "loss": 0.228,
+      "step": 40070
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.6336318034785055e-05,
+      "loss": 0.1147,
+      "step": 40075
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.633584923350992e-05,
+      "loss": 0.4093,
+      "step": 40080
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.633538043223478e-05,
+      "loss": 0.1311,
+      "step": 40085
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.633491163095964e-05,
+      "loss": 0.0689,
+      "step": 40090
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.63344428296845e-05,
+      "loss": 0.0738,
+      "step": 40095
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.633397402840936e-05,
+      "loss": 0.0874,
+      "step": 40100
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.6333505227134218e-05,
+      "loss": 0.1441,
+      "step": 40105
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.6333036425859078e-05,
+      "loss": 0.1107,
+      "step": 40110
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.6332567624583938e-05,
+      "loss": 0.161,
+      "step": 40115
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.63320988233088e-05,
+      "loss": 0.1873,
+      "step": 40120
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.633163002203366e-05,
+      "loss": 0.1899,
+      "step": 40125
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.633116122075852e-05,
+      "loss": 0.161,
+      "step": 40130
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.633069241948338e-05,
+      "loss": 0.2658,
+      "step": 40135
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.6330223618208245e-05,
+      "loss": 0.06,
+      "step": 40140
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.6329754816933105e-05,
+      "loss": 0.0345,
+      "step": 40145
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.6329286015657964e-05,
+      "loss": 0.084,
+      "step": 40150
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.6328817214382824e-05,
+      "loss": 0.1249,
+      "step": 40155
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.6328348413107684e-05,
+      "loss": 0.12,
+      "step": 40160
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.6327879611832544e-05,
+      "loss": 0.0907,
+      "step": 40165
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.6327410810557408e-05,
+      "loss": 0.1695,
+      "step": 40170
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.6326942009282268e-05,
+      "loss": 0.2406,
+      "step": 40175
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.6326473208007127e-05,
+      "loss": 0.3145,
+      "step": 40180
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6326004406731987e-05,
+      "loss": 0.2744,
+      "step": 40185
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6325535605456847e-05,
+      "loss": 0.0679,
+      "step": 40190
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6325066804181707e-05,
+      "loss": 0.0223,
+      "step": 40195
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6324598002906567e-05,
+      "loss": 0.0352,
+      "step": 40200
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.632412920163143e-05,
+      "loss": 0.0538,
+      "step": 40205
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.632366040035629e-05,
+      "loss": 0.0675,
+      "step": 40210
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.632319159908115e-05,
+      "loss": 0.1217,
+      "step": 40215
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6322722797806014e-05,
+      "loss": 0.109,
+      "step": 40220
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6322253996530874e-05,
+      "loss": 0.1773,
+      "step": 40225
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6321785195255734e-05,
+      "loss": 0.6679,
+      "step": 40230
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6321316393980594e-05,
+      "loss": 0.2886,
+      "step": 40235
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6320847592705453e-05,
+      "loss": 0.0511,
+      "step": 40240
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6320378791430313e-05,
+      "loss": 0.0669,
+      "step": 40245
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6319909990155173e-05,
+      "loss": 0.0664,
+      "step": 40250
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6319441188880033e-05,
+      "loss": 0.0805,
+      "step": 40255
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6318972387604897e-05,
+      "loss": 0.1264,
+      "step": 40260
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6318503586329756e-05,
+      "loss": 0.1922,
+      "step": 40265
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6318034785054616e-05,
+      "loss": 0.1564,
+      "step": 40270
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6317565983779476e-05,
+      "loss": 0.1241,
+      "step": 40275
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6317097182504336e-05,
+      "loss": 0.2404,
+      "step": 40280
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.63166283812292e-05,
+      "loss": 0.2399,
+      "step": 40285
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.631615957995406e-05,
+      "loss": 0.0875,
+      "step": 40290
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.631569077867892e-05,
+      "loss": 0.0858,
+      "step": 40295
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.631522197740378e-05,
+      "loss": 0.0512,
+      "step": 40300
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6314753176128643e-05,
+      "loss": 0.0822,
+      "step": 40305
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6314284374853503e-05,
+      "loss": 0.2328,
+      "step": 40310
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6313815573578363e-05,
+      "loss": 0.1609,
+      "step": 40315
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6313346772303223e-05,
+      "loss": 0.1715,
+      "step": 40320
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6312877971028082e-05,
+      "loss": 0.1636,
+      "step": 40325
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6312409169752942e-05,
+      "loss": 0.2306,
+      "step": 40330
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6311940368477802e-05,
+      "loss": 0.2073,
+      "step": 40335
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6311471567202662e-05,
+      "loss": 0.0513,
+      "step": 40340
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6311002765927522e-05,
+      "loss": 0.0635,
+      "step": 40345
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6310533964652386e-05,
+      "loss": 0.0524,
+      "step": 40350
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6310065163377245e-05,
+      "loss": 0.0711,
+      "step": 40355
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6309596362102105e-05,
+      "loss": 0.0976,
+      "step": 40360
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.630912756082697e-05,
+      "loss": 0.1112,
+      "step": 40365
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.630865875955183e-05,
+      "loss": 0.1517,
+      "step": 40370
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.630818995827669e-05,
+      "loss": 0.181,
+      "step": 40375
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.630772115700155e-05,
+      "loss": 0.2969,
+      "step": 40380
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.630725235572641e-05,
+      "loss": 0.2803,
+      "step": 40385
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.630678355445127e-05,
+      "loss": 0.0327,
+      "step": 40390
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.630631475317613e-05,
+      "loss": 0.0737,
+      "step": 40395
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.630584595190099e-05,
+      "loss": 0.0605,
+      "step": 40400
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.630537715062585e-05,
+      "loss": 0.1205,
+      "step": 40405
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.630490834935071e-05,
+      "loss": 0.0968,
+      "step": 40410
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.630443954807557e-05,
+      "loss": 0.0997,
+      "step": 40415
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.630397074680043e-05,
+      "loss": 0.0928,
+      "step": 40420
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.630350194552529e-05,
+      "loss": 0.1773,
+      "step": 40425
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6303033144250155e-05,
+      "loss": 0.2593,
+      "step": 40430
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6302564342975015e-05,
+      "loss": 0.2245,
+      "step": 40435
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6302095541699874e-05,
+      "loss": 0.0403,
+      "step": 40440
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6301626740424738e-05,
+      "loss": 0.0584,
+      "step": 40445
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6301157939149598e-05,
+      "loss": 0.0841,
+      "step": 40450
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6300689137874458e-05,
+      "loss": 0.1322,
+      "step": 40455
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6300220336599318e-05,
+      "loss": 0.1035,
+      "step": 40460
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6299751535324178e-05,
+      "loss": 0.0839,
+      "step": 40465
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6299282734049037e-05,
+      "loss": 0.1201,
+      "step": 40470
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6298813932773897e-05,
+      "loss": 0.2239,
+      "step": 40475
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6298345131498757e-05,
+      "loss": 0.3387,
+      "step": 40480
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6297876330223617e-05,
+      "loss": 0.2674,
+      "step": 40485
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.629740752894848e-05,
+      "loss": 0.0619,
+      "step": 40490
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.629693872767334e-05,
+      "loss": 0.0873,
+      "step": 40495
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.62964699263982e-05,
+      "loss": 0.0728,
+      "step": 40500
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.629600112512306e-05,
+      "loss": 0.0871,
+      "step": 40505
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6295532323847924e-05,
+      "loss": 0.0998,
+      "step": 40510
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6295063522572784e-05,
+      "loss": 0.1339,
+      "step": 40515
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6294594721297644e-05,
+      "loss": 0.1606,
+      "step": 40520
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6294125920022504e-05,
+      "loss": 0.2007,
+      "step": 40525
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6293657118747363e-05,
+      "loss": 0.2369,
+      "step": 40530
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6293188317472223e-05,
+      "loss": 0.3223,
+      "step": 40535
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6292719516197087e-05,
+      "loss": 0.0264,
+      "step": 40540
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6292250714921947e-05,
+      "loss": 0.0578,
+      "step": 40545
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6291781913646807e-05,
+      "loss": 0.0684,
+      "step": 40550
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6291313112371667e-05,
+      "loss": 0.0905,
+      "step": 40555
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6290844311096526e-05,
+      "loss": 0.1341,
+      "step": 40560
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6290375509821386e-05,
+      "loss": 0.0828,
+      "step": 40565
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.628990670854625e-05,
+      "loss": 0.1134,
+      "step": 40570
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.628943790727111e-05,
+      "loss": 0.1483,
+      "step": 40575
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.628896910599597e-05,
+      "loss": 0.2516,
+      "step": 40580
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6288500304720833e-05,
+      "loss": 0.2559,
+      "step": 40585
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6288031503445693e-05,
+      "loss": 0.0355,
+      "step": 40590
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6287562702170553e-05,
+      "loss": 0.0847,
+      "step": 40595
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6287093900895413e-05,
+      "loss": 0.0847,
+      "step": 40600
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6286625099620273e-05,
+      "loss": 0.0802,
+      "step": 40605
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6286156298345133e-05,
+      "loss": 0.0599,
+      "step": 40610
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6285687497069993e-05,
+      "loss": 0.0885,
+      "step": 40615
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6285218695794852e-05,
+      "loss": 0.161,
+      "step": 40620
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6284749894519712e-05,
+      "loss": 0.2128,
+      "step": 40625
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6284281093244576e-05,
+      "loss": 0.2853,
+      "step": 40630
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6283812291969436e-05,
+      "loss": 0.2575,
+      "step": 40635
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6283343490694296e-05,
+      "loss": 0.0866,
+      "step": 40640
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6282874689419155e-05,
+      "loss": 0.0454,
+      "step": 40645
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.628240588814402e-05,
+      "loss": 0.1026,
+      "step": 40650
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.628193708686888e-05,
+      "loss": 0.0982,
+      "step": 40655
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.628146828559374e-05,
+      "loss": 0.071,
+      "step": 40660
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.62809994843186e-05,
+      "loss": 0.1744,
+      "step": 40665
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.628053068304346e-05,
+      "loss": 0.19,
+      "step": 40670
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.628006188176832e-05,
+      "loss": 0.2066,
+      "step": 40675
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6279593080493182e-05,
+      "loss": 0.3175,
+      "step": 40680
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6279124279218042e-05,
+      "loss": 0.3947,
+      "step": 40685
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.62786554779429e-05,
+      "loss": 0.0417,
+      "step": 40690
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.627818667666776e-05,
+      "loss": 0.05,
+      "step": 40695
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.627771787539262e-05,
+      "loss": 0.0853,
+      "step": 40700
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.627724907411748e-05,
+      "loss": 0.1473,
+      "step": 40705
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.627678027284234e-05,
+      "loss": 0.0938,
+      "step": 40710
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6276311471567205e-05,
+      "loss": 0.1339,
+      "step": 40715
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6275842670292065e-05,
+      "loss": 0.2072,
+      "step": 40720
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6275373869016925e-05,
+      "loss": 0.1397,
+      "step": 40725
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6274905067741788e-05,
+      "loss": 0.2763,
+      "step": 40730
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6274436266466648e-05,
+      "loss": 0.2513,
+      "step": 40735
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6273967465191508e-05,
+      "loss": 0.0345,
+      "step": 40740
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6273498663916368e-05,
+      "loss": 0.021,
+      "step": 40745
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6273029862641228e-05,
+      "loss": 0.0595,
+      "step": 40750
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6272561061366088e-05,
+      "loss": 0.0782,
+      "step": 40755
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6272092260090948e-05,
+      "loss": 0.1077,
+      "step": 40760
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6271623458815807e-05,
+      "loss": 0.1096,
+      "step": 40765
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.627115465754067e-05,
+      "loss": 0.168,
+      "step": 40770
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.627068585626553e-05,
+      "loss": 0.2267,
+      "step": 40775
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.627021705499039e-05,
+      "loss": 0.445,
+      "step": 40780
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.626974825371525e-05,
+      "loss": 0.3413,
+      "step": 40785
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.626927945244011e-05,
+      "loss": 0.0355,
+      "step": 40790
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6268810651164974e-05,
+      "loss": 0.0478,
+      "step": 40795
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6268341849889834e-05,
+      "loss": 0.0609,
+      "step": 40800
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6267873048614694e-05,
+      "loss": 0.1098,
+      "step": 40805
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6267404247339554e-05,
+      "loss": 0.1211,
+      "step": 40810
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6266935446064414e-05,
+      "loss": 0.112,
+      "step": 40815
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6266466644789277e-05,
+      "loss": 0.2154,
+      "step": 40820
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6265997843514137e-05,
+      "loss": 0.1539,
+      "step": 40825
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6265529042238997e-05,
+      "loss": 0.2905,
+      "step": 40830
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6265060240963857e-05,
+      "loss": 0.1497,
+      "step": 40835
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6264591439688717e-05,
+      "loss": 0.0331,
+      "step": 40840
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6264122638413577e-05,
+      "loss": 0.0605,
+      "step": 40845
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6263653837138436e-05,
+      "loss": 0.0846,
+      "step": 40850
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6263185035863296e-05,
+      "loss": 0.0709,
+      "step": 40855
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.626271623458816e-05,
+      "loss": 0.0923,
+      "step": 40860
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.626224743331302e-05,
+      "loss": 0.1565,
+      "step": 40865
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.626177863203788e-05,
+      "loss": 0.1856,
+      "step": 40870
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6261309830762743e-05,
+      "loss": 0.2154,
+      "step": 40875
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6260841029487603e-05,
+      "loss": 0.4401,
+      "step": 40880
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6260372228212463e-05,
+      "loss": 0.2305,
+      "step": 40885
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6259903426937323e-05,
+      "loss": 0.0609,
+      "step": 40890
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6259434625662183e-05,
+      "loss": 0.0389,
+      "step": 40895
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6258965824387043e-05,
+      "loss": 0.1174,
+      "step": 40900
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6258497023111903e-05,
+      "loss": 0.0794,
+      "step": 40905
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6258028221836766e-05,
+      "loss": 0.0679,
+      "step": 40910
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6257559420561626e-05,
+      "loss": 0.2702,
+      "step": 40915
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6257090619286486e-05,
+      "loss": 0.1589,
+      "step": 40920
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6256621818011346e-05,
+      "loss": 0.2221,
+      "step": 40925
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6256153016736206e-05,
+      "loss": 0.3977,
+      "step": 40930
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.625568421546107e-05,
+      "loss": 0.2443,
+      "step": 40935
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.625521541418593e-05,
+      "loss": 0.0319,
+      "step": 40940
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.625474661291079e-05,
+      "loss": 0.0884,
+      "step": 40945
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.625427781163565e-05,
+      "loss": 0.0723,
+      "step": 40950
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6253809010360512e-05,
+      "loss": 0.0731,
+      "step": 40955
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6253340209085372e-05,
+      "loss": 0.1685,
+      "step": 40960
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6252871407810232e-05,
+      "loss": 0.1155,
+      "step": 40965
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6252402606535092e-05,
+      "loss": 0.1788,
+      "step": 40970
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6251933805259952e-05,
+      "loss": 0.2254,
+      "step": 40975
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6251465003984812e-05,
+      "loss": 0.2898,
+      "step": 40980
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.625099620270967e-05,
+      "loss": 0.2073,
+      "step": 40985
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.625052740143453e-05,
+      "loss": 0.0784,
+      "step": 40990
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.625005860015939e-05,
+      "loss": 0.0628,
+      "step": 40995
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6249589798884255e-05,
+      "loss": 0.1004,
+      "step": 41000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6249120997609115e-05,
+      "loss": 0.1191,
+      "step": 41005
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6248652196333975e-05,
+      "loss": 0.2033,
+      "step": 41010
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6248183395058838e-05,
+      "loss": 0.1165,
+      "step": 41015
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6247714593783698e-05,
+      "loss": 0.1041,
+      "step": 41020
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6247245792508558e-05,
+      "loss": 0.3128,
+      "step": 41025
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6246776991233418e-05,
+      "loss": 0.2066,
+      "step": 41030
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6246308189958278e-05,
+      "loss": 0.1945,
+      "step": 41035
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6245839388683138e-05,
+      "loss": 0.0559,
+      "step": 41040
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.6245370587407998e-05,
+      "loss": 0.0498,
+      "step": 41045
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.624490178613286e-05,
+      "loss": 0.0774,
+      "step": 41050
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.624443298485772e-05,
+      "loss": 0.134,
+      "step": 41055
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.624396418358258e-05,
+      "loss": 0.0899,
+      "step": 41060
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.624349538230744e-05,
+      "loss": 0.0978,
+      "step": 41065
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.62430265810323e-05,
+      "loss": 0.1787,
+      "step": 41070
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.624255777975716e-05,
+      "loss": 0.2584,
+      "step": 41075
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.6242088978482024e-05,
+      "loss": 0.4001,
+      "step": 41080
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.6241620177206884e-05,
+      "loss": 0.2551,
+      "step": 41085
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.6241151375931744e-05,
+      "loss": 0.0141,
+      "step": 41090
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.6240682574656607e-05,
+      "loss": 0.0182,
+      "step": 41095
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.6240213773381467e-05,
+      "loss": 0.0712,
+      "step": 41100
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.6239744972106327e-05,
+      "loss": 0.0826,
+      "step": 41105
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.6239276170831187e-05,
+      "loss": 0.0982,
+      "step": 41110
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.6238807369556047e-05,
+      "loss": 0.0902,
+      "step": 41115
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.6238338568280907e-05,
+      "loss": 0.1409,
+      "step": 41120
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.6237869767005767e-05,
+      "loss": 0.0901,
+      "step": 41125
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.6237400965730627e-05,
+      "loss": 0.2791,
+      "step": 41130
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.6236932164455487e-05,
+      "loss": 0.3241,
+      "step": 41135
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.6236463363180347e-05,
+      "loss": 0.0624,
+      "step": 41140
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.623599456190521e-05,
+      "loss": 0.0258,
+      "step": 41145
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.623552576063007e-05,
+      "loss": 0.0758,
+      "step": 41150
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.623505695935493e-05,
+      "loss": 0.0531,
+      "step": 41155
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.6234588158079793e-05,
+      "loss": 0.0854,
+      "step": 41160
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.6234119356804653e-05,
+      "loss": 0.1209,
+      "step": 41165
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.6233650555529513e-05,
+      "loss": 0.1532,
+      "step": 41170
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.6233181754254373e-05,
+      "loss": 0.1413,
+      "step": 41175
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.6232712952979233e-05,
+      "loss": 0.2444,
+      "step": 41180
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.6232244151704093e-05,
+      "loss": 0.2288,
+      "step": 41185
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.6231775350428956e-05,
+      "loss": 0.0482,
+      "step": 41190
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.6231306549153816e-05,
+      "loss": 0.0709,
+      "step": 41195
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.6230837747878676e-05,
+      "loss": 0.1429,
+      "step": 41200
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.6230368946603536e-05,
+      "loss": 0.0677,
+      "step": 41205
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.6229900145328396e-05,
+      "loss": 0.1367,
+      "step": 41210
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.6229431344053256e-05,
+      "loss": 0.091,
+      "step": 41215
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.6228962542778116e-05,
+      "loss": 0.1481,
+      "step": 41220
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.622849374150298e-05,
+      "loss": 0.1727,
+      "step": 41225
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.622802494022784e-05,
+      "loss": 0.2958,
+      "step": 41230
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.62275561389527e-05,
+      "loss": 0.4008,
+      "step": 41235
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.6227087337677562e-05,
+      "loss": 0.0504,
+      "step": 41240
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.6226618536402422e-05,
+      "loss": 0.0635,
+      "step": 41245
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.6226149735127282e-05,
+      "loss": 0.0534,
+      "step": 41250
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.6225680933852142e-05,
+      "loss": 0.0808,
+      "step": 41255
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.6225212132577002e-05,
+      "loss": 0.1256,
+      "step": 41260
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.6224743331301862e-05,
+      "loss": 0.2142,
+      "step": 41265
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.6224274530026722e-05,
+      "loss": 0.1487,
+      "step": 41270
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.622380572875158e-05,
+      "loss": 0.2907,
+      "step": 41275
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.622333692747644e-05,
+      "loss": 0.4025,
+      "step": 41280
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.6222868126201305e-05,
+      "loss": 0.2121,
+      "step": 41285
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.6222399324926165e-05,
+      "loss": 0.0294,
+      "step": 41290
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.6221930523651025e-05,
+      "loss": 0.061,
+      "step": 41295
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.6221461722375885e-05,
+      "loss": 0.0598,
+      "step": 41300
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.6220992921100748e-05,
+      "loss": 0.1083,
+      "step": 41305
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.6220524119825608e-05,
+      "loss": 0.1211,
+      "step": 41310
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.6220055318550468e-05,
+      "loss": 0.1688,
+      "step": 41315
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.6219586517275328e-05,
+      "loss": 0.1301,
+      "step": 41320
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.6219117716000188e-05,
+      "loss": 0.1663,
+      "step": 41325
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.621864891472505e-05,
+      "loss": 0.1977,
+      "step": 41330
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.621818011344991e-05,
+      "loss": 0.2396,
+      "step": 41335
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.621771131217477e-05,
+      "loss": 0.0523,
+      "step": 41340
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.621724251089963e-05,
+      "loss": 0.0638,
+      "step": 41345
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.621677370962449e-05,
+      "loss": 0.0662,
+      "step": 41350
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.621630490834935e-05,
+      "loss": 0.1036,
+      "step": 41355
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.621583610707421e-05,
+      "loss": 0.0878,
+      "step": 41360
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.6215367305799074e-05,
+      "loss": 0.1043,
+      "step": 41365
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.6214898504523934e-05,
+      "loss": 0.1287,
+      "step": 41370
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.6214429703248794e-05,
+      "loss": 0.2675,
+      "step": 41375
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.6213960901973654e-05,
+      "loss": 0.2608,
+      "step": 41380
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.6213492100698517e-05,
+      "loss": 0.2334,
+      "step": 41385
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.6213023299423377e-05,
+      "loss": 0.0367,
+      "step": 41390
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.6212554498148237e-05,
+      "loss": 0.0891,
+      "step": 41395
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.6212085696873097e-05,
+      "loss": 0.0596,
+      "step": 41400
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.6211616895597957e-05,
+      "loss": 0.1102,
+      "step": 41405
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.6211148094322817e-05,
+      "loss": 0.0846,
+      "step": 41410
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.6210679293047677e-05,
+      "loss": 0.2469,
+      "step": 41415
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.621021049177254e-05,
+      "loss": 0.1392,
+      "step": 41420
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.62097416904974e-05,
+      "loss": 0.223,
+      "step": 41425
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.620927288922226e-05,
+      "loss": 0.3501,
+      "step": 41430
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.620880408794712e-05,
+      "loss": 0.1642,
+      "step": 41435
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.620833528667198e-05,
+      "loss": 0.0356,
+      "step": 41440
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.6207866485396843e-05,
+      "loss": 0.1539,
+      "step": 41445
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.6207397684121703e-05,
+      "loss": 0.1197,
+      "step": 41450
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.6206928882846563e-05,
+      "loss": 0.0775,
+      "step": 41455
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.6206460081571423e-05,
+      "loss": 0.0948,
+      "step": 41460
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.6205991280296283e-05,
+      "loss": 0.1299,
+      "step": 41465
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.6205522479021146e-05,
+      "loss": 0.1459,
+      "step": 41470
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.6205053677746006e-05,
+      "loss": 0.187,
+      "step": 41475
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.6204584876470866e-05,
+      "loss": 0.3319,
+      "step": 41480
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.6204116075195726e-05,
+      "loss": 0.1926,
+      "step": 41485
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.6203647273920586e-05,
+      "loss": 0.0347,
+      "step": 41490
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.6203178472645446e-05,
+      "loss": 0.0729,
+      "step": 41495
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.6202709671370306e-05,
+      "loss": 0.0343,
+      "step": 41500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.6202240870095166e-05,
+      "loss": 0.1035,
+      "step": 41505
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.620177206882003e-05,
+      "loss": 0.0586,
+      "step": 41510
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.620130326754489e-05,
+      "loss": 0.1304,
+      "step": 41515
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.620083446626975e-05,
+      "loss": 0.0746,
+      "step": 41520
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.6200365664994612e-05,
+      "loss": 0.2772,
+      "step": 41525
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.6199896863719472e-05,
+      "loss": 0.3602,
+      "step": 41530
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.6199428062444332e-05,
+      "loss": 0.2218,
+      "step": 41535
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.6198959261169192e-05,
+      "loss": 0.0288,
+      "step": 41540
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.6198490459894052e-05,
+      "loss": 0.0953,
+      "step": 41545
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.6198021658618912e-05,
+      "loss": 0.0825,
+      "step": 41550
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.6197552857343772e-05,
+      "loss": 0.0721,
+      "step": 41555
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.6197084056068635e-05,
+      "loss": 0.0856,
+      "step": 41560
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.6196615254793495e-05,
+      "loss": 0.1164,
+      "step": 41565
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.6196146453518355e-05,
+      "loss": 0.1362,
+      "step": 41570
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.6195677652243215e-05,
+      "loss": 0.1411,
+      "step": 41575
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.6195208850968075e-05,
+      "loss": 0.3744,
+      "step": 41580
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.6194740049692935e-05,
+      "loss": 0.218,
+      "step": 41585
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.6194271248417798e-05,
+      "loss": 0.029,
+      "step": 41590
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.6193802447142658e-05,
+      "loss": 0.0284,
+      "step": 41595
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.6193333645867518e-05,
+      "loss": 0.1183,
+      "step": 41600
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.6192864844592378e-05,
+      "loss": 0.1304,
+      "step": 41605
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.619239604331724e-05,
+      "loss": 0.1123,
+      "step": 41610
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.61919272420421e-05,
+      "loss": 0.1057,
+      "step": 41615
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.619145844076696e-05,
+      "loss": 0.2111,
+      "step": 41620
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.619098963949182e-05,
+      "loss": 0.1933,
+      "step": 41625
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.619052083821668e-05,
+      "loss": 0.2101,
+      "step": 41630
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.619005203694154e-05,
+      "loss": 0.4121,
+      "step": 41635
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.61895832356664e-05,
+      "loss": 0.0507,
+      "step": 41640
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.618911443439126e-05,
+      "loss": 0.0774,
+      "step": 41645
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.618864563311612e-05,
+      "loss": 0.0937,
+      "step": 41650
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.6188176831840984e-05,
+      "loss": 0.104,
+      "step": 41655
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.6187708030565844e-05,
+      "loss": 0.0718,
+      "step": 41660
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.6187239229290704e-05,
+      "loss": 0.2121,
+      "step": 41665
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.6186770428015567e-05,
+      "loss": 0.1222,
+      "step": 41670
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.6186301626740427e-05,
+      "loss": 0.1626,
+      "step": 41675
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.6185832825465287e-05,
+      "loss": 0.2618,
+      "step": 41680
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.6185364024190147e-05,
+      "loss": 0.1554,
+      "step": 41685
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.6184895222915007e-05,
+      "loss": 0.0277,
+      "step": 41690
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.6184426421639867e-05,
+      "loss": 0.0889,
+      "step": 41695
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.618395762036473e-05,
+      "loss": 0.0996,
+      "step": 41700
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.618348881908959e-05,
+      "loss": 0.0605,
+      "step": 41705
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.618302001781445e-05,
+      "loss": 0.1412,
+      "step": 41710
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.618255121653931e-05,
+      "loss": 0.1696,
+      "step": 41715
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.618208241526417e-05,
+      "loss": 0.1668,
+      "step": 41720
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.618161361398903e-05,
+      "loss": 0.198,
+      "step": 41725
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.618114481271389e-05,
+      "loss": 0.3409,
+      "step": 41730
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.6180676011438753e-05,
+      "loss": 0.1873,
+      "step": 41735
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.6180207210163613e-05,
+      "loss": 0.0503,
+      "step": 41740
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.6179738408888473e-05,
+      "loss": 0.0428,
+      "step": 41745
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.6179269607613336e-05,
+      "loss": 0.0873,
+      "step": 41750
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.6178800806338196e-05,
+      "loss": 0.0582,
+      "step": 41755
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.6178332005063056e-05,
+      "loss": 0.1057,
+      "step": 41760
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.6177863203787916e-05,
+      "loss": 0.217,
+      "step": 41765
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.6177394402512776e-05,
+      "loss": 0.0945,
+      "step": 41770
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.6176925601237636e-05,
+      "loss": 0.2702,
+      "step": 41775
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.6176456799962496e-05,
+      "loss": 0.2901,
+      "step": 41780
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.6175987998687356e-05,
+      "loss": 0.2701,
+      "step": 41785
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.6175519197412216e-05,
+      "loss": 0.0123,
+      "step": 41790
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.617505039613708e-05,
+      "loss": 0.0755,
+      "step": 41795
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.617458159486194e-05,
+      "loss": 0.058,
+      "step": 41800
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.61741127935868e-05,
+      "loss": 0.0831,
+      "step": 41805
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.617364399231166e-05,
+      "loss": 0.0736,
+      "step": 41810
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.6173175191036522e-05,
+      "loss": 0.1257,
+      "step": 41815
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.6172706389761382e-05,
+      "loss": 0.147,
+      "step": 41820
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.6172237588486242e-05,
+      "loss": 0.1401,
+      "step": 41825
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.6171768787211102e-05,
+      "loss": 0.3274,
+      "step": 41830
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.6171299985935962e-05,
+      "loss": 0.2624,
+      "step": 41835
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.6170831184660825e-05,
+      "loss": 0.0651,
+      "step": 41840
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.6170362383385685e-05,
+      "loss": 0.0283,
+      "step": 41845
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.6169893582110545e-05,
+      "loss": 0.0477,
+      "step": 41850
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.6169424780835405e-05,
+      "loss": 0.061,
+      "step": 41855
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.6168955979560265e-05,
+      "loss": 0.1343,
+      "step": 41860
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.6168487178285125e-05,
+      "loss": 0.1538,
+      "step": 41865
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.6168018377009985e-05,
+      "loss": 0.21,
+      "step": 41870
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.6167549575734848e-05,
+      "loss": 0.1877,
+      "step": 41875
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.6167080774459708e-05,
+      "loss": 0.3532,
+      "step": 41880
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.6166611973184568e-05,
+      "loss": 0.3586,
+      "step": 41885
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.6166143171909428e-05,
+      "loss": 0.0907,
+      "step": 41890
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.616567437063429e-05,
+      "loss": 0.0542,
+      "step": 41895
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.616520556935915e-05,
+      "loss": 0.0861,
+      "step": 41900
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.616473676808401e-05,
+      "loss": 0.0796,
+      "step": 41905
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.616426796680887e-05,
+      "loss": 0.1031,
+      "step": 41910
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.616379916553373e-05,
+      "loss": 0.0803,
+      "step": 41915
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.616333036425859e-05,
+      "loss": 0.1137,
+      "step": 41920
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.616286156298345e-05,
+      "loss": 0.2324,
+      "step": 41925
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.616239276170831e-05,
+      "loss": 0.2309,
+      "step": 41930
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.6161923960433174e-05,
+      "loss": 0.2706,
+      "step": 41935
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.6161455159158034e-05,
+      "loss": 0.0404,
+      "step": 41940
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.6160986357882894e-05,
+      "loss": 0.0287,
+      "step": 41945
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.6160517556607754e-05,
+      "loss": 0.1108,
+      "step": 41950
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.6160048755332617e-05,
+      "loss": 0.0581,
+      "step": 41955
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.6159579954057477e-05,
+      "loss": 0.1313,
+      "step": 41960
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.6159111152782337e-05,
+      "loss": 0.1299,
+      "step": 41965
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.6158642351507197e-05,
+      "loss": 0.1494,
+      "step": 41970
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.6158173550232057e-05,
+      "loss": 0.1915,
+      "step": 41975
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.615770474895692e-05,
+      "loss": 0.3193,
+      "step": 41980
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.615723594768178e-05,
+      "loss": 0.1987,
+      "step": 41985
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.615676714640664e-05,
+      "loss": 0.0334,
+      "step": 41990
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.61562983451315e-05,
+      "loss": 0.0295,
+      "step": 41995
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.615582954385636e-05,
+      "loss": 0.0684,
+      "step": 42000
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.615536074258122e-05,
+      "loss": 0.1138,
+      "step": 42005
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.615489194130608e-05,
+      "loss": 0.1429,
+      "step": 42010
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.615442314003094e-05,
+      "loss": 0.0981,
+      "step": 42015
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.6153954338755803e-05,
+      "loss": 0.1291,
+      "step": 42020
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.6153485537480663e-05,
+      "loss": 0.1244,
+      "step": 42025
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.6153016736205523e-05,
+      "loss": 0.2555,
+      "step": 42030
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.6152547934930386e-05,
+      "loss": 0.261,
+      "step": 42035
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.6152079133655246e-05,
+      "loss": 0.0461,
+      "step": 42040
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.6151610332380106e-05,
+      "loss": 0.0987,
+      "step": 42045
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.6151141531104966e-05,
+      "loss": 0.0703,
+      "step": 42050
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.6150672729829826e-05,
+      "loss": 0.0625,
+      "step": 42055
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.6150203928554686e-05,
+      "loss": 0.0667,
+      "step": 42060
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.6149735127279546e-05,
+      "loss": 0.1145,
+      "step": 42065
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.614926632600441e-05,
+      "loss": 0.1158,
+      "step": 42070
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.614879752472927e-05,
+      "loss": 0.1947,
+      "step": 42075
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.614832872345413e-05,
+      "loss": 0.3018,
+      "step": 42080
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.614785992217899e-05,
+      "loss": 0.3088,
+      "step": 42085
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.614739112090385e-05,
+      "loss": 0.0916,
+      "step": 42090
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.614692231962871e-05,
+      "loss": 0.0213,
+      "step": 42095
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.6146453518353572e-05,
+      "loss": 0.0558,
+      "step": 42100
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.6145984717078432e-05,
+      "loss": 0.0763,
+      "step": 42105
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.6145515915803292e-05,
+      "loss": 0.1617,
+      "step": 42110
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.6145047114528152e-05,
+      "loss": 0.0917,
+      "step": 42115
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.6144578313253015e-05,
+      "loss": 0.2151,
+      "step": 42120
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.6144109511977875e-05,
+      "loss": 0.1872,
+      "step": 42125
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.6143640710702735e-05,
+      "loss": 0.2114,
+      "step": 42130
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.6143171909427595e-05,
+      "loss": 0.2838,
+      "step": 42135
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.6142703108152455e-05,
+      "loss": 0.0481,
+      "step": 42140
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.6142234306877315e-05,
+      "loss": 0.0809,
+      "step": 42145
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.6141765505602175e-05,
+      "loss": 0.0956,
+      "step": 42150
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.6141296704327035e-05,
+      "loss": 0.1684,
+      "step": 42155
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.6140827903051895e-05,
+      "loss": 0.0701,
+      "step": 42160
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.6140359101776758e-05,
+      "loss": 0.1664,
+      "step": 42165
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.6139890300501618e-05,
+      "loss": 0.0904,
+      "step": 42170
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.6139421499226478e-05,
+      "loss": 0.1441,
+      "step": 42175
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.613895269795134e-05,
+      "loss": 0.2687,
+      "step": 42180
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.61384838966762e-05,
+      "loss": 0.184,
+      "step": 42185
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.613801509540106e-05,
+      "loss": 0.0325,
+      "step": 42190
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.613754629412592e-05,
+      "loss": 0.0627,
+      "step": 42195
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.613707749285078e-05,
+      "loss": 0.0675,
+      "step": 42200
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.613660869157564e-05,
+      "loss": 0.0689,
+      "step": 42205
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.6136139890300504e-05,
+      "loss": 0.0447,
+      "step": 42210
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.6135671089025364e-05,
+      "loss": 0.1771,
+      "step": 42215
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.6135202287750224e-05,
+      "loss": 0.1027,
+      "step": 42220
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.6134733486475084e-05,
+      "loss": 0.1719,
+      "step": 42225
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.6134264685199944e-05,
+      "loss": 0.4006,
+      "step": 42230
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.6133795883924804e-05,
+      "loss": 0.4029,
+      "step": 42235
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.6133327082649664e-05,
+      "loss": 0.0602,
+      "step": 42240
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.6132858281374527e-05,
+      "loss": 0.1014,
+      "step": 42245
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.6132389480099387e-05,
+      "loss": 0.1154,
+      "step": 42250
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.6131920678824247e-05,
+      "loss": 0.0592,
+      "step": 42255
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.613145187754911e-05,
+      "loss": 0.1081,
+      "step": 42260
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.613098307627397e-05,
+      "loss": 0.1918,
+      "step": 42265
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.613051427499883e-05,
+      "loss": 0.14,
+      "step": 42270
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.613004547372369e-05,
+      "loss": 0.1181,
+      "step": 42275
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.612957667244855e-05,
+      "loss": 0.3222,
+      "step": 42280
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.612910787117341e-05,
+      "loss": 0.1738,
+      "step": 42285
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.612863906989827e-05,
+      "loss": 0.0451,
+      "step": 42290
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.612817026862313e-05,
+      "loss": 0.0387,
+      "step": 42295
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.612770146734799e-05,
+      "loss": 0.062,
+      "step": 42300
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.6127232666072853e-05,
+      "loss": 0.1021,
+      "step": 42305
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.6126763864797713e-05,
+      "loss": 0.1734,
+      "step": 42310
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.6126295063522573e-05,
+      "loss": 0.1267,
+      "step": 42315
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.6125826262247433e-05,
+      "loss": 0.156,
+      "step": 42320
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.6125357460972296e-05,
+      "loss": 0.1571,
+      "step": 42325
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6124888659697156e-05,
+      "loss": 0.3486,
+      "step": 42330
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6124419858422016e-05,
+      "loss": 0.2881,
+      "step": 42335
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6123951057146876e-05,
+      "loss": 0.0239,
+      "step": 42340
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6123482255871736e-05,
+      "loss": 0.0531,
+      "step": 42345
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.61230134545966e-05,
+      "loss": 0.0873,
+      "step": 42350
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.612254465332146e-05,
+      "loss": 0.0827,
+      "step": 42355
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.612207585204632e-05,
+      "loss": 0.0947,
+      "step": 42360
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.612160705077118e-05,
+      "loss": 0.0902,
+      "step": 42365
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.612113824949604e-05,
+      "loss": 0.1753,
+      "step": 42370
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.61206694482209e-05,
+      "loss": 0.1915,
+      "step": 42375
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.612020064694576e-05,
+      "loss": 0.3008,
+      "step": 42380
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6119731845670622e-05,
+      "loss": 0.3702,
+      "step": 42385
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6119263044395482e-05,
+      "loss": 0.0867,
+      "step": 42390
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6118794243120342e-05,
+      "loss": 0.0771,
+      "step": 42395
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6118325441845206e-05,
+      "loss": 0.0406,
+      "step": 42400
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6117856640570066e-05,
+      "loss": 0.0628,
+      "step": 42405
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6117387839294926e-05,
+      "loss": 0.0704,
+      "step": 42410
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6116919038019785e-05,
+      "loss": 0.075,
+      "step": 42415
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6116450236744645e-05,
+      "loss": 0.1656,
+      "step": 42420
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6115981435469505e-05,
+      "loss": 0.1958,
+      "step": 42425
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6115512634194365e-05,
+      "loss": 0.2968,
+      "step": 42430
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6115043832919225e-05,
+      "loss": 0.2463,
+      "step": 42435
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6114575031644085e-05,
+      "loss": 0.0686,
+      "step": 42440
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.611410623036895e-05,
+      "loss": 0.0637,
+      "step": 42445
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.611363742909381e-05,
+      "loss": 0.0961,
+      "step": 42450
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.611316862781867e-05,
+      "loss": 0.0605,
+      "step": 42455
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6112699826543528e-05,
+      "loss": 0.0825,
+      "step": 42460
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.611223102526839e-05,
+      "loss": 0.1368,
+      "step": 42465
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.611176222399325e-05,
+      "loss": 0.1426,
+      "step": 42470
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.611129342271811e-05,
+      "loss": 0.1738,
+      "step": 42475
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.611082462144297e-05,
+      "loss": 0.271,
+      "step": 42480
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.611035582016783e-05,
+      "loss": 0.2163,
+      "step": 42485
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6109887018892695e-05,
+      "loss": 0.048,
+      "step": 42490
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6109418217617555e-05,
+      "loss": 0.034,
+      "step": 42495
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6108949416342414e-05,
+      "loss": 0.0202,
+      "step": 42500
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6108480615067274e-05,
+      "loss": 0.0828,
+      "step": 42505
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6108011813792134e-05,
+      "loss": 0.0536,
+      "step": 42510
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6107543012516994e-05,
+      "loss": 0.084,
+      "step": 42515
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6107074211241854e-05,
+      "loss": 0.2224,
+      "step": 42520
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6106605409966714e-05,
+      "loss": 0.1419,
+      "step": 42525
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6106136608691577e-05,
+      "loss": 0.3436,
+      "step": 42530
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6105667807416437e-05,
+      "loss": 0.2449,
+      "step": 42535
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6105199006141297e-05,
+      "loss": 0.0223,
+      "step": 42540
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.610473020486616e-05,
+      "loss": 0.0835,
+      "step": 42545
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.610426140359102e-05,
+      "loss": 0.05,
+      "step": 42550
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.610379260231588e-05,
+      "loss": 0.1297,
+      "step": 42555
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.610332380104074e-05,
+      "loss": 0.0907,
+      "step": 42560
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.61028549997656e-05,
+      "loss": 0.1403,
+      "step": 42565
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.610238619849046e-05,
+      "loss": 0.1567,
+      "step": 42570
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.610191739721532e-05,
+      "loss": 0.096,
+      "step": 42575
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.610144859594018e-05,
+      "loss": 0.3812,
+      "step": 42580
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6100979794665044e-05,
+      "loss": 0.2637,
+      "step": 42585
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6100510993389903e-05,
+      "loss": 0.0086,
+      "step": 42590
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6100042192114763e-05,
+      "loss": 0.0589,
+      "step": 42595
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6099573390839623e-05,
+      "loss": 0.1128,
+      "step": 42600
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6099104589564483e-05,
+      "loss": 0.117,
+      "step": 42605
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6098635788289347e-05,
+      "loss": 0.0931,
+      "step": 42610
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6098166987014207e-05,
+      "loss": 0.0807,
+      "step": 42615
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6097698185739066e-05,
+      "loss": 0.2223,
+      "step": 42620
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6097229384463926e-05,
+      "loss": 0.2597,
+      "step": 42625
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.609676058318879e-05,
+      "loss": 0.3632,
+      "step": 42630
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.609629178191365e-05,
+      "loss": 0.1469,
+      "step": 42635
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.609582298063851e-05,
+      "loss": 0.033,
+      "step": 42640
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.609535417936337e-05,
+      "loss": 0.0802,
+      "step": 42645
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.609488537808823e-05,
+      "loss": 0.0277,
+      "step": 42650
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.609441657681309e-05,
+      "loss": 0.1024,
+      "step": 42655
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.609394777553795e-05,
+      "loss": 0.0589,
+      "step": 42660
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.609347897426281e-05,
+      "loss": 0.0769,
+      "step": 42665
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6093010172987673e-05,
+      "loss": 0.1042,
+      "step": 42670
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6092541371712533e-05,
+      "loss": 0.3239,
+      "step": 42675
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6092072570437392e-05,
+      "loss": 0.3277,
+      "step": 42680
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6091603769162252e-05,
+      "loss": 0.3928,
+      "step": 42685
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6091134967887116e-05,
+      "loss": 0.0315,
+      "step": 42690
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6090666166611976e-05,
+      "loss": 0.0425,
+      "step": 42695
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6090197365336836e-05,
+      "loss": 0.0487,
+      "step": 42700
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6089728564061695e-05,
+      "loss": 0.0903,
+      "step": 42705
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6089259762786555e-05,
+      "loss": 0.1076,
+      "step": 42710
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6088790961511415e-05,
+      "loss": 0.169,
+      "step": 42715
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.608832216023628e-05,
+      "loss": 0.1955,
+      "step": 42720
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.608785335896114e-05,
+      "loss": 0.2111,
+      "step": 42725
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6087384557686e-05,
+      "loss": 0.2163,
+      "step": 42730
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.608691575641086e-05,
+      "loss": 0.2275,
+      "step": 42735
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.608644695513572e-05,
+      "loss": 0.0338,
+      "step": 42740
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.608597815386058e-05,
+      "loss": 0.0571,
+      "step": 42745
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.608550935258544e-05,
+      "loss": 0.1106,
+      "step": 42750
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.60850405513103e-05,
+      "loss": 0.1291,
+      "step": 42755
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.608457175003516e-05,
+      "loss": 0.0905,
+      "step": 42760
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.608410294876002e-05,
+      "loss": 0.1497,
+      "step": 42765
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6083634147484885e-05,
+      "loss": 0.1673,
+      "step": 42770
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6083165346209745e-05,
+      "loss": 0.2101,
+      "step": 42775
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6082696544934605e-05,
+      "loss": 0.4179,
+      "step": 42780
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6082227743659465e-05,
+      "loss": 0.274,
+      "step": 42785
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6081758942384325e-05,
+      "loss": 0.0273,
+      "step": 42790
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6081290141109184e-05,
+      "loss": 0.0521,
+      "step": 42795
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6080821339834044e-05,
+      "loss": 0.1252,
+      "step": 42800
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6080352538558904e-05,
+      "loss": 0.0884,
+      "step": 42805
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6079883737283764e-05,
+      "loss": 0.1067,
+      "step": 42810
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6079414936008628e-05,
+      "loss": 0.1663,
+      "step": 42815
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6078946134733488e-05,
+      "loss": 0.0931,
+      "step": 42820
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6078477333458347e-05,
+      "loss": 0.2319,
+      "step": 42825
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.607800853218321e-05,
+      "loss": 0.3311,
+      "step": 42830
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.607753973090807e-05,
+      "loss": 0.2856,
+      "step": 42835
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.607707092963293e-05,
+      "loss": 0.0445,
+      "step": 42840
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.607660212835779e-05,
+      "loss": 0.07,
+      "step": 42845
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.607613332708265e-05,
+      "loss": 0.0974,
+      "step": 42850
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.607566452580751e-05,
+      "loss": 0.151,
+      "step": 42855
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6075195724532374e-05,
+      "loss": 0.1649,
+      "step": 42860
+    },
+    {
+      "epoch": 2.0,
+      "eval_cer": 0.013818165196324908,
+      "eval_loss": 0.07233013957738876,
+      "eval_runtime": 378.9809,
+      "eval_samples_per_second": 50.266,
+      "eval_steps_per_second": 12.568,
+      "eval_wer": 0.11786914335334316,
+      "step": 42862
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6074726923257234e-05,
+      "loss": 0.2466,
+      "step": 42865
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6074258121982094e-05,
+      "loss": 0.0398,
+      "step": 42870
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6073789320706954e-05,
+      "loss": 0.0557,
+      "step": 42875
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6073320519431813e-05,
+      "loss": 0.055,
+      "step": 42880
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6072851718156673e-05,
+      "loss": 0.1247,
+      "step": 42885
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6072382916881533e-05,
+      "loss": 0.0794,
+      "step": 42890
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6071914115606397e-05,
+      "loss": 0.1031,
+      "step": 42895
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6071445314331257e-05,
+      "loss": 0.1369,
+      "step": 42900
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6070976513056117e-05,
+      "loss": 0.1704,
+      "step": 42905
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.607050771178098e-05,
+      "loss": 0.2859,
+      "step": 42910
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.607003891050584e-05,
+      "loss": 0.3003,
+      "step": 42915
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.60695701092307e-05,
+      "loss": 0.0153,
+      "step": 42920
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.606910130795556e-05,
+      "loss": 0.0411,
+      "step": 42925
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.606863250668042e-05,
+      "loss": 0.038,
+      "step": 42930
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.606816370540528e-05,
+      "loss": 0.0782,
+      "step": 42935
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.606769490413014e-05,
+      "loss": 0.1823,
+      "step": 42940
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6067226102855e-05,
+      "loss": 0.0842,
+      "step": 42945
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.606675730157986e-05,
+      "loss": 0.1666,
+      "step": 42950
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6066288500304723e-05,
+      "loss": 0.2304,
+      "step": 42955
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6065819699029583e-05,
+      "loss": 0.1845,
+      "step": 42960
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6065350897754443e-05,
+      "loss": 0.3603,
+      "step": 42965
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6064882096479302e-05,
+      "loss": 0.1044,
+      "step": 42970
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6064413295204166e-05,
+      "loss": 0.0342,
+      "step": 42975
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6063944493929026e-05,
+      "loss": 0.0837,
+      "step": 42980
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6063475692653886e-05,
+      "loss": 0.0801,
+      "step": 42985
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6063006891378746e-05,
+      "loss": 0.1308,
+      "step": 42990
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6062538090103606e-05,
+      "loss": 0.1231,
+      "step": 42995
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.606206928882847e-05,
+      "loss": 0.1921,
+      "step": 43000
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.606160048755333e-05,
+      "loss": 0.2517,
+      "step": 43005
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.606113168627819e-05,
+      "loss": 0.2169,
+      "step": 43010
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.606066288500305e-05,
+      "loss": 0.3231,
+      "step": 43015
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.606019408372791e-05,
+      "loss": 0.049,
+      "step": 43020
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.605972528245277e-05,
+      "loss": 0.0529,
+      "step": 43025
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.605925648117763e-05,
+      "loss": 0.0778,
+      "step": 43030
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.605878767990249e-05,
+      "loss": 0.1165,
+      "step": 43035
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6058318878627352e-05,
+      "loss": 0.0635,
+      "step": 43040
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.605785007735221e-05,
+      "loss": 0.1498,
+      "step": 43045
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.605738127607707e-05,
+      "loss": 0.1469,
+      "step": 43050
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6056912474801935e-05,
+      "loss": 0.1534,
+      "step": 43055
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6056443673526795e-05,
+      "loss": 0.1837,
+      "step": 43060
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6055974872251655e-05,
+      "loss": 0.3012,
+      "step": 43065
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6055506070976515e-05,
+      "loss": 0.0679,
+      "step": 43070
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6055037269701375e-05,
+      "loss": 0.0562,
+      "step": 43075
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6054568468426235e-05,
+      "loss": 0.1606,
+      "step": 43080
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6054099667151094e-05,
+      "loss": 0.0786,
+      "step": 43085
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6053630865875954e-05,
+      "loss": 0.0877,
+      "step": 43090
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6053162064600818e-05,
+      "loss": 0.1015,
+      "step": 43095
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6052693263325678e-05,
+      "loss": 0.1345,
+      "step": 43100
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6052224462050538e-05,
+      "loss": 0.1603,
+      "step": 43105
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6051755660775398e-05,
+      "loss": 0.2164,
+      "step": 43110
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6051286859500257e-05,
+      "loss": 0.2828,
+      "step": 43115
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.605081805822512e-05,
+      "loss": 0.0067,
+      "step": 43120
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.605034925694998e-05,
+      "loss": 0.0261,
+      "step": 43125
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.604988045567484e-05,
+      "loss": 0.0532,
+      "step": 43130
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.60494116543997e-05,
+      "loss": 0.1057,
+      "step": 43135
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6048942853124564e-05,
+      "loss": 0.119,
+      "step": 43140
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6048474051849424e-05,
+      "loss": 0.0828,
+      "step": 43145
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6048005250574284e-05,
+      "loss": 0.1604,
+      "step": 43150
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6047536449299144e-05,
+      "loss": 0.1541,
+      "step": 43155
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6047067648024004e-05,
+      "loss": 0.3743,
+      "step": 43160
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6046598846748864e-05,
+      "loss": 0.3275,
+      "step": 43165
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6046130045473724e-05,
+      "loss": 0.0713,
+      "step": 43170
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6045661244198583e-05,
+      "loss": 0.1079,
+      "step": 43175
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6045192442923447e-05,
+      "loss": 0.0531,
+      "step": 43180
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6044723641648307e-05,
+      "loss": 0.0501,
+      "step": 43185
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6044254840373167e-05,
+      "loss": 0.1146,
+      "step": 43190
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6043786039098027e-05,
+      "loss": 0.1149,
+      "step": 43195
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.604331723782289e-05,
+      "loss": 0.2056,
+      "step": 43200
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.604284843654775e-05,
+      "loss": 0.1313,
+      "step": 43205
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.604237963527261e-05,
+      "loss": 0.2624,
+      "step": 43210
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.604191083399747e-05,
+      "loss": 0.3201,
+      "step": 43215
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.604144203272233e-05,
+      "loss": 0.0401,
+      "step": 43220
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.604097323144719e-05,
+      "loss": 0.1277,
+      "step": 43225
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.604050443017205e-05,
+      "loss": 0.0803,
+      "step": 43230
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6040035628896913e-05,
+      "loss": 0.092,
+      "step": 43235
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6039566827621773e-05,
+      "loss": 0.1218,
+      "step": 43240
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6039098026346633e-05,
+      "loss": 0.0504,
+      "step": 43245
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6038629225071493e-05,
+      "loss": 0.1807,
+      "step": 43250
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6038160423796353e-05,
+      "loss": 0.1809,
+      "step": 43255
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6037691622521216e-05,
+      "loss": 0.3912,
+      "step": 43260
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6037222821246076e-05,
+      "loss": 0.2558,
+      "step": 43265
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6036754019970936e-05,
+      "loss": 0.096,
+      "step": 43270
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6036285218695796e-05,
+      "loss": 0.0349,
+      "step": 43275
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.603581641742066e-05,
+      "loss": 0.0889,
+      "step": 43280
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.603534761614552e-05,
+      "loss": 0.0881,
+      "step": 43285
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.603487881487038e-05,
+      "loss": 0.1415,
+      "step": 43290
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.603441001359524e-05,
+      "loss": 0.1533,
+      "step": 43295
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.60339412123201e-05,
+      "loss": 0.2526,
+      "step": 43300
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.603347241104496e-05,
+      "loss": 0.2226,
+      "step": 43305
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.603300360976982e-05,
+      "loss": 0.2101,
+      "step": 43310
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.603253480849468e-05,
+      "loss": 0.291,
+      "step": 43315
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.603206600721954e-05,
+      "loss": 0.0347,
+      "step": 43320
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6031597205944402e-05,
+      "loss": 0.0621,
+      "step": 43325
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6031128404669262e-05,
+      "loss": 0.0558,
+      "step": 43330
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.603065960339412e-05,
+      "loss": 0.0694,
+      "step": 43335
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6030190802118985e-05,
+      "loss": 0.1065,
+      "step": 43340
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6029722000843845e-05,
+      "loss": 0.0916,
+      "step": 43345
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6029253199568705e-05,
+      "loss": 0.1291,
+      "step": 43350
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6028784398293565e-05,
+      "loss": 0.1963,
+      "step": 43355
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6028315597018425e-05,
+      "loss": 0.2028,
+      "step": 43360
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6027846795743285e-05,
+      "loss": 0.2797,
+      "step": 43365
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6027377994468148e-05,
+      "loss": 0.082,
+      "step": 43370
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6026909193193008e-05,
+      "loss": 0.0206,
+      "step": 43375
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6026440391917868e-05,
+      "loss": 0.0824,
+      "step": 43380
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6025971590642728e-05,
+      "loss": 0.0229,
+      "step": 43385
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6025502789367588e-05,
+      "loss": 0.0882,
+      "step": 43390
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6025033988092448e-05,
+      "loss": 0.093,
+      "step": 43395
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6024565186817308e-05,
+      "loss": 0.162,
+      "step": 43400
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.602409638554217e-05,
+      "loss": 0.175,
+      "step": 43405
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.602362758426703e-05,
+      "loss": 0.1903,
+      "step": 43410
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.602315878299189e-05,
+      "loss": 0.2453,
+      "step": 43415
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6022689981716754e-05,
+      "loss": 0.0193,
+      "step": 43420
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6022221180441614e-05,
+      "loss": 0.0242,
+      "step": 43425
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6021752379166474e-05,
+      "loss": 0.0957,
+      "step": 43430
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6021283577891334e-05,
+      "loss": 0.0662,
+      "step": 43435
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6020814776616194e-05,
+      "loss": 0.1751,
+      "step": 43440
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6020345975341054e-05,
+      "loss": 0.0665,
+      "step": 43445
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6019877174065914e-05,
+      "loss": 0.1302,
+      "step": 43450
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6019408372790774e-05,
+      "loss": 0.1834,
+      "step": 43455
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6018939571515634e-05,
+      "loss": 0.2339,
+      "step": 43460
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6018470770240497e-05,
+      "loss": 0.2381,
+      "step": 43465
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6018001968965357e-05,
+      "loss": 0.055,
+      "step": 43470
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6017533167690217e-05,
+      "loss": 0.0649,
+      "step": 43475
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6017064366415077e-05,
+      "loss": 0.0983,
+      "step": 43480
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.601659556513994e-05,
+      "loss": 0.0965,
+      "step": 43485
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.60161267638648e-05,
+      "loss": 0.133,
+      "step": 43490
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.601565796258966e-05,
+      "loss": 0.0916,
+      "step": 43495
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.601518916131452e-05,
+      "loss": 0.1319,
+      "step": 43500
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.601472036003938e-05,
+      "loss": 0.2172,
+      "step": 43505
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6014251558764243e-05,
+      "loss": 0.3607,
+      "step": 43510
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6013782757489103e-05,
+      "loss": 0.4852,
+      "step": 43515
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6013313956213963e-05,
+      "loss": 0.0733,
+      "step": 43520
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6012845154938823e-05,
+      "loss": 0.0488,
+      "step": 43525
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6012376353663683e-05,
+      "loss": 0.0537,
+      "step": 43530
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6011907552388543e-05,
+      "loss": 0.0909,
+      "step": 43535
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6011438751113403e-05,
+      "loss": 0.0844,
+      "step": 43540
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6010969949838263e-05,
+      "loss": 0.12,
+      "step": 43545
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6010501148563126e-05,
+      "loss": 0.1497,
+      "step": 43550
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6010032347287986e-05,
+      "loss": 0.231,
+      "step": 43555
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6009563546012846e-05,
+      "loss": 0.2282,
+      "step": 43560
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.600909474473771e-05,
+      "loss": 0.3212,
+      "step": 43565
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.600862594346257e-05,
+      "loss": 0.0352,
+      "step": 43570
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.600815714218743e-05,
+      "loss": 0.0511,
+      "step": 43575
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.600768834091229e-05,
+      "loss": 0.1067,
+      "step": 43580
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.600721953963715e-05,
+      "loss": 0.0649,
+      "step": 43585
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.600675073836201e-05,
+      "loss": 0.0548,
+      "step": 43590
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.600628193708687e-05,
+      "loss": 0.0562,
+      "step": 43595
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.600581313581173e-05,
+      "loss": 0.1239,
+      "step": 43600
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6005344334536592e-05,
+      "loss": 0.195,
+      "step": 43605
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6004875533261452e-05,
+      "loss": 0.3106,
+      "step": 43610
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6004406731986312e-05,
+      "loss": 0.257,
+      "step": 43615
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6003937930711172e-05,
+      "loss": 0.0264,
+      "step": 43620
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6003469129436032e-05,
+      "loss": 0.0908,
+      "step": 43625
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6003000328160895e-05,
+      "loss": 0.0849,
+      "step": 43630
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6002531526885755e-05,
+      "loss": 0.0908,
+      "step": 43635
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6002062725610615e-05,
+      "loss": 0.0562,
+      "step": 43640
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6001593924335475e-05,
+      "loss": 0.0963,
+      "step": 43645
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6001125123060338e-05,
+      "loss": 0.1417,
+      "step": 43650
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6000656321785198e-05,
+      "loss": 0.1321,
+      "step": 43655
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6000187520510058e-05,
+      "loss": 0.2516,
+      "step": 43660
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.5999718719234918e-05,
+      "loss": 0.2598,
+      "step": 43665
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.5999249917959778e-05,
+      "loss": 0.0211,
+      "step": 43670
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.5998781116684638e-05,
+      "loss": 0.1165,
+      "step": 43675
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.5998312315409498e-05,
+      "loss": 0.0921,
+      "step": 43680
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.5997843514134358e-05,
+      "loss": 0.087,
+      "step": 43685
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.599737471285922e-05,
+      "loss": 0.0793,
+      "step": 43690
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.599690591158408e-05,
+      "loss": 0.097,
+      "step": 43695
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.599643711030894e-05,
+      "loss": 0.1542,
+      "step": 43700
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.59959683090338e-05,
+      "loss": 0.2022,
+      "step": 43705
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.5995499507758664e-05,
+      "loss": 0.292,
+      "step": 43710
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.5995030706483524e-05,
+      "loss": 0.3516,
+      "step": 43715
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.5994561905208384e-05,
+      "loss": 0.0639,
+      "step": 43720
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.5994093103933244e-05,
+      "loss": 0.0484,
+      "step": 43725
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.5993624302658104e-05,
+      "loss": 0.0859,
+      "step": 43730
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.5993155501382964e-05,
+      "loss": 0.0987,
+      "step": 43735
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.5992686700107824e-05,
+      "loss": 0.0731,
+      "step": 43740
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.5992217898832687e-05,
+      "loss": 0.1355,
+      "step": 43745
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.5991749097557547e-05,
+      "loss": 0.1258,
+      "step": 43750
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.5991280296282407e-05,
+      "loss": 0.1247,
+      "step": 43755
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.5990811495007267e-05,
+      "loss": 0.2308,
+      "step": 43760
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.5990342693732127e-05,
+      "loss": 0.3262,
+      "step": 43765
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.598987389245699e-05,
+      "loss": 0.0529,
+      "step": 43770
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.598940509118185e-05,
+      "loss": 0.0394,
+      "step": 43775
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.598893628990671e-05,
+      "loss": 0.0503,
+      "step": 43780
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.598846748863157e-05,
+      "loss": 0.0712,
+      "step": 43785
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.5987998687356433e-05,
+      "loss": 0.0774,
+      "step": 43790
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.5987529886081293e-05,
+      "loss": 0.1446,
+      "step": 43795
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.5987061084806153e-05,
+      "loss": 0.1228,
+      "step": 43800
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.5986592283531013e-05,
+      "loss": 0.2051,
+      "step": 43805
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.5986123482255873e-05,
+      "loss": 0.2713,
+      "step": 43810
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.5985654680980733e-05,
+      "loss": 0.298,
+      "step": 43815
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.5985185879705593e-05,
+      "loss": 0.0534,
+      "step": 43820
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.5984717078430453e-05,
+      "loss": 0.039,
+      "step": 43825
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5984248277155313e-05,
+      "loss": 0.09,
+      "step": 43830
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5983779475880176e-05,
+      "loss": 0.1248,
+      "step": 43835
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5983310674605036e-05,
+      "loss": 0.1089,
+      "step": 43840
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5982841873329896e-05,
+      "loss": 0.1591,
+      "step": 43845
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.598237307205476e-05,
+      "loss": 0.0722,
+      "step": 43850
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.598190427077962e-05,
+      "loss": 0.2335,
+      "step": 43855
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.598143546950448e-05,
+      "loss": 0.2227,
+      "step": 43860
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.598096666822934e-05,
+      "loss": 0.4275,
+      "step": 43865
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.59804978669542e-05,
+      "loss": 0.0902,
+      "step": 43870
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.598002906567906e-05,
+      "loss": 0.0339,
+      "step": 43875
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.597956026440392e-05,
+      "loss": 0.0849,
+      "step": 43880
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5979091463128782e-05,
+      "loss": 0.0877,
+      "step": 43885
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5978622661853642e-05,
+      "loss": 0.0713,
+      "step": 43890
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5978153860578502e-05,
+      "loss": 0.0822,
+      "step": 43895
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5977685059303362e-05,
+      "loss": 0.1345,
+      "step": 43900
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5977216258028222e-05,
+      "loss": 0.1678,
+      "step": 43905
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5976747456753082e-05,
+      "loss": 0.2043,
+      "step": 43910
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5976278655477945e-05,
+      "loss": 0.2602,
+      "step": 43915
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5975809854202805e-05,
+      "loss": 0.0879,
+      "step": 43920
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5975341052927665e-05,
+      "loss": 0.0494,
+      "step": 43925
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.597487225165253e-05,
+      "loss": 0.0878,
+      "step": 43930
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5974403450377388e-05,
+      "loss": 0.0827,
+      "step": 43935
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5973934649102248e-05,
+      "loss": 0.1332,
+      "step": 43940
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5973465847827108e-05,
+      "loss": 0.1587,
+      "step": 43945
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5972997046551968e-05,
+      "loss": 0.1737,
+      "step": 43950
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5972528245276828e-05,
+      "loss": 0.2788,
+      "step": 43955
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5972059444001688e-05,
+      "loss": 0.3036,
+      "step": 43960
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5971590642726548e-05,
+      "loss": 0.2977,
+      "step": 43965
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5971121841451408e-05,
+      "loss": 0.0365,
+      "step": 43970
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.597065304017627e-05,
+      "loss": 0.0257,
+      "step": 43975
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.597018423890113e-05,
+      "loss": 0.0948,
+      "step": 43980
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.596971543762599e-05,
+      "loss": 0.0868,
+      "step": 43985
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.596924663635085e-05,
+      "loss": 0.0878,
+      "step": 43990
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5968777835075714e-05,
+      "loss": 0.0816,
+      "step": 43995
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5968309033800574e-05,
+      "loss": 0.0745,
+      "step": 44000
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5967840232525434e-05,
+      "loss": 0.266,
+      "step": 44005
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5967371431250294e-05,
+      "loss": 0.178,
+      "step": 44010
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5966902629975154e-05,
+      "loss": 0.2893,
+      "step": 44015
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5966433828700017e-05,
+      "loss": 0.0461,
+      "step": 44020
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5965965027424877e-05,
+      "loss": 0.0394,
+      "step": 44025
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5965496226149737e-05,
+      "loss": 0.0646,
+      "step": 44030
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5965027424874597e-05,
+      "loss": 0.0966,
+      "step": 44035
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5964558623599457e-05,
+      "loss": 0.1131,
+      "step": 44040
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5964089822324317e-05,
+      "loss": 0.1476,
+      "step": 44045
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5963621021049177e-05,
+      "loss": 0.1769,
+      "step": 44050
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5963152219774037e-05,
+      "loss": 0.2321,
+      "step": 44055
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.59626834184989e-05,
+      "loss": 0.2395,
+      "step": 44060
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.596221461722376e-05,
+      "loss": 0.316,
+      "step": 44065
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.596174581594862e-05,
+      "loss": 0.006,
+      "step": 44070
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5961277014673483e-05,
+      "loss": 0.0283,
+      "step": 44075
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5960808213398343e-05,
+      "loss": 0.0903,
+      "step": 44080
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5960339412123203e-05,
+      "loss": 0.047,
+      "step": 44085
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5959870610848063e-05,
+      "loss": 0.1183,
+      "step": 44090
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5959401809572923e-05,
+      "loss": 0.1107,
+      "step": 44095
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5958933008297783e-05,
+      "loss": 0.1509,
+      "step": 44100
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5958464207022643e-05,
+      "loss": 0.2046,
+      "step": 44105
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5957995405747503e-05,
+      "loss": 0.3222,
+      "step": 44110
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5957526604472366e-05,
+      "loss": 0.3044,
+      "step": 44115
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5957057803197226e-05,
+      "loss": 0.0688,
+      "step": 44120
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5956589001922086e-05,
+      "loss": 0.0558,
+      "step": 44125
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5956120200646946e-05,
+      "loss": 0.073,
+      "step": 44130
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.595565139937181e-05,
+      "loss": 0.0916,
+      "step": 44135
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.595518259809667e-05,
+      "loss": 0.1182,
+      "step": 44140
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.595471379682153e-05,
+      "loss": 0.0756,
+      "step": 44145
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.595424499554639e-05,
+      "loss": 0.1211,
+      "step": 44150
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.595377619427125e-05,
+      "loss": 0.1624,
+      "step": 44155
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5953307392996112e-05,
+      "loss": 0.2841,
+      "step": 44160
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5952838591720972e-05,
+      "loss": 0.3485,
+      "step": 44165
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5952369790445832e-05,
+      "loss": 0.092,
+      "step": 44170
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5951900989170692e-05,
+      "loss": 0.0326,
+      "step": 44175
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5951432187895552e-05,
+      "loss": 0.0569,
+      "step": 44180
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5950963386620412e-05,
+      "loss": 0.1249,
+      "step": 44185
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5950494585345272e-05,
+      "loss": 0.0785,
+      "step": 44190
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5950025784070132e-05,
+      "loss": 0.0755,
+      "step": 44195
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5949556982794995e-05,
+      "loss": 0.2121,
+      "step": 44200
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5949088181519855e-05,
+      "loss": 0.2509,
+      "step": 44205
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5948619380244715e-05,
+      "loss": 0.2829,
+      "step": 44210
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.594815057896958e-05,
+      "loss": 0.3905,
+      "step": 44215
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.594768177769444e-05,
+      "loss": 0.0941,
+      "step": 44220
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5947212976419298e-05,
+      "loss": 0.0883,
+      "step": 44225
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5946744175144158e-05,
+      "loss": 0.0664,
+      "step": 44230
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5946275373869018e-05,
+      "loss": 0.048,
+      "step": 44235
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5945806572593878e-05,
+      "loss": 0.1531,
+      "step": 44240
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5945337771318738e-05,
+      "loss": 0.0949,
+      "step": 44245
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5944868970043598e-05,
+      "loss": 0.1509,
+      "step": 44250
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.594440016876846e-05,
+      "loss": 0.221,
+      "step": 44255
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.594393136749332e-05,
+      "loss": 0.2352,
+      "step": 44260
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.594346256621818e-05,
+      "loss": 0.1831,
+      "step": 44265
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.594299376494304e-05,
+      "loss": 0.0654,
+      "step": 44270
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.59425249636679e-05,
+      "loss": 0.0374,
+      "step": 44275
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5942056162392764e-05,
+      "loss": 0.0869,
+      "step": 44280
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5941587361117624e-05,
+      "loss": 0.1089,
+      "step": 44285
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5941118559842484e-05,
+      "loss": 0.0679,
+      "step": 44290
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5940649758567344e-05,
+      "loss": 0.1117,
+      "step": 44295
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5940180957292207e-05,
+      "loss": 0.1786,
+      "step": 44300
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5939712156017067e-05,
+      "loss": 0.1472,
+      "step": 44305
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5939243354741927e-05,
+      "loss": 0.2769,
+      "step": 44310
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5938774553466787e-05,
+      "loss": 0.1834,
+      "step": 44315
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5938305752191647e-05,
+      "loss": 0.0698,
+      "step": 44320
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5937836950916507e-05,
+      "loss": 0.0685,
+      "step": 44325
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5937368149641367e-05,
+      "loss": 0.0576,
+      "step": 44330
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5936899348366227e-05,
+      "loss": 0.1033,
+      "step": 44335
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5936430547091087e-05,
+      "loss": 0.0882,
+      "step": 44340
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.593596174581595e-05,
+      "loss": 0.1051,
+      "step": 44345
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.593549294454081e-05,
+      "loss": 0.137,
+      "step": 44350
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.593502414326567e-05,
+      "loss": 0.1292,
+      "step": 44355
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5934555341990533e-05,
+      "loss": 0.3229,
+      "step": 44360
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5934086540715393e-05,
+      "loss": 0.4082,
+      "step": 44365
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5933617739440253e-05,
+      "loss": 0.0691,
+      "step": 44370
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5933148938165113e-05,
+      "loss": 0.0598,
+      "step": 44375
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5932680136889973e-05,
+      "loss": 0.0883,
+      "step": 44380
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5932211335614833e-05,
+      "loss": 0.0679,
+      "step": 44385
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5931742534339693e-05,
+      "loss": 0.0568,
+      "step": 44390
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5931273733064556e-05,
+      "loss": 0.1136,
+      "step": 44395
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5930804931789416e-05,
+      "loss": 0.1673,
+      "step": 44400
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5930336130514276e-05,
+      "loss": 0.151,
+      "step": 44405
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5929867329239136e-05,
+      "loss": 0.2195,
+      "step": 44410
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5929398527963996e-05,
+      "loss": 0.3555,
+      "step": 44415
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5928929726688856e-05,
+      "loss": 0.0534,
+      "step": 44420
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.592846092541372e-05,
+      "loss": 0.0423,
+      "step": 44425
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.592799212413858e-05,
+      "loss": 0.0974,
+      "step": 44430
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.592752332286344e-05,
+      "loss": 0.1153,
+      "step": 44435
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5927054521588303e-05,
+      "loss": 0.1223,
+      "step": 44440
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5926585720313162e-05,
+      "loss": 0.0456,
+      "step": 44445
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5926116919038022e-05,
+      "loss": 0.24,
+      "step": 44450
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5925648117762882e-05,
+      "loss": 0.2007,
+      "step": 44455
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5925179316487742e-05,
+      "loss": 0.3084,
+      "step": 44460
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5924710515212602e-05,
+      "loss": 0.2493,
+      "step": 44465
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5924241713937462e-05,
+      "loss": 0.0436,
+      "step": 44470
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5923772912662322e-05,
+      "loss": 0.0289,
+      "step": 44475
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5923304111387182e-05,
+      "loss": 0.0585,
+      "step": 44480
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5922835310112045e-05,
+      "loss": 0.0515,
+      "step": 44485
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5922366508836905e-05,
+      "loss": 0.1049,
+      "step": 44490
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5921897707561765e-05,
+      "loss": 0.1265,
+      "step": 44495
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5921428906286625e-05,
+      "loss": 0.0542,
+      "step": 44500
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.592096010501149e-05,
+      "loss": 0.3128,
+      "step": 44505
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.592049130373635e-05,
+      "loss": 0.2474,
+      "step": 44510
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.592002250246121e-05,
+      "loss": 0.3596,
+      "step": 44515
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5919553701186068e-05,
+      "loss": 0.1306,
+      "step": 44520
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5919084899910928e-05,
+      "loss": 0.0428,
+      "step": 44525
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5918616098635788e-05,
+      "loss": 0.0753,
+      "step": 44530
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.591814729736065e-05,
+      "loss": 0.12,
+      "step": 44535
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.591767849608551e-05,
+      "loss": 0.1302,
+      "step": 44540
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.591720969481037e-05,
+      "loss": 0.1034,
+      "step": 44545
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.591674089353523e-05,
+      "loss": 0.1406,
+      "step": 44550
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.591627209226009e-05,
+      "loss": 0.1504,
+      "step": 44555
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.591580329098495e-05,
+      "loss": 0.3588,
+      "step": 44560
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5915334489709814e-05,
+      "loss": 0.3762,
+      "step": 44565
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5914865688434674e-05,
+      "loss": 0.0344,
+      "step": 44570
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5914396887159534e-05,
+      "loss": 0.0487,
+      "step": 44575
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5913928085884394e-05,
+      "loss": 0.0769,
+      "step": 44580
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5913459284609258e-05,
+      "loss": 0.0923,
+      "step": 44585
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5912990483334117e-05,
+      "loss": 0.1312,
+      "step": 44590
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5912521682058977e-05,
+      "loss": 0.0953,
+      "step": 44595
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5912052880783837e-05,
+      "loss": 0.2189,
+      "step": 44600
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5911584079508697e-05,
+      "loss": 0.3004,
+      "step": 44605
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5911115278233557e-05,
+      "loss": 0.2755,
+      "step": 44610
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5910646476958417e-05,
+      "loss": 0.3,
+      "step": 44615
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5910177675683277e-05,
+      "loss": 0.0662,
+      "step": 44620
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.590970887440814e-05,
+      "loss": 0.0399,
+      "step": 44625
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5909240073133e-05,
+      "loss": 0.0311,
+      "step": 44630
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.590877127185786e-05,
+      "loss": 0.0786,
+      "step": 44635
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.590830247058272e-05,
+      "loss": 0.1089,
+      "step": 44640
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5907833669307584e-05,
+      "loss": 0.1173,
+      "step": 44645
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5907364868032443e-05,
+      "loss": 0.1884,
+      "step": 44650
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5906896066757303e-05,
+      "loss": 0.1855,
+      "step": 44655
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5906427265482163e-05,
+      "loss": 0.1961,
+      "step": 44660
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5905958464207023e-05,
+      "loss": 0.3027,
+      "step": 44665
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5905489662931887e-05,
+      "loss": 0.0615,
+      "step": 44670
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5905020861656747e-05,
+      "loss": 0.0229,
+      "step": 44675
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5904552060381606e-05,
+      "loss": 0.0722,
+      "step": 44680
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5904083259106466e-05,
+      "loss": 0.0763,
+      "step": 44685
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5903614457831326e-05,
+      "loss": 0.0572,
+      "step": 44690
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5903145656556186e-05,
+      "loss": 0.1163,
+      "step": 44695
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5902676855281046e-05,
+      "loss": 0.1347,
+      "step": 44700
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5902208054005906e-05,
+      "loss": 0.1086,
+      "step": 44705
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.590173925273077e-05,
+      "loss": 0.2285,
+      "step": 44710
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.590127045145563e-05,
+      "loss": 0.3338,
+      "step": 44715
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.590080165018049e-05,
+      "loss": 0.0527,
+      "step": 44720
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5900332848905353e-05,
+      "loss": 0.0404,
+      "step": 44725
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5899864047630213e-05,
+      "loss": 0.031,
+      "step": 44730
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5899395246355072e-05,
+      "loss": 0.0831,
+      "step": 44735
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5898926445079932e-05,
+      "loss": 0.0364,
+      "step": 44740
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5898457643804792e-05,
+      "loss": 0.1629,
+      "step": 44745
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5897988842529652e-05,
+      "loss": 0.166,
+      "step": 44750
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5897520041254512e-05,
+      "loss": 0.2196,
+      "step": 44755
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5897051239979372e-05,
+      "loss": 0.2828,
+      "step": 44760
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5896582438704235e-05,
+      "loss": 0.2774,
+      "step": 44765
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5896113637429095e-05,
+      "loss": 0.0914,
+      "step": 44770
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5895644836153955e-05,
+      "loss": 0.0176,
+      "step": 44775
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5895176034878815e-05,
+      "loss": 0.1039,
+      "step": 44780
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5894707233603675e-05,
+      "loss": 0.1057,
+      "step": 44785
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.589423843232854e-05,
+      "loss": 0.0865,
+      "step": 44790
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.58937696310534e-05,
+      "loss": 0.1795,
+      "step": 44795
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.589330082977826e-05,
+      "loss": 0.2326,
+      "step": 44800
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.589283202850312e-05,
+      "loss": 0.1466,
+      "step": 44805
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.589236322722798e-05,
+      "loss": 0.2534,
+      "step": 44810
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.589189442595284e-05,
+      "loss": 0.2671,
+      "step": 44815
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.58914256246777e-05,
+      "loss": 0.0678,
+      "step": 44820
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.589095682340256e-05,
+      "loss": 0.043,
+      "step": 44825
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.589048802212742e-05,
+      "loss": 0.0897,
+      "step": 44830
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.589001922085228e-05,
+      "loss": 0.0819,
+      "step": 44835
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.588955041957714e-05,
+      "loss": 0.0692,
+      "step": 44840
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5889081618302e-05,
+      "loss": 0.1084,
+      "step": 44845
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.588861281702686e-05,
+      "loss": 0.1194,
+      "step": 44850
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5888144015751724e-05,
+      "loss": 0.1409,
+      "step": 44855
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5887675214476584e-05,
+      "loss": 0.3668,
+      "step": 44860
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5887206413201444e-05,
+      "loss": 0.2747,
+      "step": 44865
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5886737611926308e-05,
+      "loss": 0.0384,
+      "step": 44870
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5886268810651168e-05,
+      "loss": 0.0351,
+      "step": 44875
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5885800009376028e-05,
+      "loss": 0.1084,
+      "step": 44880
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5885331208100887e-05,
+      "loss": 0.0684,
+      "step": 44885
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5884862406825747e-05,
+      "loss": 0.0758,
+      "step": 44890
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5884393605550607e-05,
+      "loss": 0.1907,
+      "step": 44895
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5883924804275467e-05,
+      "loss": 0.1878,
+      "step": 44900
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.588345600300033e-05,
+      "loss": 0.1311,
+      "step": 44905
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.588298720172519e-05,
+      "loss": 0.2296,
+      "step": 44910
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.588251840045005e-05,
+      "loss": 0.2933,
+      "step": 44915
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.588204959917491e-05,
+      "loss": 0.0965,
+      "step": 44920
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.588158079789977e-05,
+      "loss": 0.0439,
+      "step": 44925
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.588111199662463e-05,
+      "loss": 0.0551,
+      "step": 44930
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5880643195349494e-05,
+      "loss": 0.1475,
+      "step": 44935
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5880174394074353e-05,
+      "loss": 0.0817,
+      "step": 44940
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5879705592799213e-05,
+      "loss": 0.1122,
+      "step": 44945
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5879236791524077e-05,
+      "loss": 0.2288,
+      "step": 44950
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5878767990248937e-05,
+      "loss": 0.184,
+      "step": 44955
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5878299188973797e-05,
+      "loss": 0.2115,
+      "step": 44960
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5877830387698657e-05,
+      "loss": 0.3778,
+      "step": 44965
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5877361586423516e-05,
+      "loss": 0.0626,
+      "step": 44970
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5876892785148376e-05,
+      "loss": 0.0159,
+      "step": 44975
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5876423983873236e-05,
+      "loss": 0.0709,
+      "step": 44980
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5875955182598096e-05,
+      "loss": 0.1002,
+      "step": 44985
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5875486381322956e-05,
+      "loss": 0.0633,
+      "step": 44990
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.587501758004782e-05,
+      "loss": 0.1328,
+      "step": 44995
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.587454877877268e-05,
+      "loss": 0.1123,
+      "step": 45000
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.587407997749754e-05,
+      "loss": 0.258,
+      "step": 45005
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.58736111762224e-05,
+      "loss": 0.167,
+      "step": 45010
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5873142374947263e-05,
+      "loss": 0.263,
+      "step": 45015
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5872673573672123e-05,
+      "loss": 0.0767,
+      "step": 45020
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5872204772396983e-05,
+      "loss": 0.107,
+      "step": 45025
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5871735971121842e-05,
+      "loss": 0.0827,
+      "step": 45030
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5871267169846702e-05,
+      "loss": 0.129,
+      "step": 45035
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5870798368571562e-05,
+      "loss": 0.0815,
+      "step": 45040
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5870329567296426e-05,
+      "loss": 0.1138,
+      "step": 45045
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5869860766021286e-05,
+      "loss": 0.0898,
+      "step": 45050
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5869391964746146e-05,
+      "loss": 0.2097,
+      "step": 45055
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5868923163471005e-05,
+      "loss": 0.3592,
+      "step": 45060
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5868454362195865e-05,
+      "loss": 0.3279,
+      "step": 45065
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5867985560920725e-05,
+      "loss": 0.0615,
+      "step": 45070
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.586751675964559e-05,
+      "loss": 0.0598,
+      "step": 45075
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.586704795837045e-05,
+      "loss": 0.0647,
+      "step": 45080
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.586657915709531e-05,
+      "loss": 0.0312,
+      "step": 45085
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.586611035582017e-05,
+      "loss": 0.0893,
+      "step": 45090
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5865641554545032e-05,
+      "loss": 0.0462,
+      "step": 45095
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5865172753269892e-05,
+      "loss": 0.1772,
+      "step": 45100
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.586470395199475e-05,
+      "loss": 0.2507,
+      "step": 45105
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.586423515071961e-05,
+      "loss": 0.2887,
+      "step": 45110
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.586376634944447e-05,
+      "loss": 0.2688,
+      "step": 45115
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.586329754816933e-05,
+      "loss": 0.0636,
+      "step": 45120
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.586282874689419e-05,
+      "loss": 0.051,
+      "step": 45125
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.586235994561905e-05,
+      "loss": 0.0607,
+      "step": 45130
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.5861891144343915e-05,
+      "loss": 0.0952,
+      "step": 45135
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.5861422343068775e-05,
+      "loss": 0.0968,
+      "step": 45140
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.5860953541793634e-05,
+      "loss": 0.0867,
+      "step": 45145
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.5860484740518494e-05,
+      "loss": 0.0972,
+      "step": 45150
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.5860015939243358e-05,
+      "loss": 0.2867,
+      "step": 45155
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.5859547137968218e-05,
+      "loss": 0.1727,
+      "step": 45160
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.5859078336693078e-05,
+      "loss": 0.4698,
+      "step": 45165
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.5858609535417938e-05,
+      "loss": 0.0394,
+      "step": 45170
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.5858140734142797e-05,
+      "loss": 0.0625,
+      "step": 45175
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.5857671932867657e-05,
+      "loss": 0.0684,
+      "step": 45180
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.585720313159252e-05,
+      "loss": 0.069,
+      "step": 45185
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.585673433031738e-05,
+      "loss": 0.0847,
+      "step": 45190
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.585626552904224e-05,
+      "loss": 0.1978,
+      "step": 45195
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.58557967277671e-05,
+      "loss": 0.1196,
+      "step": 45200
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.585532792649196e-05,
+      "loss": 0.2047,
+      "step": 45205
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.585485912521682e-05,
+      "loss": 0.2853,
+      "step": 45210
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.585439032394168e-05,
+      "loss": 0.2437,
+      "step": 45215
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.5853921522666544e-05,
+      "loss": 0.087,
+      "step": 45220
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.5853452721391404e-05,
+      "loss": 0.0486,
+      "step": 45225
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.5852983920116264e-05,
+      "loss": 0.1025,
+      "step": 45230
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.5852515118841127e-05,
+      "loss": 0.0609,
+      "step": 45235
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.5852046317565987e-05,
+      "loss": 0.11,
+      "step": 45240
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.5851577516290847e-05,
+      "loss": 0.1482,
+      "step": 45245
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.5851108715015707e-05,
+      "loss": 0.1591,
+      "step": 45250
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.5850639913740567e-05,
+      "loss": 0.2293,
+      "step": 45255
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.5850171112465427e-05,
+      "loss": 0.2316,
+      "step": 45260
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.5849702311190286e-05,
+      "loss": 0.3227,
+      "step": 45265
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.5849233509915146e-05,
+      "loss": 0.0498,
+      "step": 45270
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.584876470864001e-05,
+      "loss": 0.0601,
+      "step": 45275
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.584829590736487e-05,
+      "loss": 0.0298,
+      "step": 45280
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.584782710608973e-05,
+      "loss": 0.0737,
+      "step": 45285
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.584735830481459e-05,
+      "loss": 0.0653,
+      "step": 45290
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.584688950353945e-05,
+      "loss": 0.1548,
+      "step": 45295
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.5846420702264313e-05,
+      "loss": 0.1249,
+      "step": 45300
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.5845951900989173e-05,
+      "loss": 0.1399,
+      "step": 45305
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.5845483099714033e-05,
+      "loss": 0.2704,
+      "step": 45310
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.5845014298438893e-05,
+      "loss": 0.2491,
+      "step": 45315
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.5844545497163756e-05,
+      "loss": 0.0577,
+      "step": 45320
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.5844076695888616e-05,
+      "loss": 0.0291,
+      "step": 45325
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.5843607894613476e-05,
+      "loss": 0.0801,
+      "step": 45330
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.5843139093338336e-05,
+      "loss": 0.062,
+      "step": 45335
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.5842670292063196e-05,
+      "loss": 0.0672,
+      "step": 45340
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.5842201490788056e-05,
+      "loss": 0.1409,
+      "step": 45345
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.5841732689512915e-05,
+      "loss": 0.1184,
+      "step": 45350
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.5841263888237775e-05,
+      "loss": 0.2517,
+      "step": 45355
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.5840795086962635e-05,
+      "loss": 0.1505,
+      "step": 45360
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.58403262856875e-05,
+      "loss": 0.318,
+      "step": 45365
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.583985748441236e-05,
+      "loss": 0.0626,
+      "step": 45370
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.583938868313722e-05,
+      "loss": 0.0521,
+      "step": 45375
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.5838919881862082e-05,
+      "loss": 0.1245,
+      "step": 45380
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.5838451080586942e-05,
+      "loss": 0.0721,
+      "step": 45385
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.5837982279311802e-05,
+      "loss": 0.1457,
+      "step": 45390
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.583751347803666e-05,
+      "loss": 0.132,
+      "step": 45395
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.583704467676152e-05,
+      "loss": 0.1566,
+      "step": 45400
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.583657587548638e-05,
+      "loss": 0.2234,
+      "step": 45405
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.583610707421124e-05,
+      "loss": 0.1806,
+      "step": 45410
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.5835638272936105e-05,
+      "loss": 0.384,
+      "step": 45415
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.5835169471660965e-05,
+      "loss": 0.0281,
+      "step": 45420
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.5834700670385825e-05,
+      "loss": 0.0603,
+      "step": 45425
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.5834231869110685e-05,
+      "loss": 0.0306,
+      "step": 45430
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.5833763067835545e-05,
+      "loss": 0.0526,
+      "step": 45435
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.5833294266560404e-05,
+      "loss": 0.1193,
+      "step": 45440
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.5832825465285268e-05,
+      "loss": 0.0691,
+      "step": 45445
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.5832356664010128e-05,
+      "loss": 0.1326,
+      "step": 45450
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.5831887862734988e-05,
+      "loss": 0.1811,
+      "step": 45455
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.583141906145985e-05,
+      "loss": 0.3118,
+      "step": 45460
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.583095026018471e-05,
+      "loss": 0.2734,
+      "step": 45465
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.583048145890957e-05,
+      "loss": 0.0796,
+      "step": 45470
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.583001265763443e-05,
+      "loss": 0.0566,
+      "step": 45475
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.582954385635929e-05,
+      "loss": 0.0649,
+      "step": 45480
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.582907505508415e-05,
+      "loss": 0.07,
+      "step": 45485
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.582860625380901e-05,
+      "loss": 0.1253,
+      "step": 45490
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.582813745253387e-05,
+      "loss": 0.0574,
+      "step": 45495
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.582766865125873e-05,
+      "loss": 0.129,
+      "step": 45500
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.5827199849983594e-05,
+      "loss": 0.2017,
+      "step": 45505
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.5826731048708454e-05,
+      "loss": 0.2227,
+      "step": 45510
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.5826262247433314e-05,
+      "loss": 0.3438,
+      "step": 45515
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.5825793446158177e-05,
+      "loss": 0.0246,
+      "step": 45520
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.5825324644883037e-05,
+      "loss": 0.0709,
+      "step": 45525
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.5824855843607897e-05,
+      "loss": 0.0477,
+      "step": 45530
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.5824387042332757e-05,
+      "loss": 0.1571,
+      "step": 45535
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.5823918241057617e-05,
+      "loss": 0.0867,
+      "step": 45540
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.5823449439782477e-05,
+      "loss": 0.1452,
+      "step": 45545
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.5822980638507337e-05,
+      "loss": 0.1558,
+      "step": 45550
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.58225118372322e-05,
+      "loss": 0.2085,
+      "step": 45555
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.582204303595706e-05,
+      "loss": 0.2106,
+      "step": 45560
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.582157423468192e-05,
+      "loss": 0.2487,
+      "step": 45565
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.582110543340678e-05,
+      "loss": 0.0285,
+      "step": 45570
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.582063663213164e-05,
+      "loss": 0.0829,
+      "step": 45575
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.58201678308565e-05,
+      "loss": 0.1037,
+      "step": 45580
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.5819699029581363e-05,
+      "loss": 0.1275,
+      "step": 45585
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.5819230228306223e-05,
+      "loss": 0.0762,
+      "step": 45590
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.5818761427031083e-05,
+      "loss": 0.1647,
+      "step": 45595
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.5818292625755946e-05,
+      "loss": 0.0876,
+      "step": 45600
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.5817823824480806e-05,
+      "loss": 0.202,
+      "step": 45605
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.5817355023205666e-05,
+      "loss": 0.2445,
+      "step": 45610
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.5816886221930526e-05,
+      "loss": 0.2188,
+      "step": 45615
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.5816417420655386e-05,
+      "loss": 0.0779,
+      "step": 45620
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.5815948619380246e-05,
+      "loss": 0.0369,
+      "step": 45625
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.5815479818105106e-05,
+      "loss": 0.0595,
+      "step": 45630
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.5815011016829966e-05,
+      "loss": 0.0622,
+      "step": 45635
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.5814542215554826e-05,
+      "loss": 0.0816,
+      "step": 45640
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.581407341427969e-05,
+      "loss": 0.0371,
+      "step": 45645
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.581360461300455e-05,
+      "loss": 0.2105,
+      "step": 45650
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.581313581172941e-05,
+      "loss": 0.2728,
+      "step": 45655
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.581266701045427e-05,
+      "loss": 0.3118,
+      "step": 45660
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.5812198209179132e-05,
+      "loss": 0.2977,
+      "step": 45665
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.5811729407903992e-05,
+      "loss": 0.0961,
+      "step": 45670
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.5811260606628852e-05,
+      "loss": 0.0623,
+      "step": 45675
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.5810791805353712e-05,
+      "loss": 0.0298,
+      "step": 45680
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.5810323004078572e-05,
+      "loss": 0.0836,
+      "step": 45685
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.580985420280343e-05,
+      "loss": 0.0684,
+      "step": 45690
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.5809385401528295e-05,
+      "loss": 0.0929,
+      "step": 45695
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.5808916600253155e-05,
+      "loss": 0.1611,
+      "step": 45700
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.5808447798978015e-05,
+      "loss": 0.1149,
+      "step": 45705
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.5807978997702875e-05,
+      "loss": 0.1897,
+      "step": 45710
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.5807510196427735e-05,
+      "loss": 0.345,
+      "step": 45715
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.5807041395152595e-05,
+      "loss": 0.0345,
+      "step": 45720
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.5806572593877455e-05,
+      "loss": 0.0692,
+      "step": 45725
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.5806103792602318e-05,
+      "loss": 0.0905,
+      "step": 45730
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.5805634991327178e-05,
+      "loss": 0.0561,
+      "step": 45735
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.5805166190052038e-05,
+      "loss": 0.064,
+      "step": 45740
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.58046973887769e-05,
+      "loss": 0.1136,
+      "step": 45745
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.580422858750176e-05,
+      "loss": 0.1444,
+      "step": 45750
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.580375978622662e-05,
+      "loss": 0.226,
+      "step": 45755
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.580329098495148e-05,
+      "loss": 0.4227,
+      "step": 45760
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.580282218367634e-05,
+      "loss": 0.3622,
+      "step": 45765
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.58023533824012e-05,
+      "loss": 0.0763,
+      "step": 45770
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.580188458112606e-05,
+      "loss": 0.0369,
+      "step": 45775
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.580141577985092e-05,
+      "loss": 0.045,
+      "step": 45780
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.5800946978575784e-05,
+      "loss": 0.04,
+      "step": 45785
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.5800478177300644e-05,
+      "loss": 0.0845,
+      "step": 45790
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.5800009376025504e-05,
+      "loss": 0.126,
+      "step": 45795
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.5799540574750364e-05,
+      "loss": 0.1245,
+      "step": 45800
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.5799071773475224e-05,
+      "loss": 0.2071,
+      "step": 45805
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.5798602972200087e-05,
+      "loss": 0.2901,
+      "step": 45810
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.5798134170924947e-05,
+      "loss": 0.3112,
+      "step": 45815
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.5797665369649807e-05,
+      "loss": 0.0668,
+      "step": 45820
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.5797196568374667e-05,
+      "loss": 0.0743,
+      "step": 45825
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.5796727767099527e-05,
+      "loss": 0.0486,
+      "step": 45830
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.579625896582439e-05,
+      "loss": 0.0764,
+      "step": 45835
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.579579016454925e-05,
+      "loss": 0.1223,
+      "step": 45840
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.579532136327411e-05,
+      "loss": 0.1544,
+      "step": 45845
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.579485256199897e-05,
+      "loss": 0.0958,
+      "step": 45850
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.579438376072383e-05,
+      "loss": 0.2034,
+      "step": 45855
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.579391495944869e-05,
+      "loss": 0.2167,
+      "step": 45860
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.579344615817355e-05,
+      "loss": 0.6983,
+      "step": 45865
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.5792977356898413e-05,
+      "loss": 0.0472,
+      "step": 45870
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.5792508555623273e-05,
+      "loss": 0.0692,
+      "step": 45875
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.5792039754348133e-05,
+      "loss": 0.0474,
+      "step": 45880
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.5791570953072993e-05,
+      "loss": 0.0648,
+      "step": 45885
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.5791102151797856e-05,
+      "loss": 0.0829,
+      "step": 45890
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.5790633350522716e-05,
+      "loss": 0.0981,
+      "step": 45895
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.5790164549247576e-05,
+      "loss": 0.1775,
+      "step": 45900
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.5789695747972436e-05,
+      "loss": 0.2469,
+      "step": 45905
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.5789226946697296e-05,
+      "loss": 0.2104,
+      "step": 45910
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.5788758145422156e-05,
+      "loss": 0.4498,
+      "step": 45915
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.5788289344147016e-05,
+      "loss": 0.0632,
+      "step": 45920
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.578782054287188e-05,
+      "loss": 0.1408,
+      "step": 45925
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.578735174159674e-05,
+      "loss": 0.0793,
+      "step": 45930
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.57868829403216e-05,
+      "loss": 0.0435,
+      "step": 45935
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.578641413904646e-05,
+      "loss": 0.1076,
+      "step": 45940
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.578594533777132e-05,
+      "loss": 0.164,
+      "step": 45945
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.5785476536496182e-05,
+      "loss": 0.0789,
+      "step": 45950
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.5785007735221042e-05,
+      "loss": 0.1608,
+      "step": 45955
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.5784538933945902e-05,
+      "loss": 0.2281,
+      "step": 45960
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.5784070132670762e-05,
+      "loss": 0.225,
+      "step": 45965
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.5783601331395625e-05,
+      "loss": 0.0594,
+      "step": 45970
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.5783132530120485e-05,
+      "loss": 0.1095,
+      "step": 45975
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.5782663728845345e-05,
+      "loss": 0.0838,
+      "step": 45980
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.5782194927570205e-05,
+      "loss": 0.0833,
+      "step": 45985
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.5781726126295065e-05,
+      "loss": 0.1167,
+      "step": 45990
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.5781257325019925e-05,
+      "loss": 0.1286,
+      "step": 45995
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.5780788523744785e-05,
+      "loss": 0.2014,
+      "step": 46000
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.5780319722469645e-05,
+      "loss": 0.2011,
+      "step": 46005
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.5779850921194505e-05,
+      "loss": 0.2846,
+      "step": 46010
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.5779382119919368e-05,
+      "loss": 0.337,
+      "step": 46015
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.5778913318644228e-05,
+      "loss": 0.0337,
+      "step": 46020
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.5778444517369088e-05,
+      "loss": 0.0547,
+      "step": 46025
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.577797571609395e-05,
+      "loss": 0.0493,
+      "step": 46030
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.577750691481881e-05,
+      "loss": 0.1682,
+      "step": 46035
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.577703811354367e-05,
+      "loss": 0.1686,
+      "step": 46040
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.577656931226853e-05,
+      "loss": 0.1238,
+      "step": 46045
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.577610051099339e-05,
+      "loss": 0.167,
+      "step": 46050
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.577563170971825e-05,
+      "loss": 0.1678,
+      "step": 46055
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.577516290844311e-05,
+      "loss": 0.2137,
+      "step": 46060
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.5774694107167974e-05,
+      "loss": 0.2462,
+      "step": 46065
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.5774225305892834e-05,
+      "loss": 0.049,
+      "step": 46070
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.5773756504617694e-05,
+      "loss": 0.0538,
+      "step": 46075
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.5773287703342554e-05,
+      "loss": 0.1026,
+      "step": 46080
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.5772818902067414e-05,
+      "loss": 0.0767,
+      "step": 46085
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.5772350100792274e-05,
+      "loss": 0.0594,
+      "step": 46090
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.5771881299517137e-05,
+      "loss": 0.1368,
+      "step": 46095
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.5771412498241997e-05,
+      "loss": 0.1787,
+      "step": 46100
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.5770943696966857e-05,
+      "loss": 0.2133,
+      "step": 46105
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.577047489569172e-05,
+      "loss": 0.1721,
+      "step": 46110
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.577000609441658e-05,
+      "loss": 0.2872,
+      "step": 46115
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.576953729314144e-05,
+      "loss": 0.061,
+      "step": 46120
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.57690684918663e-05,
+      "loss": 0.0356,
+      "step": 46125
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.576859969059116e-05,
+      "loss": 0.0851,
+      "step": 46130
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.576813088931602e-05,
+      "loss": 0.0768,
+      "step": 46135
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.576766208804088e-05,
+      "loss": 0.0727,
+      "step": 46140
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.576719328676574e-05,
+      "loss": 0.1419,
+      "step": 46145
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.57667244854906e-05,
+      "loss": 0.0965,
+      "step": 46150
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.576625568421546e-05,
+      "loss": 0.2987,
+      "step": 46155
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.5765786882940323e-05,
+      "loss": 0.2558,
+      "step": 46160
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.5765318081665183e-05,
+      "loss": 0.3177,
+      "step": 46165
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.5764849280390043e-05,
+      "loss": 0.0826,
+      "step": 46170
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.5764380479114906e-05,
+      "loss": 0.0487,
+      "step": 46175
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.5763911677839766e-05,
+      "loss": 0.0609,
+      "step": 46180
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.5763442876564626e-05,
+      "loss": 0.0953,
+      "step": 46185
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.5762974075289486e-05,
+      "loss": 0.1276,
+      "step": 46190
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.5762505274014346e-05,
+      "loss": 0.1125,
+      "step": 46195
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.5762036472739206e-05,
+      "loss": 0.19,
+      "step": 46200
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.576156767146407e-05,
+      "loss": 0.2064,
+      "step": 46205
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.576109887018893e-05,
+      "loss": 0.1666,
+      "step": 46210
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.576063006891379e-05,
+      "loss": 0.1832,
+      "step": 46215
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.576016126763865e-05,
+      "loss": 0.0941,
+      "step": 46220
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.575969246636351e-05,
+      "loss": 0.0844,
+      "step": 46225
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.575922366508837e-05,
+      "loss": 0.0803,
+      "step": 46230
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.575875486381323e-05,
+      "loss": 0.0812,
+      "step": 46235
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.5758286062538092e-05,
+      "loss": 0.0936,
+      "step": 46240
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.5757817261262952e-05,
+      "loss": 0.112,
+      "step": 46245
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.5757348459987812e-05,
+      "loss": 0.1694,
+      "step": 46250
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.5756879658712675e-05,
+      "loss": 0.228,
+      "step": 46255
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.5756410857437535e-05,
+      "loss": 0.2182,
+      "step": 46260
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.5755942056162395e-05,
+      "loss": 0.2981,
+      "step": 46265
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.5755473254887255e-05,
+      "loss": 0.0266,
+      "step": 46270
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.5755004453612115e-05,
+      "loss": 0.0374,
+      "step": 46275
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.5754535652336975e-05,
+      "loss": 0.0811,
+      "step": 46280
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.5754066851061835e-05,
+      "loss": 0.0798,
+      "step": 46285
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.5753598049786695e-05,
+      "loss": 0.0449,
+      "step": 46290
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.5753129248511558e-05,
+      "loss": 0.1425,
+      "step": 46295
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.5752660447236418e-05,
+      "loss": 0.1629,
+      "step": 46300
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.5752191645961278e-05,
+      "loss": 0.2877,
+      "step": 46305
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.5751722844686138e-05,
+      "loss": 0.3094,
+      "step": 46310
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.5751254043410998e-05,
+      "loss": 0.2833,
+      "step": 46315
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.575078524213586e-05,
+      "loss": 0.0353,
+      "step": 46320
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.575031644086072e-05,
+      "loss": 0.0478,
+      "step": 46325
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.574984763958558e-05,
+      "loss": 0.0581,
+      "step": 46330
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.574937883831044e-05,
+      "loss": 0.0356,
+      "step": 46335
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.57489100370353e-05,
+      "loss": 0.125,
+      "step": 46340
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.5748441235760164e-05,
+      "loss": 0.1341,
+      "step": 46345
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.5747972434485024e-05,
+      "loss": 0.1914,
+      "step": 46350
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.5747503633209884e-05,
+      "loss": 0.2266,
+      "step": 46355
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.5747034831934744e-05,
+      "loss": 0.2149,
+      "step": 46360
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.5746566030659604e-05,
+      "loss": 0.1703,
+      "step": 46365
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.5746097229384464e-05,
+      "loss": 0.0314,
+      "step": 46370
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.5745628428109324e-05,
+      "loss": 0.0379,
+      "step": 46375
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.5745159626834187e-05,
+      "loss": 0.0915,
+      "step": 46380
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.5744690825559047e-05,
+      "loss": 0.0746,
+      "step": 46385
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.5744222024283907e-05,
+      "loss": 0.12,
+      "step": 46390
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.5743753223008767e-05,
+      "loss": 0.2889,
+      "step": 46395
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.574328442173363e-05,
+      "loss": 0.0758,
+      "step": 46400
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.574281562045849e-05,
+      "loss": 0.1937,
+      "step": 46405
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.574234681918335e-05,
+      "loss": 0.2275,
+      "step": 46410
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.574187801790821e-05,
+      "loss": 0.3093,
+      "step": 46415
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.574140921663307e-05,
+      "loss": 0.0719,
+      "step": 46420
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.574094041535793e-05,
+      "loss": 0.0674,
+      "step": 46425
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.574047161408279e-05,
+      "loss": 0.0495,
+      "step": 46430
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.5740002812807653e-05,
+      "loss": 0.0669,
+      "step": 46435
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.5739534011532513e-05,
+      "loss": 0.0784,
+      "step": 46440
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.5739065210257373e-05,
+      "loss": 0.1035,
+      "step": 46445
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.5738596408982233e-05,
+      "loss": 0.1268,
+      "step": 46450
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.5738127607707093e-05,
+      "loss": 0.1276,
+      "step": 46455
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.5737658806431956e-05,
+      "loss": 0.2203,
+      "step": 46460
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.5737190005156816e-05,
+      "loss": 0.2587,
+      "step": 46465
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.5736721203881676e-05,
+      "loss": 0.0469,
+      "step": 46470
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.5736252402606536e-05,
+      "loss": 0.0494,
+      "step": 46475
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.5735783601331396e-05,
+      "loss": 0.0247,
+      "step": 46480
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.573531480005626e-05,
+      "loss": 0.1164,
+      "step": 46485
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.573484599878112e-05,
+      "loss": 0.0751,
+      "step": 46490
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.573437719750598e-05,
+      "loss": 0.1269,
+      "step": 46495
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.573390839623084e-05,
+      "loss": 0.146,
+      "step": 46500
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.57334395949557e-05,
+      "loss": 0.1816,
+      "step": 46505
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.573297079368056e-05,
+      "loss": 0.2958,
+      "step": 46510
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.573250199240542e-05,
+      "loss": 0.2065,
+      "step": 46515
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.573203319113028e-05,
+      "loss": 0.0629,
+      "step": 46520
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.5731564389855142e-05,
+      "loss": 0.0846,
+      "step": 46525
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.5731095588580002e-05,
+      "loss": 0.0455,
+      "step": 46530
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.5730626787304862e-05,
+      "loss": 0.0692,
+      "step": 46535
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.5730157986029725e-05,
+      "loss": 0.1178,
+      "step": 46540
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.5729689184754585e-05,
+      "loss": 0.1149,
+      "step": 46545
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.5729220383479445e-05,
+      "loss": 0.1512,
+      "step": 46550
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.5728751582204305e-05,
+      "loss": 0.1548,
+      "step": 46555
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.5728282780929165e-05,
+      "loss": 0.177,
+      "step": 46560
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.5727813979654025e-05,
+      "loss": 0.2508,
+      "step": 46565
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.5727345178378885e-05,
+      "loss": 0.0904,
+      "step": 46570
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.5726876377103748e-05,
+      "loss": 0.0596,
+      "step": 46575
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.5726407575828608e-05,
+      "loss": 0.0812,
+      "step": 46580
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.5725938774553468e-05,
+      "loss": 0.0603,
+      "step": 46585
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.5725469973278328e-05,
+      "loss": 0.0903,
+      "step": 46590
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.5725001172003188e-05,
+      "loss": 0.1102,
+      "step": 46595
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.5724532370728048e-05,
+      "loss": 0.129,
+      "step": 46600
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.572406356945291e-05,
+      "loss": 0.2716,
+      "step": 46605
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.572359476817777e-05,
+      "loss": 0.2493,
+      "step": 46610
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.572312596690263e-05,
+      "loss": 0.2971,
+      "step": 46615
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.5722657165627494e-05,
+      "loss": 0.0593,
+      "step": 46620
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.5722188364352354e-05,
+      "loss": 0.0314,
+      "step": 46625
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.5721719563077214e-05,
+      "loss": 0.0768,
+      "step": 46630
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.5721250761802074e-05,
+      "loss": 0.0952,
+      "step": 46635
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.5720781960526934e-05,
+      "loss": 0.0829,
+      "step": 46640
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.5720313159251794e-05,
+      "loss": 0.0766,
+      "step": 46645
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.5719844357976654e-05,
+      "loss": 0.0711,
+      "step": 46650
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.5719375556701514e-05,
+      "loss": 0.1738,
+      "step": 46655
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.5718906755426374e-05,
+      "loss": 0.1849,
+      "step": 46660
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.5718437954151234e-05,
+      "loss": 0.362,
+      "step": 46665
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.5717969152876097e-05,
+      "loss": 0.0848,
+      "step": 46670
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.5717500351600957e-05,
+      "loss": 0.0446,
+      "step": 46675
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.5717031550325817e-05,
+      "loss": 0.0797,
+      "step": 46680
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.571656274905068e-05,
+      "loss": 0.107,
+      "step": 46685
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.571609394777554e-05,
+      "loss": 0.1182,
+      "step": 46690
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.57156251465004e-05,
+      "loss": 0.1912,
+      "step": 46695
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.571515634522526e-05,
+      "loss": 0.1293,
+      "step": 46700
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.571468754395012e-05,
+      "loss": 0.1855,
+      "step": 46705
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.571421874267498e-05,
+      "loss": 0.2209,
+      "step": 46710
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.5713749941399843e-05,
+      "loss": 0.4509,
+      "step": 46715
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.5713281140124703e-05,
+      "loss": 0.0601,
+      "step": 46720
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.5712812338849563e-05,
+      "loss": 0.0555,
+      "step": 46725
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.5712343537574423e-05,
+      "loss": 0.0454,
+      "step": 46730
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.5711874736299283e-05,
+      "loss": 0.0598,
+      "step": 46735
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.5711405935024143e-05,
+      "loss": 0.0963,
+      "step": 46740
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.5710937133749003e-05,
+      "loss": 0.1205,
+      "step": 46745
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.5710468332473866e-05,
+      "loss": 0.1481,
+      "step": 46750
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.5709999531198726e-05,
+      "loss": 0.2012,
+      "step": 46755
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.5709530729923586e-05,
+      "loss": 0.2832,
+      "step": 46760
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.570906192864845e-05,
+      "loss": 0.3036,
+      "step": 46765
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.570859312737331e-05,
+      "loss": 0.0773,
+      "step": 46770
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.570812432609817e-05,
+      "loss": 0.077,
+      "step": 46775
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.570765552482303e-05,
+      "loss": 0.054,
+      "step": 46780
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.570718672354789e-05,
+      "loss": 0.112,
+      "step": 46785
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.570671792227275e-05,
+      "loss": 0.0699,
+      "step": 46790
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.570624912099761e-05,
+      "loss": 0.1254,
+      "step": 46795
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.570578031972247e-05,
+      "loss": 0.1808,
+      "step": 46800
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.570531151844733e-05,
+      "loss": 0.1349,
+      "step": 46805
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.5704842717172192e-05,
+      "loss": 0.2154,
+      "step": 46810
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.5704373915897052e-05,
+      "loss": 0.223,
+      "step": 46815
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.5703905114621912e-05,
+      "loss": 0.0263,
+      "step": 46820
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.5703436313346772e-05,
+      "loss": 0.0534,
+      "step": 46825
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.5702967512071635e-05,
+      "loss": 0.0935,
+      "step": 46830
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.5702498710796495e-05,
+      "loss": 0.1136,
+      "step": 46835
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.5702029909521355e-05,
+      "loss": 0.1008,
+      "step": 46840
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.5701561108246215e-05,
+      "loss": 0.2067,
+      "step": 46845
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.5701092306971075e-05,
+      "loss": 0.1734,
+      "step": 46850
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.570062350569594e-05,
+      "loss": 0.1801,
+      "step": 46855
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.57001547044208e-05,
+      "loss": 0.269,
+      "step": 46860
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.569968590314566e-05,
+      "loss": 0.2084,
+      "step": 46865
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.5699217101870518e-05,
+      "loss": 0.0311,
+      "step": 46870
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.5698748300595378e-05,
+      "loss": 0.0641,
+      "step": 46875
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.5698279499320238e-05,
+      "loss": 0.0508,
+      "step": 46880
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.5697810698045098e-05,
+      "loss": 0.0901,
+      "step": 46885
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.569734189676996e-05,
+      "loss": 0.1451,
+      "step": 46890
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.569687309549482e-05,
+      "loss": 0.0871,
+      "step": 46895
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.569640429421968e-05,
+      "loss": 0.1306,
+      "step": 46900
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.569593549294454e-05,
+      "loss": 0.1422,
+      "step": 46905
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.5695466691669405e-05,
+      "loss": 0.2327,
+      "step": 46910
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.5694997890394264e-05,
+      "loss": 0.2564,
+      "step": 46915
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.5694529089119124e-05,
+      "loss": 0.0679,
+      "step": 46920
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.5694060287843984e-05,
+      "loss": 0.0566,
+      "step": 46925
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.5693591486568844e-05,
+      "loss": 0.0412,
+      "step": 46930
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.5693122685293704e-05,
+      "loss": 0.0899,
+      "step": 46935
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.5692653884018564e-05,
+      "loss": 0.0676,
+      "step": 46940
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.5692185082743427e-05,
+      "loss": 0.0635,
+      "step": 46945
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.5691716281468287e-05,
+      "loss": 0.1505,
+      "step": 46950
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.5691247480193147e-05,
+      "loss": 0.138,
+      "step": 46955
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.5690778678918007e-05,
+      "loss": 0.3916,
+      "step": 46960
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.5690309877642867e-05,
+      "loss": 0.3525,
+      "step": 46965
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.568984107636773e-05,
+      "loss": 0.0513,
+      "step": 46970
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.568937227509259e-05,
+      "loss": 0.0724,
+      "step": 46975
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.568890347381745e-05,
+      "loss": 0.0959,
+      "step": 46980
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.568843467254231e-05,
+      "loss": 0.1125,
+      "step": 46985
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.568796587126717e-05,
+      "loss": 0.081,
+      "step": 46990
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.5687497069992034e-05,
+      "loss": 0.2037,
+      "step": 46995
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.5687028268716893e-05,
+      "loss": 0.1913,
+      "step": 47000
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.5686559467441753e-05,
+      "loss": 0.126,
+      "step": 47005
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.5686090666166613e-05,
+      "loss": 0.2837,
+      "step": 47010
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.5685621864891473e-05,
+      "loss": 0.2889,
+      "step": 47015
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.5685153063616333e-05,
+      "loss": 0.0453,
+      "step": 47020
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.5684684262341193e-05,
+      "loss": 0.0343,
+      "step": 47025
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.5684215461066053e-05,
+      "loss": 0.0998,
+      "step": 47030
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.5683746659790916e-05,
+      "loss": 0.0703,
+      "step": 47035
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.5683277858515776e-05,
+      "loss": 0.0683,
+      "step": 47040
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.5682809057240636e-05,
+      "loss": 0.1142,
+      "step": 47045
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.56823402559655e-05,
+      "loss": 0.1156,
+      "step": 47050
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.568187145469036e-05,
+      "loss": 0.0922,
+      "step": 47055
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.568140265341522e-05,
+      "loss": 0.1995,
+      "step": 47060
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.568093385214008e-05,
+      "loss": 0.3706,
+      "step": 47065
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.568046505086494e-05,
+      "loss": 0.0824,
+      "step": 47070
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.56799962495898e-05,
+      "loss": 0.0288,
+      "step": 47075
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.567952744831466e-05,
+      "loss": 0.0757,
+      "step": 47080
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.5679058647039523e-05,
+      "loss": 0.1502,
+      "step": 47085
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.5678589845764382e-05,
+      "loss": 0.1346,
+      "step": 47090
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.5678121044489242e-05,
+      "loss": 0.2792,
+      "step": 47095
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.5677652243214102e-05,
+      "loss": 0.1159,
+      "step": 47100
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.5677183441938962e-05,
+      "loss": 0.0856,
+      "step": 47105
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.5676714640663822e-05,
+      "loss": 0.2502,
+      "step": 47110
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.5676245839388686e-05,
+      "loss": 0.3666,
+      "step": 47115
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.5675777038113545e-05,
+      "loss": 0.0493,
+      "step": 47120
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.5675308236838405e-05,
+      "loss": 0.0624,
+      "step": 47125
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.5674839435563265e-05,
+      "loss": 0.0632,
+      "step": 47130
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.567437063428813e-05,
+      "loss": 0.0216,
+      "step": 47135
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.567390183301299e-05,
+      "loss": 0.0983,
+      "step": 47140
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.567343303173785e-05,
+      "loss": 0.1241,
+      "step": 47145
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.567296423046271e-05,
+      "loss": 0.1748,
+      "step": 47150
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.567249542918757e-05,
+      "loss": 0.1428,
+      "step": 47155
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.5672026627912428e-05,
+      "loss": 0.1892,
+      "step": 47160
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.5671557826637288e-05,
+      "loss": 0.2356,
+      "step": 47165
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.5671089025362148e-05,
+      "loss": 0.0544,
+      "step": 47170
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.5670620224087008e-05,
+      "loss": 0.0387,
+      "step": 47175
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.567015142281187e-05,
+      "loss": 0.0373,
+      "step": 47180
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.566968262153673e-05,
+      "loss": 0.0462,
+      "step": 47185
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.566921382026159e-05,
+      "loss": 0.0844,
+      "step": 47190
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.5668745018986455e-05,
+      "loss": 0.1622,
+      "step": 47195
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.5668276217711315e-05,
+      "loss": 0.1004,
+      "step": 47200
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.5667807416436174e-05,
+      "loss": 0.1573,
+      "step": 47205
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.5667338615161034e-05,
+      "loss": 0.235,
+      "step": 47210
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.5666869813885894e-05,
+      "loss": 0.3367,
+      "step": 47215
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.5666401012610754e-05,
+      "loss": 0.0501,
+      "step": 47220
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.5665932211335618e-05,
+      "loss": 0.0445,
+      "step": 47225
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.5665463410060478e-05,
+      "loss": 0.0511,
+      "step": 47230
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.5664994608785337e-05,
+      "loss": 0.1131,
+      "step": 47235
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.5664525807510197e-05,
+      "loss": 0.0836,
+      "step": 47240
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.5664057006235057e-05,
+      "loss": 0.0873,
+      "step": 47245
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.5663588204959917e-05,
+      "loss": 0.092,
+      "step": 47250
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.5663119403684777e-05,
+      "loss": 0.1788,
+      "step": 47255
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.566265060240964e-05,
+      "loss": 0.1563,
+      "step": 47260
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.56621818011345e-05,
+      "loss": 0.3004,
+      "step": 47265
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.566171299985936e-05,
+      "loss": 0.0335,
+      "step": 47270
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.5661244198584224e-05,
+      "loss": 0.053,
+      "step": 47275
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.5660775397309084e-05,
+      "loss": 0.0392,
+      "step": 47280
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.5660306596033944e-05,
+      "loss": 0.056,
+      "step": 47285
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.5659837794758804e-05,
+      "loss": 0.1362,
+      "step": 47290
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.5659368993483663e-05,
+      "loss": 0.1154,
+      "step": 47295
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.5658900192208523e-05,
+      "loss": 0.1525,
+      "step": 47300
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.5658431390933383e-05,
+      "loss": 0.247,
+      "step": 47305
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.5657962589658243e-05,
+      "loss": 0.2236,
+      "step": 47310
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.5657493788383103e-05,
+      "loss": 0.2756,
+      "step": 47315
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.5657024987107967e-05,
+      "loss": 0.0349,
+      "step": 47320
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.5656556185832826e-05,
+      "loss": 0.0435,
+      "step": 47325
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.5656087384557686e-05,
+      "loss": 0.0459,
+      "step": 47330
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.565561858328255e-05,
+      "loss": 0.1041,
+      "step": 47335
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.565514978200741e-05,
+      "loss": 0.0852,
+      "step": 47340
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.565468098073227e-05,
+      "loss": 0.1001,
+      "step": 47345
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.565421217945713e-05,
+      "loss": 0.1142,
+      "step": 47350
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.565374337818199e-05,
+      "loss": 0.0887,
+      "step": 47355
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.565327457690685e-05,
+      "loss": 0.2389,
+      "step": 47360
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.5652805775631713e-05,
+      "loss": 0.2464,
+      "step": 47365
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.5652336974356573e-05,
+      "loss": 0.032,
+      "step": 47370
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.5651868173081433e-05,
+      "loss": 0.0374,
+      "step": 47375
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.5651399371806292e-05,
+      "loss": 0.0476,
+      "step": 47380
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.5650930570531152e-05,
+      "loss": 0.0691,
+      "step": 47385
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.5650461769256012e-05,
+      "loss": 0.1024,
+      "step": 47390
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.5649992967980872e-05,
+      "loss": 0.0723,
+      "step": 47395
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.5649524166705736e-05,
+      "loss": 0.1205,
+      "step": 47400
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.5649055365430596e-05,
+      "loss": 0.1828,
+      "step": 47405
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.5648586564155455e-05,
+      "loss": 0.2641,
+      "step": 47410
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.564811776288032e-05,
+      "loss": 0.217,
+      "step": 47415
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.564764896160518e-05,
+      "loss": 0.0275,
+      "step": 47420
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.564718016033004e-05,
+      "loss": 0.0234,
+      "step": 47425
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.56467113590549e-05,
+      "loss": 0.019,
+      "step": 47430
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.564624255777976e-05,
+      "loss": 0.0693,
+      "step": 47435
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.564577375650462e-05,
+      "loss": 0.1484,
+      "step": 47440
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.564530495522948e-05,
+      "loss": 0.0962,
+      "step": 47445
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.564483615395434e-05,
+      "loss": 0.1063,
+      "step": 47450
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.5644367352679198e-05,
+      "loss": 0.1541,
+      "step": 47455
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.564389855140406e-05,
+      "loss": 0.2506,
+      "step": 47460
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.564342975012892e-05,
+      "loss": 0.2357,
+      "step": 47465
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.564296094885378e-05,
+      "loss": 0.043,
+      "step": 47470
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.564249214757864e-05,
+      "loss": 0.0458,
+      "step": 47475
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.5642023346303505e-05,
+      "loss": 0.0539,
+      "step": 47480
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.5641554545028365e-05,
+      "loss": 0.0616,
+      "step": 47485
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.5641085743753225e-05,
+      "loss": 0.15,
+      "step": 47490
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.5640616942478085e-05,
+      "loss": 0.1106,
+      "step": 47495
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.5640148141202944e-05,
+      "loss": 0.1981,
+      "step": 47500
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.5639679339927808e-05,
+      "loss": 0.1841,
+      "step": 47505
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.5639210538652668e-05,
+      "loss": 0.1291,
+      "step": 47510
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.5638741737377528e-05,
+      "loss": 0.4161,
+      "step": 47515
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.5638272936102388e-05,
+      "loss": 0.0439,
+      "step": 47520
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.5637804134827248e-05,
+      "loss": 0.0376,
+      "step": 47525
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.5637335333552107e-05,
+      "loss": 0.0504,
+      "step": 47530
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.5636866532276967e-05,
+      "loss": 0.0765,
+      "step": 47535
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.5636397731001827e-05,
+      "loss": 0.0547,
+      "step": 47540
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.563592892972669e-05,
+      "loss": 0.0697,
+      "step": 47545
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.563546012845155e-05,
+      "loss": 0.1192,
+      "step": 47550
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.563499132717641e-05,
+      "loss": 0.1758,
+      "step": 47555
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.5634522525901274e-05,
+      "loss": 0.3129,
+      "step": 47560
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.5634053724626134e-05,
+      "loss": 0.2956,
+      "step": 47565
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.5633584923350994e-05,
+      "loss": 0.1004,
+      "step": 47570
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.5633116122075854e-05,
+      "loss": 0.033,
+      "step": 47575
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.5632647320800714e-05,
+      "loss": 0.0694,
+      "step": 47580
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.5632178519525573e-05,
+      "loss": 0.0848,
+      "step": 47585
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.5631709718250433e-05,
+      "loss": 0.1193,
+      "step": 47590
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.5631240916975297e-05,
+      "loss": 0.1105,
+      "step": 47595
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.5630772115700157e-05,
+      "loss": 0.2483,
+      "step": 47600
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.5630303314425017e-05,
+      "loss": 0.186,
+      "step": 47605
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.5629834513149877e-05,
+      "loss": 0.243,
+      "step": 47610
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.5629365711874736e-05,
+      "loss": 0.2054,
+      "step": 47615
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.5628896910599596e-05,
+      "loss": 0.0592,
+      "step": 47620
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.562842810932446e-05,
+      "loss": 0.0247,
+      "step": 47625
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.562795930804932e-05,
+      "loss": 0.0993,
+      "step": 47630
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.562749050677418e-05,
+      "loss": 0.0961,
+      "step": 47635
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.562702170549904e-05,
+      "loss": 0.1097,
+      "step": 47640
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.5626552904223903e-05,
+      "loss": 0.0517,
+      "step": 47645
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.5626084102948763e-05,
+      "loss": 0.103,
+      "step": 47650
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.5625615301673623e-05,
+      "loss": 0.1968,
+      "step": 47655
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.5625146500398483e-05,
+      "loss": 0.4059,
+      "step": 47660
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.5624677699123343e-05,
+      "loss": 0.3454,
+      "step": 47665
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.5624208897848203e-05,
+      "loss": 0.07,
+      "step": 47670
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.5623740096573062e-05,
+      "loss": 0.0739,
+      "step": 47675
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.5623271295297922e-05,
+      "loss": 0.0511,
+      "step": 47680
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.5622802494022786e-05,
+      "loss": 0.0457,
+      "step": 47685
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.5622333692747646e-05,
+      "loss": 0.0793,
+      "step": 47690
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.5621864891472506e-05,
+      "loss": 0.1131,
+      "step": 47695
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.5621396090197366e-05,
+      "loss": 0.1825,
+      "step": 47700
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.562092728892223e-05,
+      "loss": 0.1641,
+      "step": 47705
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.562045848764709e-05,
+      "loss": 0.3252,
+      "step": 47710
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.561998968637195e-05,
+      "loss": 0.2352,
+      "step": 47715
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.561952088509681e-05,
+      "loss": 0.0826,
+      "step": 47720
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.561905208382167e-05,
+      "loss": 0.0546,
+      "step": 47725
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.561858328254653e-05,
+      "loss": 0.1073,
+      "step": 47730
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.5618114481271392e-05,
+      "loss": 0.0686,
+      "step": 47735
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.5617645679996252e-05,
+      "loss": 0.118,
+      "step": 47740
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.561717687872111e-05,
+      "loss": 0.132,
+      "step": 47745
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.561670807744597e-05,
+      "loss": 0.1841,
+      "step": 47750
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.561623927617083e-05,
+      "loss": 0.1671,
+      "step": 47755
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.561577047489569e-05,
+      "loss": 0.3364,
+      "step": 47760
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.5615301673620555e-05,
+      "loss": 0.2793,
+      "step": 47765
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.5614832872345415e-05,
+      "loss": 0.0919,
+      "step": 47770
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.5614364071070275e-05,
+      "loss": 0.0524,
+      "step": 47775
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.5613895269795135e-05,
+      "loss": 0.107,
+      "step": 47780
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.5613426468519998e-05,
+      "loss": 0.0647,
+      "step": 47785
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.5612957667244858e-05,
+      "loss": 0.1325,
+      "step": 47790
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.5612488865969718e-05,
+      "loss": 0.1216,
+      "step": 47795
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.5612020064694578e-05,
+      "loss": 0.1204,
+      "step": 47800
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.5611551263419438e-05,
+      "loss": 0.1161,
+      "step": 47805
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.5611082462144298e-05,
+      "loss": 0.1865,
+      "step": 47810
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.5610613660869158e-05,
+      "loss": 0.2131,
+      "step": 47815
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.5610144859594017e-05,
+      "loss": 0.0688,
+      "step": 47820
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.5609676058318877e-05,
+      "loss": 0.0357,
+      "step": 47825
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.560920725704374e-05,
+      "loss": 0.0453,
+      "step": 47830
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.56087384557686e-05,
+      "loss": 0.1079,
+      "step": 47835
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.560826965449346e-05,
+      "loss": 0.0614,
+      "step": 47840
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.5607800853218324e-05,
+      "loss": 0.1924,
+      "step": 47845
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.5607332051943184e-05,
+      "loss": 0.1522,
+      "step": 47850
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.5606863250668044e-05,
+      "loss": 0.1627,
+      "step": 47855
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.5606394449392904e-05,
+      "loss": 0.28,
+      "step": 47860
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.5605925648117764e-05,
+      "loss": 0.2299,
+      "step": 47865
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.5605456846842624e-05,
+      "loss": 0.087,
+      "step": 47870
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.5604988045567487e-05,
+      "loss": 0.0141,
+      "step": 47875
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.5604519244292347e-05,
+      "loss": 0.0714,
+      "step": 47880
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.5604050443017207e-05,
+      "loss": 0.0633,
+      "step": 47885
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.5603581641742067e-05,
+      "loss": 0.1126,
+      "step": 47890
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.5603112840466927e-05,
+      "loss": 0.0534,
+      "step": 47895
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.5602644039191787e-05,
+      "loss": 0.1151,
+      "step": 47900
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.5602175237916647e-05,
+      "loss": 0.2126,
+      "step": 47905
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.560170643664151e-05,
+      "loss": 0.1806,
+      "step": 47910
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.560123763536637e-05,
+      "loss": 0.355,
+      "step": 47915
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.560076883409123e-05,
+      "loss": 0.0739,
+      "step": 47920
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.5600300032816093e-05,
+      "loss": 0.0643,
+      "step": 47925
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.5599831231540953e-05,
+      "loss": 0.0717,
+      "step": 47930
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.5599362430265813e-05,
+      "loss": 0.0994,
+      "step": 47935
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.5598893628990673e-05,
+      "loss": 0.1345,
+      "step": 47940
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.5598424827715533e-05,
+      "loss": 0.1039,
+      "step": 47945
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.5597956026440393e-05,
+      "loss": 0.1926,
+      "step": 47950
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.5597487225165253e-05,
+      "loss": 0.1395,
+      "step": 47955
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.5597018423890113e-05,
+      "loss": 0.3645,
+      "step": 47960
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.5596549622614972e-05,
+      "loss": 0.3706,
+      "step": 47965
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.5596080821339836e-05,
+      "loss": 0.0203,
+      "step": 47970
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.5595612020064696e-05,
+      "loss": 0.0991,
+      "step": 47975
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.5595143218789556e-05,
+      "loss": 0.0464,
+      "step": 47980
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.5594674417514416e-05,
+      "loss": 0.036,
+      "step": 47985
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.559420561623928e-05,
+      "loss": 0.1018,
+      "step": 47990
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.559373681496414e-05,
+      "loss": 0.1252,
+      "step": 47995
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.5593268013689e-05,
+      "loss": 0.093,
+      "step": 48000
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.559279921241386e-05,
+      "loss": 0.1209,
+      "step": 48005
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.559233041113872e-05,
+      "loss": 0.3554,
+      "step": 48010
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.5591861609863582e-05,
+      "loss": 0.3472,
+      "step": 48015
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.5591392808588442e-05,
+      "loss": 0.0574,
+      "step": 48020
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.5590924007313302e-05,
+      "loss": 0.074,
+      "step": 48025
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.5590455206038162e-05,
+      "loss": 0.1075,
+      "step": 48030
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.5589986404763022e-05,
+      "loss": 0.0526,
+      "step": 48035
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.558951760348788e-05,
+      "loss": 0.0969,
+      "step": 48040
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.558904880221274e-05,
+      "loss": 0.1061,
+      "step": 48045
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.55885800009376e-05,
+      "loss": 0.2251,
+      "step": 48050
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.5588111199662465e-05,
+      "loss": 0.2318,
+      "step": 48055
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.5587642398387325e-05,
+      "loss": 0.1552,
+      "step": 48060
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.5587173597112185e-05,
+      "loss": 0.3252,
+      "step": 48065
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.5586704795837048e-05,
+      "loss": 0.1035,
+      "step": 48070
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.5586235994561908e-05,
+      "loss": 0.0411,
+      "step": 48075
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.5585767193286768e-05,
+      "loss": 0.0677,
+      "step": 48080
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.5585298392011628e-05,
+      "loss": 0.0637,
+      "step": 48085
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.5584829590736488e-05,
+      "loss": 0.1163,
+      "step": 48090
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.5584360789461348e-05,
+      "loss": 0.0817,
+      "step": 48095
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.5583891988186208e-05,
+      "loss": 0.1213,
+      "step": 48100
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.5583423186911068e-05,
+      "loss": 0.2463,
+      "step": 48105
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.558295438563593e-05,
+      "loss": 0.2425,
+      "step": 48110
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.558248558436079e-05,
+      "loss": 0.2606,
+      "step": 48115
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.558201678308565e-05,
+      "loss": 0.0782,
+      "step": 48120
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.558154798181051e-05,
+      "loss": 0.0525,
+      "step": 48125
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.558107918053537e-05,
+      "loss": 0.0192,
+      "step": 48130
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.5580610379260234e-05,
+      "loss": 0.0877,
+      "step": 48135
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.5580141577985094e-05,
+      "loss": 0.1597,
+      "step": 48140
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.5579672776709954e-05,
+      "loss": 0.0772,
+      "step": 48145
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.5579203975434814e-05,
+      "loss": 0.1402,
+      "step": 48150
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.5578735174159677e-05,
+      "loss": 0.1657,
+      "step": 48155
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.5578266372884537e-05,
+      "loss": 0.3195,
+      "step": 48160
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.5577797571609397e-05,
+      "loss": 0.2577,
+      "step": 48165
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.5577328770334257e-05,
+      "loss": 0.0815,
+      "step": 48170
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.5576859969059117e-05,
+      "loss": 0.0641,
+      "step": 48175
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.5576391167783977e-05,
+      "loss": 0.0169,
+      "step": 48180
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.5575922366508837e-05,
+      "loss": 0.1095,
+      "step": 48185
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.5575453565233697e-05,
+      "loss": 0.0848,
+      "step": 48190
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.557498476395856e-05,
+      "loss": 0.0564,
+      "step": 48195
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.557451596268342e-05,
+      "loss": 0.2123,
+      "step": 48200
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.557404716140828e-05,
+      "loss": 0.154,
+      "step": 48205
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.557357836013314e-05,
+      "loss": 0.2392,
+      "step": 48210
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.5573109558858003e-05,
+      "loss": 0.3414,
+      "step": 48215
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.5572640757582863e-05,
+      "loss": 0.094,
+      "step": 48220
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.5572171956307723e-05,
+      "loss": 0.0229,
+      "step": 48225
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.5571703155032583e-05,
+      "loss": 0.0471,
+      "step": 48230
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.5571234353757443e-05,
+      "loss": 0.1453,
+      "step": 48235
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.5570765552482303e-05,
+      "loss": 0.0759,
+      "step": 48240
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.5570296751207166e-05,
+      "loss": 0.1313,
+      "step": 48245
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.5569827949932026e-05,
+      "loss": 0.1588,
+      "step": 48250
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.5569359148656886e-05,
+      "loss": 0.1738,
+      "step": 48255
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.5568890347381746e-05,
+      "loss": 0.2611,
+      "step": 48260
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.5568421546106606e-05,
+      "loss": 0.2127,
+      "step": 48265
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.5567952744831466e-05,
+      "loss": 0.0951,
+      "step": 48270
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.556748394355633e-05,
+      "loss": 0.0701,
+      "step": 48275
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.556701514228119e-05,
+      "loss": 0.0602,
+      "step": 48280
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.556654634100605e-05,
+      "loss": 0.1011,
+      "step": 48285
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.556607753973091e-05,
+      "loss": 0.075,
+      "step": 48290
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.5565608738455772e-05,
+      "loss": 0.0798,
+      "step": 48295
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.5565139937180632e-05,
+      "loss": 0.1493,
+      "step": 48300
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.5564671135905492e-05,
+      "loss": 0.2064,
+      "step": 48305
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.5564202334630352e-05,
+      "loss": 0.2354,
+      "step": 48310
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.5563733533355212e-05,
+      "loss": 0.3015,
+      "step": 48315
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.5563264732080072e-05,
+      "loss": 0.0693,
+      "step": 48320
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.5562795930804932e-05,
+      "loss": 0.0577,
+      "step": 48325
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.556232712952979e-05,
+      "loss": 0.0701,
+      "step": 48330
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.556185832825465e-05,
+      "loss": 0.0724,
+      "step": 48335
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.5561389526979515e-05,
+      "loss": 0.0785,
+      "step": 48340
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.5560920725704375e-05,
+      "loss": 0.1078,
+      "step": 48345
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.5560451924429235e-05,
+      "loss": 0.1763,
+      "step": 48350
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.5559983123154098e-05,
+      "loss": 0.182,
+      "step": 48355
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.5559514321878958e-05,
+      "loss": 0.2734,
+      "step": 48360
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.5559045520603818e-05,
+      "loss": 0.4282,
+      "step": 48365
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.5558576719328678e-05,
+      "loss": 0.0159,
+      "step": 48370
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.5558107918053538e-05,
+      "loss": 0.0192,
+      "step": 48375
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.5557639116778398e-05,
+      "loss": 0.0352,
+      "step": 48380
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.555717031550326e-05,
+      "loss": 0.0958,
+      "step": 48385
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.555670151422812e-05,
+      "loss": 0.0457,
+      "step": 48390
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.555623271295298e-05,
+      "loss": 0.0711,
+      "step": 48395
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.555576391167784e-05,
+      "loss": 0.0788,
+      "step": 48400
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.55552951104027e-05,
+      "loss": 0.1404,
+      "step": 48405
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.555482630912756e-05,
+      "loss": 0.1612,
+      "step": 48410
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.555435750785242e-05,
+      "loss": 0.1604,
+      "step": 48415
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.5553888706577284e-05,
+      "loss": 0.0949,
+      "step": 48420
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.5553419905302144e-05,
+      "loss": 0.0609,
+      "step": 48425
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.5552951104027004e-05,
+      "loss": 0.0533,
+      "step": 48430
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.5552482302751867e-05,
+      "loss": 0.0413,
+      "step": 48435
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.5552013501476727e-05,
+      "loss": 0.0969,
+      "step": 48440
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.5551544700201587e-05,
+      "loss": 0.082,
+      "step": 48445
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.5551075898926447e-05,
+      "loss": 0.125,
+      "step": 48450
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.5550607097651307e-05,
+      "loss": 0.1438,
+      "step": 48455
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.5550138296376167e-05,
+      "loss": 0.2595,
+      "step": 48460
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.5549669495101027e-05,
+      "loss": 0.3857,
+      "step": 48465
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.5549200693825887e-05,
+      "loss": 0.0179,
+      "step": 48470
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.5548731892550747e-05,
+      "loss": 0.0542,
+      "step": 48475
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.554826309127561e-05,
+      "loss": 0.0466,
+      "step": 48480
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.554779429000047e-05,
+      "loss": 0.0854,
+      "step": 48485
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.554732548872533e-05,
+      "loss": 0.1496,
+      "step": 48490
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.554685668745019e-05,
+      "loss": 0.106,
+      "step": 48495
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.5546387886175053e-05,
+      "loss": 0.2021,
+      "step": 48500
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.5545919084899913e-05,
+      "loss": 0.1652,
+      "step": 48505
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.5545450283624773e-05,
+      "loss": 0.2523,
+      "step": 48510
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.5544981482349633e-05,
+      "loss": 0.3244,
+      "step": 48515
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.5544512681074493e-05,
+      "loss": 0.1141,
+      "step": 48520
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.5544043879799356e-05,
+      "loss": 0.0334,
+      "step": 48525
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.5543575078524216e-05,
+      "loss": 0.0672,
+      "step": 48530
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.5543106277249076e-05,
+      "loss": 0.0809,
+      "step": 48535
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.5542637475973936e-05,
+      "loss": 0.1522,
+      "step": 48540
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.5542168674698796e-05,
+      "loss": 0.0487,
+      "step": 48545
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.5541699873423656e-05,
+      "loss": 0.1164,
+      "step": 48550
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.5541231072148516e-05,
+      "loss": 0.2902,
+      "step": 48555
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.5540762270873376e-05,
+      "loss": 0.2218,
+      "step": 48560
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.554029346959824e-05,
+      "loss": 0.2843,
+      "step": 48565
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.55398246683231e-05,
+      "loss": 0.0109,
+      "step": 48570
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.553935586704796e-05,
+      "loss": 0.0523,
+      "step": 48575
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.5538887065772822e-05,
+      "loss": 0.0514,
+      "step": 48580
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.5538418264497682e-05,
+      "loss": 0.0571,
+      "step": 48585
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.5537949463222542e-05,
+      "loss": 0.1356,
+      "step": 48590
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.5537480661947402e-05,
+      "loss": 0.129,
+      "step": 48595
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.5537011860672262e-05,
+      "loss": 0.1741,
+      "step": 48600
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.5536543059397122e-05,
+      "loss": 0.1214,
+      "step": 48605
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.5536074258121982e-05,
+      "loss": 0.3035,
+      "step": 48610
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.5535605456846842e-05,
+      "loss": 0.4002,
+      "step": 48615
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.5535136655571705e-05,
+      "loss": 0.0531,
+      "step": 48620
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.5534667854296565e-05,
+      "loss": 0.0398,
+      "step": 48625
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.5534199053021425e-05,
+      "loss": 0.0443,
+      "step": 48630
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.5533730251746285e-05,
+      "loss": 0.0927,
+      "step": 48635
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.5533261450471145e-05,
+      "loss": 0.177,
+      "step": 48640
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.5532792649196008e-05,
+      "loss": 0.0691,
+      "step": 48645
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.5532323847920868e-05,
+      "loss": 0.1293,
+      "step": 48650
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.5531855046645728e-05,
+      "loss": 0.1106,
+      "step": 48655
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.5531386245370588e-05,
+      "loss": 0.2808,
+      "step": 48660
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.553091744409545e-05,
+      "loss": 0.1944,
+      "step": 48665
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.553044864282031e-05,
+      "loss": 0.1225,
+      "step": 48670
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.552997984154517e-05,
+      "loss": 0.0264,
+      "step": 48675
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.552951104027003e-05,
+      "loss": 0.0261,
+      "step": 48680
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.552904223899489e-05,
+      "loss": 0.036,
+      "step": 48685
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.552857343771975e-05,
+      "loss": 0.1202,
+      "step": 48690
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.552810463644461e-05,
+      "loss": 0.1904,
+      "step": 48695
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.552763583516947e-05,
+      "loss": 0.1372,
+      "step": 48700
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.5527167033894334e-05,
+      "loss": 0.1831,
+      "step": 48705
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.5526698232619194e-05,
+      "loss": 0.4118,
+      "step": 48710
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.5526229431344054e-05,
+      "loss": 0.2792,
+      "step": 48715
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.5525760630068917e-05,
+      "loss": 0.0797,
+      "step": 48720
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.5525291828793777e-05,
+      "loss": 0.0872,
+      "step": 48725
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.5524823027518637e-05,
+      "loss": 0.0749,
+      "step": 48730
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.5524354226243497e-05,
+      "loss": 0.0277,
+      "step": 48735
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.5523885424968357e-05,
+      "loss": 0.1316,
+      "step": 48740
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.5523416623693217e-05,
+      "loss": 0.0827,
+      "step": 48745
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.5522947822418077e-05,
+      "loss": 0.0588,
+      "step": 48750
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.5522479021142937e-05,
+      "loss": 0.1939,
+      "step": 48755
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.55220102198678e-05,
+      "loss": 0.1868,
+      "step": 48760
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.552154141859266e-05,
+      "loss": 0.2015,
+      "step": 48765
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.552107261731752e-05,
+      "loss": 0.1162,
+      "step": 48770
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.552060381604238e-05,
+      "loss": 0.098,
+      "step": 48775
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.552013501476724e-05,
+      "loss": 0.0625,
+      "step": 48780
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.5519666213492103e-05,
+      "loss": 0.0749,
+      "step": 48785
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.5519197412216963e-05,
+      "loss": 0.0543,
+      "step": 48790
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.5518728610941823e-05,
+      "loss": 0.1783,
+      "step": 48795
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.5518259809666683e-05,
+      "loss": 0.1378,
+      "step": 48800
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.5517791008391546e-05,
+      "loss": 0.1705,
+      "step": 48805
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.5517322207116406e-05,
+      "loss": 0.3341,
+      "step": 48810
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.5516853405841266e-05,
+      "loss": 0.3355,
+      "step": 48815
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.5516384604566126e-05,
+      "loss": 0.0187,
+      "step": 48820
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.5515915803290986e-05,
+      "loss": 0.0375,
+      "step": 48825
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.5515447002015846e-05,
+      "loss": 0.0405,
+      "step": 48830
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.5514978200740706e-05,
+      "loss": 0.0343,
+      "step": 48835
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.5514509399465566e-05,
+      "loss": 0.0258,
+      "step": 48840
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.5514040598190426e-05,
+      "loss": 0.125,
+      "step": 48845
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.551357179691529e-05,
+      "loss": 0.1763,
+      "step": 48850
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.551310299564015e-05,
+      "loss": 0.1754,
+      "step": 48855
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.551263419436501e-05,
+      "loss": 0.2588,
+      "step": 48860
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.5512165393089872e-05,
+      "loss": 0.3456,
+      "step": 48865
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.5511696591814732e-05,
+      "loss": 0.04,
+      "step": 48870
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.5511227790539592e-05,
+      "loss": 0.0311,
+      "step": 48875
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.5510758989264452e-05,
+      "loss": 0.0711,
+      "step": 48880
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.5510290187989312e-05,
+      "loss": 0.0476,
+      "step": 48885
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.5509821386714172e-05,
+      "loss": 0.0521,
+      "step": 48890
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.5509352585439035e-05,
+      "loss": 0.1187,
+      "step": 48895
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.5508883784163895e-05,
+      "loss": 0.2008,
+      "step": 48900
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.5508414982888755e-05,
+      "loss": 0.1893,
+      "step": 48905
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.5507946181613615e-05,
+      "loss": 0.2853,
+      "step": 48910
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.5507477380338475e-05,
+      "loss": 0.364,
+      "step": 48915
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.5507008579063335e-05,
+      "loss": 0.0612,
+      "step": 48920
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.5506539777788195e-05,
+      "loss": 0.0777,
+      "step": 48925
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.5506070976513058e-05,
+      "loss": 0.0366,
+      "step": 48930
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.5505602175237918e-05,
+      "loss": 0.0794,
+      "step": 48935
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.5505133373962778e-05,
+      "loss": 0.0836,
+      "step": 48940
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.550466457268764e-05,
+      "loss": 0.0946,
+      "step": 48945
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.55041957714125e-05,
+      "loss": 0.1121,
+      "step": 48950
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.550372697013736e-05,
+      "loss": 0.1959,
+      "step": 48955
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.550325816886222e-05,
+      "loss": 0.359,
+      "step": 48960
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.550278936758708e-05,
+      "loss": 0.2793,
+      "step": 48965
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.550232056631194e-05,
+      "loss": 0.0281,
+      "step": 48970
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.55018517650368e-05,
+      "loss": 0.0542,
+      "step": 48975
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.550138296376166e-05,
+      "loss": 0.0876,
+      "step": 48980
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.550091416248652e-05,
+      "loss": 0.086,
+      "step": 48985
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.5500445361211384e-05,
+      "loss": 0.108,
+      "step": 48990
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.5499976559936244e-05,
+      "loss": 0.1401,
+      "step": 48995
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.5499507758661104e-05,
+      "loss": 0.191,
+      "step": 49000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.5499038957385964e-05,
+      "loss": 0.1504,
+      "step": 49005
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.5498570156110827e-05,
+      "loss": 0.2,
+      "step": 49010
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.5498101354835687e-05,
+      "loss": 0.3442,
+      "step": 49015
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.5497632553560547e-05,
+      "loss": 0.072,
+      "step": 49020
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.5497163752285407e-05,
+      "loss": 0.0782,
+      "step": 49025
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.5496694951010267e-05,
+      "loss": 0.0862,
+      "step": 49030
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.549622614973513e-05,
+      "loss": 0.056,
+      "step": 49035
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.549575734845999e-05,
+      "loss": 0.0957,
+      "step": 49040
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.549528854718485e-05,
+      "loss": 0.1298,
+      "step": 49045
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.549481974590971e-05,
+      "loss": 0.0739,
+      "step": 49050
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.549435094463457e-05,
+      "loss": 0.1974,
+      "step": 49055
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.549388214335943e-05,
+      "loss": 0.2498,
+      "step": 49060
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.549341334208429e-05,
+      "loss": 0.3423,
+      "step": 49065
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.5492944540809153e-05,
+      "loss": 0.0346,
+      "step": 49070
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.5492475739534013e-05,
+      "loss": 0.0509,
+      "step": 49075
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.5492006938258873e-05,
+      "loss": 0.0455,
+      "step": 49080
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.5491538136983733e-05,
+      "loss": 0.0597,
+      "step": 49085
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.5491069335708596e-05,
+      "loss": 0.0448,
+      "step": 49090
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.5490600534433456e-05,
+      "loss": 0.0904,
+      "step": 49095
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.5490131733158316e-05,
+      "loss": 0.1396,
+      "step": 49100
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.5489662931883176e-05,
+      "loss": 0.1287,
+      "step": 49105
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.5489194130608036e-05,
+      "loss": 0.2406,
+      "step": 49110
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.5488725329332896e-05,
+      "loss": 0.2453,
+      "step": 49115
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.5488256528057756e-05,
+      "loss": 0.0674,
+      "step": 49120
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.5487787726782616e-05,
+      "loss": 0.0443,
+      "step": 49125
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.548731892550748e-05,
+      "loss": 0.1009,
+      "step": 49130
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.548685012423234e-05,
+      "loss": 0.0738,
+      "step": 49135
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.54863813229572e-05,
+      "loss": 0.1236,
+      "step": 49140
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.548591252168206e-05,
+      "loss": 0.1246,
+      "step": 49145
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.5485443720406922e-05,
+      "loss": 0.117,
+      "step": 49150
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.5484974919131782e-05,
+      "loss": 0.1785,
+      "step": 49155
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.5484506117856642e-05,
+      "loss": 0.2126,
+      "step": 49160
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.5484037316581502e-05,
+      "loss": 0.3216,
+      "step": 49165
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.5483568515306362e-05,
+      "loss": 0.0795,
+      "step": 49170
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.5483099714031226e-05,
+      "loss": 0.021,
+      "step": 49175
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.5482630912756085e-05,
+      "loss": 0.045,
+      "step": 49180
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.5482162111480945e-05,
+      "loss": 0.0943,
+      "step": 49185
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.5481693310205805e-05,
+      "loss": 0.116,
+      "step": 49190
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.5481224508930665e-05,
+      "loss": 0.0777,
+      "step": 49195
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.5480755707655525e-05,
+      "loss": 0.1557,
+      "step": 49200
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.5480286906380385e-05,
+      "loss": 0.1894,
+      "step": 49205
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.5479818105105245e-05,
+      "loss": 0.317,
+      "step": 49210
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.547934930383011e-05,
+      "loss": 0.3218,
+      "step": 49215
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.5478880502554968e-05,
+      "loss": 0.0294,
+      "step": 49220
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.5478411701279828e-05,
+      "loss": 0.0354,
+      "step": 49225
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.547794290000469e-05,
+      "loss": 0.0338,
+      "step": 49230
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.547747409872955e-05,
+      "loss": 0.1213,
+      "step": 49235
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.547700529745441e-05,
+      "loss": 0.0969,
+      "step": 49240
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.547653649617927e-05,
+      "loss": 0.0449,
+      "step": 49245
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.547606769490413e-05,
+      "loss": 0.0821,
+      "step": 49250
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.547559889362899e-05,
+      "loss": 0.2227,
+      "step": 49255
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.547513009235385e-05,
+      "loss": 0.2439,
+      "step": 49260
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.547466129107871e-05,
+      "loss": 0.1495,
+      "step": 49265
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.5474192489803574e-05,
+      "loss": 0.034,
+      "step": 49270
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.5473723688528434e-05,
+      "loss": 0.0257,
+      "step": 49275
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.5473254887253294e-05,
+      "loss": 0.1189,
+      "step": 49280
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.5472786085978154e-05,
+      "loss": 0.0593,
+      "step": 49285
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.5472317284703014e-05,
+      "loss": 0.0709,
+      "step": 49290
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.5471848483427877e-05,
+      "loss": 0.1199,
+      "step": 49295
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.5471379682152737e-05,
+      "loss": 0.1563,
+      "step": 49300
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.5470910880877597e-05,
+      "loss": 0.1083,
+      "step": 49305
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.5470442079602457e-05,
+      "loss": 0.3005,
+      "step": 49310
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.546997327832732e-05,
+      "loss": 0.2143,
+      "step": 49315
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.546950447705218e-05,
+      "loss": 0.0602,
+      "step": 49320
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.546903567577704e-05,
+      "loss": 0.021,
+      "step": 49325
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.54685668745019e-05,
+      "loss": 0.1049,
+      "step": 49330
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.546809807322676e-05,
+      "loss": 0.1159,
+      "step": 49335
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.546762927195162e-05,
+      "loss": 0.0707,
+      "step": 49340
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.546716047067648e-05,
+      "loss": 0.1285,
+      "step": 49345
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.546669166940134e-05,
+      "loss": 0.154,
+      "step": 49350
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.54662228681262e-05,
+      "loss": 0.1687,
+      "step": 49355
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.5465754066851063e-05,
+      "loss": 0.3047,
+      "step": 49360
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.5465285265575923e-05,
+      "loss": 0.2414,
+      "step": 49365
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.5464816464300783e-05,
+      "loss": 0.0811,
+      "step": 49370
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.5464347663025647e-05,
+      "loss": 0.0398,
+      "step": 49375
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.5463878861750507e-05,
+      "loss": 0.071,
+      "step": 49380
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.5463410060475366e-05,
+      "loss": 0.0606,
+      "step": 49385
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.5462941259200226e-05,
+      "loss": 0.2161,
+      "step": 49390
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.5462472457925086e-05,
+      "loss": 0.1271,
+      "step": 49395
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.5462003656649946e-05,
+      "loss": 0.084,
+      "step": 49400
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.5461534855374806e-05,
+      "loss": 0.2245,
+      "step": 49405
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.546106605409967e-05,
+      "loss": 0.3096,
+      "step": 49410
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.546059725282453e-05,
+      "loss": 0.2058,
+      "step": 49415
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.546012845154939e-05,
+      "loss": 0.041,
+      "step": 49420
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.545965965027425e-05,
+      "loss": 0.0828,
+      "step": 49425
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.545919084899911e-05,
+      "loss": 0.0496,
+      "step": 49430
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.545872204772397e-05,
+      "loss": 0.0631,
+      "step": 49435
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.5458253246448832e-05,
+      "loss": 0.0788,
+      "step": 49440
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.5457784445173692e-05,
+      "loss": 0.1385,
+      "step": 49445
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.5457315643898552e-05,
+      "loss": 0.2284,
+      "step": 49450
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.5456846842623416e-05,
+      "loss": 0.2322,
+      "step": 49455
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.5456378041348276e-05,
+      "loss": 0.3846,
+      "step": 49460
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.5455909240073136e-05,
+      "loss": 0.2495,
+      "step": 49465
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.5455440438797995e-05,
+      "loss": 0.0653,
+      "step": 49470
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.5454971637522855e-05,
+      "loss": 0.0465,
+      "step": 49475
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.5454502836247715e-05,
+      "loss": 0.091,
+      "step": 49480
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.5454034034972575e-05,
+      "loss": 0.0576,
+      "step": 49485
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.5453565233697435e-05,
+      "loss": 0.0574,
+      "step": 49490
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.5453096432422295e-05,
+      "loss": 0.0683,
+      "step": 49495
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.545262763114716e-05,
+      "loss": 0.1513,
+      "step": 49500
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.545215882987202e-05,
+      "loss": 0.1737,
+      "step": 49505
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.545169002859688e-05,
+      "loss": 0.2737,
+      "step": 49510
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.5451221227321738e-05,
+      "loss": 0.2102,
+      "step": 49515
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.54507524260466e-05,
+      "loss": 0.0284,
+      "step": 49520
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.545028362477146e-05,
+      "loss": 0.0773,
+      "step": 49525
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.544981482349632e-05,
+      "loss": 0.0148,
+      "step": 49530
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.544934602222118e-05,
+      "loss": 0.062,
+      "step": 49535
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.544887722094604e-05,
+      "loss": 0.0813,
+      "step": 49540
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.5448408419670905e-05,
+      "loss": 0.0957,
+      "step": 49545
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.5447939618395765e-05,
+      "loss": 0.1741,
+      "step": 49550
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.5447470817120625e-05,
+      "loss": 0.1808,
+      "step": 49555
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.5447002015845484e-05,
+      "loss": 0.3045,
+      "step": 49560
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.5446533214570344e-05,
+      "loss": 0.2576,
+      "step": 49565
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.5446064413295204e-05,
+      "loss": 0.0866,
+      "step": 49570
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.5445595612020064e-05,
+      "loss": 0.0214,
+      "step": 49575
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.5445126810744928e-05,
+      "loss": 0.0444,
+      "step": 49580
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.5444658009469788e-05,
+      "loss": 0.046,
+      "step": 49585
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.5444189208194647e-05,
+      "loss": 0.1132,
+      "step": 49590
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.5443720406919507e-05,
+      "loss": 0.0736,
+      "step": 49595
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.544325160564437e-05,
+      "loss": 0.0877,
+      "step": 49600
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.544278280436923e-05,
+      "loss": 0.1461,
+      "step": 49605
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.544231400309409e-05,
+      "loss": 0.2255,
+      "step": 49610
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.544184520181895e-05,
+      "loss": 0.2188,
+      "step": 49615
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.544137640054381e-05,
+      "loss": 0.0241,
+      "step": 49620
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.544090759926867e-05,
+      "loss": 0.0118,
+      "step": 49625
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.544043879799353e-05,
+      "loss": 0.0784,
+      "step": 49630
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.543996999671839e-05,
+      "loss": 0.1511,
+      "step": 49635
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.5439501195443254e-05,
+      "loss": 0.1046,
+      "step": 49640
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.5439032394168113e-05,
+      "loss": 0.1218,
+      "step": 49645
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.5438563592892973e-05,
+      "loss": 0.1129,
+      "step": 49650
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.5438094791617833e-05,
+      "loss": 0.1344,
+      "step": 49655
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.5437625990342697e-05,
+      "loss": 0.2268,
+      "step": 49660
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.5437157189067557e-05,
+      "loss": 0.3166,
+      "step": 49665
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.5436688387792417e-05,
+      "loss": 0.0485,
+      "step": 49670
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.5436219586517276e-05,
+      "loss": 0.0671,
+      "step": 49675
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.5435750785242136e-05,
+      "loss": 0.059,
+      "step": 49680
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.5435281983967e-05,
+      "loss": 0.0861,
+      "step": 49685
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.543481318269186e-05,
+      "loss": 0.0956,
+      "step": 49690
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.543434438141672e-05,
+      "loss": 0.1467,
+      "step": 49695
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.543387558014158e-05,
+      "loss": 0.1193,
+      "step": 49700
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.543340677886644e-05,
+      "loss": 0.1714,
+      "step": 49705
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.54329379775913e-05,
+      "loss": 0.229,
+      "step": 49710
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.543246917631616e-05,
+      "loss": 0.2369,
+      "step": 49715
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.543200037504102e-05,
+      "loss": 0.0291,
+      "step": 49720
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.5431531573765883e-05,
+      "loss": 0.0872,
+      "step": 49725
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.5431062772490743e-05,
+      "loss": 0.1216,
+      "step": 49730
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.5430593971215602e-05,
+      "loss": 0.0756,
+      "step": 49735
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.5430125169940466e-05,
+      "loss": 0.1129,
+      "step": 49740
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.5429656368665326e-05,
+      "loss": 0.1254,
+      "step": 49745
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.5429187567390186e-05,
+      "loss": 0.1671,
+      "step": 49750
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.5428718766115046e-05,
+      "loss": 0.1629,
+      "step": 49755
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.5428249964839906e-05,
+      "loss": 0.2814,
+      "step": 49760
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.5427781163564765e-05,
+      "loss": 0.2398,
+      "step": 49765
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.5427312362289625e-05,
+      "loss": 0.0732,
+      "step": 49770
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.5426843561014485e-05,
+      "loss": 0.0563,
+      "step": 49775
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.542637475973935e-05,
+      "loss": 0.0592,
+      "step": 49780
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.542590595846421e-05,
+      "loss": 0.0781,
+      "step": 49785
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.542543715718907e-05,
+      "loss": 0.0921,
+      "step": 49790
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.542496835591393e-05,
+      "loss": 0.1204,
+      "step": 49795
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.542449955463879e-05,
+      "loss": 0.1716,
+      "step": 49800
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.542403075336365e-05,
+      "loss": 0.208,
+      "step": 49805
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.542356195208851e-05,
+      "loss": 0.3305,
+      "step": 49810
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.542309315081337e-05,
+      "loss": 0.2823,
+      "step": 49815
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.542262434953823e-05,
+      "loss": 0.0389,
+      "step": 49820
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.5422155548263095e-05,
+      "loss": 0.0554,
+      "step": 49825
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.5421686746987955e-05,
+      "loss": 0.0942,
+      "step": 49830
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.5421217945712815e-05,
+      "loss": 0.0994,
+      "step": 49835
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.5420749144437675e-05,
+      "loss": 0.097,
+      "step": 49840
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.5420280343162535e-05,
+      "loss": 0.1022,
+      "step": 49845
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.5419811541887394e-05,
+      "loss": 0.0807,
+      "step": 49850
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.5419342740612254e-05,
+      "loss": 0.1672,
+      "step": 49855
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.5418873939337114e-05,
+      "loss": 0.2764,
+      "step": 49860
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.5418405138061974e-05,
+      "loss": 0.3174,
+      "step": 49865
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.5417936336786838e-05,
+      "loss": 0.1249,
+      "step": 49870
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.5417467535511698e-05,
+      "loss": 0.0402,
+      "step": 49875
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.5416998734236557e-05,
+      "loss": 0.0741,
+      "step": 49880
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.541652993296142e-05,
+      "loss": 0.1277,
+      "step": 49885
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.541606113168628e-05,
+      "loss": 0.1288,
+      "step": 49890
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.541559233041114e-05,
+      "loss": 0.1828,
+      "step": 49895
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.5415123529136e-05,
+      "loss": 0.1366,
+      "step": 49900
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.541465472786086e-05,
+      "loss": 0.2104,
+      "step": 49905
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.541418592658572e-05,
+      "loss": 0.1688,
+      "step": 49910
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.541371712531058e-05,
+      "loss": 0.2507,
+      "step": 49915
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.5413248324035444e-05,
+      "loss": 0.0244,
+      "step": 49920
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.5412779522760304e-05,
+      "loss": 0.0705,
+      "step": 49925
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.5412310721485164e-05,
+      "loss": 0.0406,
+      "step": 49930
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.5411841920210024e-05,
+      "loss": 0.0665,
+      "step": 49935
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.5411373118934883e-05,
+      "loss": 0.044,
+      "step": 49940
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.5410904317659743e-05,
+      "loss": 0.0675,
+      "step": 49945
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.5410435516384607e-05,
+      "loss": 0.0992,
+      "step": 49950
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.5409966715109467e-05,
+      "loss": 0.1772,
+      "step": 49955
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.5409497913834327e-05,
+      "loss": 0.1362,
+      "step": 49960
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.540902911255919e-05,
+      "loss": 0.2604,
+      "step": 49965
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.540856031128405e-05,
+      "loss": 0.0627,
+      "step": 49970
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.540809151000891e-05,
+      "loss": 0.0178,
+      "step": 49975
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.540762270873377e-05,
+      "loss": 0.0792,
+      "step": 49980
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.540715390745863e-05,
+      "loss": 0.0241,
+      "step": 49985
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.540668510618349e-05,
+      "loss": 0.1072,
+      "step": 49990
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.540621630490835e-05,
+      "loss": 0.0948,
+      "step": 49995
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.540574750363321e-05,
+      "loss": 0.1007,
+      "step": 50000
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.540527870235807e-05,
+      "loss": 0.2221,
+      "step": 50005
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.5404809901082933e-05,
+      "loss": 0.3763,
+      "step": 50010
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.5404341099807793e-05,
+      "loss": 0.2037,
+      "step": 50015
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.5403872298532653e-05,
+      "loss": 0.0519,
+      "step": 50020
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.5403403497257512e-05,
+      "loss": 0.0638,
+      "step": 50025
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.5402934695982376e-05,
+      "loss": 0.106,
+      "step": 50030
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.5402465894707236e-05,
+      "loss": 0.0515,
+      "step": 50035
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.5401997093432096e-05,
+      "loss": 0.0843,
+      "step": 50040
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.5401528292156956e-05,
+      "loss": 0.1201,
+      "step": 50045
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.5401059490881816e-05,
+      "loss": 0.1472,
+      "step": 50050
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.5400590689606675e-05,
+      "loss": 0.1796,
+      "step": 50055
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.540012188833154e-05,
+      "loss": 0.2752,
+      "step": 50060
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.53996530870564e-05,
+      "loss": 0.191,
+      "step": 50065
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.539918428578126e-05,
+      "loss": 0.023,
+      "step": 50070
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.539871548450612e-05,
+      "loss": 0.0786,
+      "step": 50075
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.539824668323098e-05,
+      "loss": 0.0499,
+      "step": 50080
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.539777788195584e-05,
+      "loss": 0.06,
+      "step": 50085
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.5397309080680702e-05,
+      "loss": 0.0837,
+      "step": 50090
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.5396840279405562e-05,
+      "loss": 0.1011,
+      "step": 50095
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.539637147813042e-05,
+      "loss": 0.2519,
+      "step": 50100
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.5395902676855285e-05,
+      "loss": 0.2367,
+      "step": 50105
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.5395433875580145e-05,
+      "loss": 0.2337,
+      "step": 50110
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.5394965074305005e-05,
+      "loss": 0.3048,
+      "step": 50115
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.5394496273029865e-05,
+      "loss": 0.1067,
+      "step": 50120
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.5394027471754725e-05,
+      "loss": 0.0296,
+      "step": 50125
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.5393558670479585e-05,
+      "loss": 0.0417,
+      "step": 50130
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.5393089869204445e-05,
+      "loss": 0.0882,
+      "step": 50135
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.5392621067929305e-05,
+      "loss": 0.1051,
+      "step": 50140
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.5392152266654164e-05,
+      "loss": 0.118,
+      "step": 50145
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.5391683465379028e-05,
+      "loss": 0.1182,
+      "step": 50150
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.5391214664103888e-05,
+      "loss": 0.1522,
+      "step": 50155
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.5390745862828748e-05,
+      "loss": 0.2011,
+      "step": 50160
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.5390277061553608e-05,
+      "loss": 0.2483,
+      "step": 50165
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.538980826027847e-05,
+      "loss": 0.0936,
+      "step": 50170
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.538933945900333e-05,
+      "loss": 0.0621,
+      "step": 50175
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.538887065772819e-05,
+      "loss": 0.0768,
+      "step": 50180
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.538840185645305e-05,
+      "loss": 0.1631,
+      "step": 50185
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.538793305517791e-05,
+      "loss": 0.0566,
+      "step": 50190
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.538746425390277e-05,
+      "loss": 0.1157,
+      "step": 50195
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.5386995452627634e-05,
+      "loss": 0.1238,
+      "step": 50200
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.5386526651352494e-05,
+      "loss": 0.1757,
+      "step": 50205
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.5386057850077354e-05,
+      "loss": 0.2642,
+      "step": 50210
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.5385589048802214e-05,
+      "loss": 0.3062,
+      "step": 50215
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.5385120247527074e-05,
+      "loss": 0.0431,
+      "step": 50220
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.5384651446251934e-05,
+      "loss": 0.0448,
+      "step": 50225
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.5384182644976793e-05,
+      "loss": 0.0536,
+      "step": 50230
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.5383713843701657e-05,
+      "loss": 0.0219,
+      "step": 50235
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.5383245042426517e-05,
+      "loss": 0.0811,
+      "step": 50240
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.5382776241151377e-05,
+      "loss": 0.124,
+      "step": 50245
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.538230743987624e-05,
+      "loss": 0.1731,
+      "step": 50250
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.53818386386011e-05,
+      "loss": 0.1248,
+      "step": 50255
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.538136983732596e-05,
+      "loss": 0.211,
+      "step": 50260
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.538090103605082e-05,
+      "loss": 0.2802,
+      "step": 50265
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.538043223477568e-05,
+      "loss": 0.0651,
+      "step": 50270
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.537996343350054e-05,
+      "loss": 0.0739,
+      "step": 50275
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.53794946322254e-05,
+      "loss": 0.0697,
+      "step": 50280
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.537902583095026e-05,
+      "loss": 0.066,
+      "step": 50285
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.5378557029675123e-05,
+      "loss": 0.0599,
+      "step": 50290
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.5378088228399983e-05,
+      "loss": 0.0544,
+      "step": 50295
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.5377619427124843e-05,
+      "loss": 0.0874,
+      "step": 50300
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.5377150625849703e-05,
+      "loss": 0.1567,
+      "step": 50305
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.5376681824574563e-05,
+      "loss": 0.2568,
+      "step": 50310
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.5376213023299426e-05,
+      "loss": 0.2765,
+      "step": 50315
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.5375744222024286e-05,
+      "loss": 0.0519,
+      "step": 50320
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.5375275420749146e-05,
+      "loss": 0.0839,
+      "step": 50325
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.5374806619474006e-05,
+      "loss": 0.0378,
+      "step": 50330
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.537433781819887e-05,
+      "loss": 0.0958,
+      "step": 50335
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.537386901692373e-05,
+      "loss": 0.0688,
+      "step": 50340
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.537340021564859e-05,
+      "loss": 0.1647,
+      "step": 50345
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.537293141437345e-05,
+      "loss": 0.1321,
+      "step": 50350
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.537246261309831e-05,
+      "loss": 0.1933,
+      "step": 50355
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.537199381182317e-05,
+      "loss": 0.2768,
+      "step": 50360
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.537152501054803e-05,
+      "loss": 0.3416,
+      "step": 50365
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.537105620927289e-05,
+      "loss": 0.0425,
+      "step": 50370
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.537058740799775e-05,
+      "loss": 0.0616,
+      "step": 50375
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.5370118606722612e-05,
+      "loss": 0.0403,
+      "step": 50380
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.5369649805447472e-05,
+      "loss": 0.0957,
+      "step": 50385
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.536918100417233e-05,
+      "loss": 0.0697,
+      "step": 50390
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.5368712202897195e-05,
+      "loss": 0.0837,
+      "step": 50395
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.5368243401622055e-05,
+      "loss": 0.1271,
+      "step": 50400
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.5367774600346915e-05,
+      "loss": 0.1135,
+      "step": 50405
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.5367305799071775e-05,
+      "loss": 0.2313,
+      "step": 50410
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.5366836997796635e-05,
+      "loss": 0.2247,
+      "step": 50415
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.5366368196521495e-05,
+      "loss": 0.0565,
+      "step": 50420
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.5365899395246355e-05,
+      "loss": 0.0344,
+      "step": 50425
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.5365430593971218e-05,
+      "loss": 0.0838,
+      "step": 50430
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.5364961792696078e-05,
+      "loss": 0.0572,
+      "step": 50435
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.5364492991420938e-05,
+      "loss": 0.0976,
+      "step": 50440
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.5364024190145798e-05,
+      "loss": 0.1598,
+      "step": 50445
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.5363555388870658e-05,
+      "loss": 0.1353,
+      "step": 50450
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.536308658759552e-05,
+      "loss": 0.2411,
+      "step": 50455
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.536261778632038e-05,
+      "loss": 0.2845,
+      "step": 50460
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.536214898504524e-05,
+      "loss": 0.3948,
+      "step": 50465
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.53616801837701e-05,
+      "loss": 0.0147,
+      "step": 50470
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.5361211382494964e-05,
+      "loss": 0.065,
+      "step": 50475
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.5360742581219824e-05,
+      "loss": 0.0895,
+      "step": 50480
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.5360273779944684e-05,
+      "loss": 0.0594,
+      "step": 50485
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.5359804978669544e-05,
+      "loss": 0.1196,
+      "step": 50490
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.5359336177394404e-05,
+      "loss": 0.1975,
+      "step": 50495
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.5358867376119264e-05,
+      "loss": 0.1263,
+      "step": 50500
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.5358398574844124e-05,
+      "loss": 0.2439,
+      "step": 50505
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.5357929773568984e-05,
+      "loss": 0.1744,
+      "step": 50510
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.5357460972293844e-05,
+      "loss": 0.2682,
+      "step": 50515
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.5356992171018707e-05,
+      "loss": 0.0798,
+      "step": 50520
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.5356523369743567e-05,
+      "loss": 0.025,
+      "step": 50525
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.5356054568468427e-05,
+      "loss": 0.0204,
+      "step": 50530
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.535558576719329e-05,
+      "loss": 0.0742,
+      "step": 50535
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.535511696591815e-05,
+      "loss": 0.0824,
+      "step": 50540
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.535464816464301e-05,
+      "loss": 0.1327,
+      "step": 50545
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.535417936336787e-05,
+      "loss": 0.0983,
+      "step": 50550
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.535371056209273e-05,
+      "loss": 0.1686,
+      "step": 50555
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.535324176081759e-05,
+      "loss": 0.5493,
+      "step": 50560
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.535277295954245e-05,
+      "loss": 0.2809,
+      "step": 50565
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.5352304158267313e-05,
+      "loss": 0.084,
+      "step": 50570
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.5351835356992173e-05,
+      "loss": 0.0162,
+      "step": 50575
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.5351366555717033e-05,
+      "loss": 0.1348,
+      "step": 50580
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.5350897754441893e-05,
+      "loss": 0.0913,
+      "step": 50585
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.5350428953166753e-05,
+      "loss": 0.0912,
+      "step": 50590
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.5349960151891613e-05,
+      "loss": 0.1417,
+      "step": 50595
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.5349491350616476e-05,
+      "loss": 0.0634,
+      "step": 50600
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.5349022549341336e-05,
+      "loss": 0.2053,
+      "step": 50605
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.5348553748066196e-05,
+      "loss": 0.3073,
+      "step": 50610
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.534808494679106e-05,
+      "loss": 0.2726,
+      "step": 50615
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.534761614551592e-05,
+      "loss": 0.0641,
+      "step": 50620
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.534714734424078e-05,
+      "loss": 0.0314,
+      "step": 50625
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.534667854296564e-05,
+      "loss": 0.0893,
+      "step": 50630
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.53462097416905e-05,
+      "loss": 0.062,
+      "step": 50635
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.534574094041536e-05,
+      "loss": 0.0985,
+      "step": 50640
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.534527213914022e-05,
+      "loss": 0.1125,
+      "step": 50645
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.534480333786508e-05,
+      "loss": 0.0805,
+      "step": 50650
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.534433453658994e-05,
+      "loss": 0.1704,
+      "step": 50655
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.5343865735314802e-05,
+      "loss": 0.3254,
+      "step": 50660
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.5343396934039662e-05,
+      "loss": 0.1932,
+      "step": 50665
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.5342928132764522e-05,
+      "loss": 0.0345,
+      "step": 50670
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.5342459331489382e-05,
+      "loss": 0.0975,
+      "step": 50675
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.5341990530214245e-05,
+      "loss": 0.0179,
+      "step": 50680
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.5341521728939105e-05,
+      "loss": 0.1288,
+      "step": 50685
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.5341052927663965e-05,
+      "loss": 0.135,
+      "step": 50690
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.5340584126388825e-05,
+      "loss": 0.1499,
+      "step": 50695
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.5340115325113685e-05,
+      "loss": 0.1307,
+      "step": 50700
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.5339646523838545e-05,
+      "loss": 0.1184,
+      "step": 50705
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.5339177722563408e-05,
+      "loss": 0.2781,
+      "step": 50710
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.5338708921288268e-05,
+      "loss": 0.2641,
+      "step": 50715
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.5338240120013128e-05,
+      "loss": 0.0174,
+      "step": 50720
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.5337771318737988e-05,
+      "loss": 0.0609,
+      "step": 50725
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.5337302517462848e-05,
+      "loss": 0.0497,
+      "step": 50730
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.5336833716187708e-05,
+      "loss": 0.1138,
+      "step": 50735
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.5336364914912568e-05,
+      "loss": 0.0964,
+      "step": 50740
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.533589611363743e-05,
+      "loss": 0.0858,
+      "step": 50745
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.533542731236229e-05,
+      "loss": 0.1338,
+      "step": 50750
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.533495851108715e-05,
+      "loss": 0.0712,
+      "step": 50755
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.5334489709812014e-05,
+      "loss": 0.1415,
+      "step": 50760
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.5334020908536874e-05,
+      "loss": 0.3659,
+      "step": 50765
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.5333552107261734e-05,
+      "loss": 0.0337,
+      "step": 50770
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.5333083305986594e-05,
+      "loss": 0.0486,
+      "step": 50775
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.5332614504711454e-05,
+      "loss": 0.0634,
+      "step": 50780
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.5332145703436314e-05,
+      "loss": 0.0815,
+      "step": 50785
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.5331676902161174e-05,
+      "loss": 0.0665,
+      "step": 50790
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.5331208100886034e-05,
+      "loss": 0.0991,
+      "step": 50795
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.5330739299610897e-05,
+      "loss": 0.1177,
+      "step": 50800
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.5330270498335757e-05,
+      "loss": 0.2121,
+      "step": 50805
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.5329801697060617e-05,
+      "loss": 0.2237,
+      "step": 50810
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.5329332895785477e-05,
+      "loss": 0.3753,
+      "step": 50815
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.5328864094510337e-05,
+      "loss": 0.0427,
+      "step": 50820
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.53283952932352e-05,
+      "loss": 0.0655,
+      "step": 50825
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.532792649196006e-05,
+      "loss": 0.0963,
+      "step": 50830
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.532745769068492e-05,
+      "loss": 0.0541,
+      "step": 50835
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.532698888940978e-05,
+      "loss": 0.0904,
+      "step": 50840
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.532652008813464e-05,
+      "loss": 0.1261,
+      "step": 50845
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.5326051286859503e-05,
+      "loss": 0.0886,
+      "step": 50850
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.5325582485584363e-05,
+      "loss": 0.175,
+      "step": 50855
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.5325113684309223e-05,
+      "loss": 0.2556,
+      "step": 50860
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.5324644883034083e-05,
+      "loss": 0.1925,
+      "step": 50865
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.5324176081758943e-05,
+      "loss": 0.101,
+      "step": 50870
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.5323707280483803e-05,
+      "loss": 0.0538,
+      "step": 50875
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.5323238479208663e-05,
+      "loss": 0.0709,
+      "step": 50880
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.5322769677933526e-05,
+      "loss": 0.1137,
+      "step": 50885
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.5322300876658386e-05,
+      "loss": 0.0497,
+      "step": 50890
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.5321832075383246e-05,
+      "loss": 0.0955,
+      "step": 50895
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.5321363274108106e-05,
+      "loss": 0.0704,
+      "step": 50900
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.532089447283297e-05,
+      "loss": 0.101,
+      "step": 50905
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.532042567155783e-05,
+      "loss": 0.221,
+      "step": 50910
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.531995687028269e-05,
+      "loss": 0.3579,
+      "step": 50915
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.531948806900755e-05,
+      "loss": 0.051,
+      "step": 50920
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.531901926773241e-05,
+      "loss": 0.0345,
+      "step": 50925
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.531855046645727e-05,
+      "loss": 0.0612,
+      "step": 50930
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.531808166518213e-05,
+      "loss": 0.1305,
+      "step": 50935
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.5317612863906992e-05,
+      "loss": 0.0428,
+      "step": 50940
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.5317144062631852e-05,
+      "loss": 0.1188,
+      "step": 50945
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.5316675261356712e-05,
+      "loss": 0.1338,
+      "step": 50950
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.5316206460081572e-05,
+      "loss": 0.1286,
+      "step": 50955
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.5315737658806432e-05,
+      "loss": 0.2555,
+      "step": 50960
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.5315268857531295e-05,
+      "loss": 0.3541,
+      "step": 50965
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.5314800056256155e-05,
+      "loss": 0.0705,
+      "step": 50970
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.5314331254981015e-05,
+      "loss": 0.0908,
+      "step": 50975
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.5313862453705875e-05,
+      "loss": 0.0534,
+      "step": 50980
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.531339365243074e-05,
+      "loss": 0.1644,
+      "step": 50985
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.5312924851155598e-05,
+      "loss": 0.0605,
+      "step": 50990
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.5312456049880458e-05,
+      "loss": 0.1191,
+      "step": 50995
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.5311987248605318e-05,
+      "loss": 0.0936,
+      "step": 51000
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.5311518447330178e-05,
+      "loss": 0.2088,
+      "step": 51005
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.5311049646055038e-05,
+      "loss": 0.1665,
+      "step": 51010
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.5310580844779898e-05,
+      "loss": 0.3294,
+      "step": 51015
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.5310112043504758e-05,
+      "loss": 0.0376,
+      "step": 51020
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.5309643242229618e-05,
+      "loss": 0.0529,
+      "step": 51025
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.530917444095448e-05,
+      "loss": 0.0384,
+      "step": 51030
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.530870563967934e-05,
+      "loss": 0.0483,
+      "step": 51035
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.53082368384042e-05,
+      "loss": 0.0941,
+      "step": 51040
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.5307768037129064e-05,
+      "loss": 0.1144,
+      "step": 51045
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.5307299235853924e-05,
+      "loss": 0.1497,
+      "step": 51050
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.5306830434578784e-05,
+      "loss": 0.0734,
+      "step": 51055
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.5306361633303644e-05,
+      "loss": 0.3083,
+      "step": 51060
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.5305892832028504e-05,
+      "loss": 0.2441,
+      "step": 51065
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.5305424030753364e-05,
+      "loss": 0.0149,
+      "step": 51070
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.5304955229478224e-05,
+      "loss": 0.0412,
+      "step": 51075
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.5304486428203087e-05,
+      "loss": 0.0896,
+      "step": 51080
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.5304017626927947e-05,
+      "loss": 0.0703,
+      "step": 51085
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.5303548825652807e-05,
+      "loss": 0.0497,
+      "step": 51090
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.5303080024377667e-05,
+      "loss": 0.083,
+      "step": 51095
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.5302611223102527e-05,
+      "loss": 0.2109,
+      "step": 51100
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.5302142421827387e-05,
+      "loss": 0.2021,
+      "step": 51105
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.530167362055225e-05,
+      "loss": 0.3419,
+      "step": 51110
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.530120481927711e-05,
+      "loss": 0.3248,
+      "step": 51115
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.530073601800197e-05,
+      "loss": 0.096,
+      "step": 51120
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.5300267216726833e-05,
+      "loss": 0.064,
+      "step": 51125
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.5299798415451693e-05,
+      "loss": 0.0324,
+      "step": 51130
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.5299329614176553e-05,
+      "loss": 0.0749,
+      "step": 51135
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.5298860812901413e-05,
+      "loss": 0.0687,
+      "step": 51140
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.5298392011626273e-05,
+      "loss": 0.0519,
+      "step": 51145
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.5297923210351133e-05,
+      "loss": 0.1116,
+      "step": 51150
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.5297454409075993e-05,
+      "loss": 0.2051,
+      "step": 51155
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.5296985607800853e-05,
+      "loss": 0.1371,
+      "step": 51160
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.5296516806525713e-05,
+      "loss": 0.3723,
+      "step": 51165
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.5296048005250573e-05,
+      "loss": 0.0814,
+      "step": 51170
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.5295579203975436e-05,
+      "loss": 0.0467,
+      "step": 51175
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.5295110402700296e-05,
+      "loss": 0.0746,
+      "step": 51180
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.5294641601425156e-05,
+      "loss": 0.0812,
+      "step": 51185
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.529417280015002e-05,
+      "loss": 0.0522,
+      "step": 51190
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.529370399887488e-05,
+      "loss": 0.0694,
+      "step": 51195
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.529323519759974e-05,
+      "loss": 0.1579,
+      "step": 51200
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.52927663963246e-05,
+      "loss": 0.2847,
+      "step": 51205
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.529229759504946e-05,
+      "loss": 0.2189,
+      "step": 51210
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.529182879377432e-05,
+      "loss": 0.3801,
+      "step": 51215
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.5291359992499182e-05,
+      "loss": 0.0602,
+      "step": 51220
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.5290891191224042e-05,
+      "loss": 0.0781,
+      "step": 51225
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.5290422389948902e-05,
+      "loss": 0.0435,
+      "step": 51230
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.5289953588673762e-05,
+      "loss": 0.0791,
+      "step": 51235
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.5289484787398622e-05,
+      "loss": 0.1357,
+      "step": 51240
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.5289015986123482e-05,
+      "loss": 0.0868,
+      "step": 51245
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.5288547184848342e-05,
+      "loss": 0.1032,
+      "step": 51250
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.5288078383573205e-05,
+      "loss": 0.1667,
+      "step": 51255
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.5287609582298065e-05,
+      "loss": 0.3281,
+      "step": 51260
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.5287140781022925e-05,
+      "loss": 0.3135,
+      "step": 51265
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.528667197974779e-05,
+      "loss": 0.0099,
+      "step": 51270
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.528620317847265e-05,
+      "loss": 0.0385,
+      "step": 51275
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.5285734377197508e-05,
+      "loss": 0.0753,
+      "step": 51280
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.5285265575922368e-05,
+      "loss": 0.0719,
+      "step": 51285
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.5284796774647228e-05,
+      "loss": 0.0609,
+      "step": 51290
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.5284327973372088e-05,
+      "loss": 0.1466,
+      "step": 51295
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.5283859172096948e-05,
+      "loss": 0.1317,
+      "step": 51300
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.5283390370821808e-05,
+      "loss": 0.2492,
+      "step": 51305
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.528292156954667e-05,
+      "loss": 0.2003,
+      "step": 51310
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.528245276827153e-05,
+      "loss": 0.3231,
+      "step": 51315
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.528198396699639e-05,
+      "loss": 0.0387,
+      "step": 51320
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.528151516572125e-05,
+      "loss": 0.0167,
+      "step": 51325
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.528104636444611e-05,
+      "loss": 0.0384,
+      "step": 51330
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.5280577563170974e-05,
+      "loss": 0.0847,
+      "step": 51335
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.5280108761895834e-05,
+      "loss": 0.0887,
+      "step": 51340
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.5279639960620694e-05,
+      "loss": 0.1089,
+      "step": 51345
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.5279171159345554e-05,
+      "loss": 0.1063,
+      "step": 51350
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.5278702358070414e-05,
+      "loss": 0.2167,
+      "step": 51355
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.5278233556795277e-05,
+      "loss": 0.2051,
+      "step": 51360
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.5277764755520137e-05,
+      "loss": 0.2115,
+      "step": 51365
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.5277295954244997e-05,
+      "loss": 0.04,
+      "step": 51370
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.5276827152969857e-05,
+      "loss": 0.0965,
+      "step": 51375
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.5276358351694717e-05,
+      "loss": 0.0714,
+      "step": 51380
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.5275889550419577e-05,
+      "loss": 0.0565,
+      "step": 51385
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.5275420749144437e-05,
+      "loss": 0.0819,
+      "step": 51390
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.52749519478693e-05,
+      "loss": 0.1281,
+      "step": 51395
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.527448314659416e-05,
+      "loss": 0.1891,
+      "step": 51400
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.527401434531902e-05,
+      "loss": 0.2188,
+      "step": 51405
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.527354554404388e-05,
+      "loss": 0.2091,
+      "step": 51410
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.5273076742768743e-05,
+      "loss": 0.2994,
+      "step": 51415
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.5272607941493603e-05,
+      "loss": 0.0559,
+      "step": 51420
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.5272139140218463e-05,
+      "loss": 0.0707,
+      "step": 51425
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.5271670338943323e-05,
+      "loss": 0.0289,
+      "step": 51430
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.5271201537668183e-05,
+      "loss": 0.1167,
+      "step": 51435
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.5270732736393043e-05,
+      "loss": 0.0636,
+      "step": 51440
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.5270263935117903e-05,
+      "loss": 0.1197,
+      "step": 51445
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.5269795133842766e-05,
+      "loss": 0.1567,
+      "step": 51450
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.5269326332567626e-05,
+      "loss": 0.1569,
+      "step": 51455
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.5268857531292486e-05,
+      "loss": 0.1988,
+      "step": 51460
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.5268388730017346e-05,
+      "loss": 0.2972,
+      "step": 51465
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.5267919928742206e-05,
+      "loss": 0.0234,
+      "step": 51470
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.526745112746707e-05,
+      "loss": 0.0534,
+      "step": 51475
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.526698232619193e-05,
+      "loss": 0.1281,
+      "step": 51480
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.526651352491679e-05,
+      "loss": 0.0836,
+      "step": 51485
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.526604472364165e-05,
+      "loss": 0.0708,
+      "step": 51490
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.526557592236651e-05,
+      "loss": 0.0852,
+      "step": 51495
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.5265107121091372e-05,
+      "loss": 0.1376,
+      "step": 51500
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.5264638319816232e-05,
+      "loss": 0.1672,
+      "step": 51505
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.5264169518541092e-05,
+      "loss": 0.31,
+      "step": 51510
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.5263700717265952e-05,
+      "loss": 0.2633,
+      "step": 51515
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.5263231915990812e-05,
+      "loss": 0.0777,
+      "step": 51520
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.5262763114715672e-05,
+      "loss": 0.0532,
+      "step": 51525
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.5262294313440532e-05,
+      "loss": 0.0739,
+      "step": 51530
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.5261825512165392e-05,
+      "loss": 0.2191,
+      "step": 51535
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.5261356710890255e-05,
+      "loss": 0.0397,
+      "step": 51540
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.5260887909615115e-05,
+      "loss": 0.0308,
+      "step": 51545
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.5260419108339975e-05,
+      "loss": 0.1165,
+      "step": 51550
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.525995030706484e-05,
+      "loss": 0.1375,
+      "step": 51555
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.52594815057897e-05,
+      "loss": 0.3485,
+      "step": 51560
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.525901270451456e-05,
+      "loss": 0.4187,
+      "step": 51565
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.5258543903239418e-05,
+      "loss": 0.0766,
+      "step": 51570
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.5258075101964278e-05,
+      "loss": 0.0273,
+      "step": 51575
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.5257606300689138e-05,
+      "loss": 0.0801,
+      "step": 51580
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.5257137499413998e-05,
+      "loss": 0.0657,
+      "step": 51585
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.5256668698138861e-05,
+      "loss": 0.0234,
+      "step": 51590
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.5256199896863721e-05,
+      "loss": 0.1634,
+      "step": 51595
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.5255731095588581e-05,
+      "loss": 0.1124,
+      "step": 51600
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.5255262294313441e-05,
+      "loss": 0.2432,
+      "step": 51605
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.5254793493038303e-05,
+      "loss": 0.3987,
+      "step": 51610
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.5254324691763163e-05,
+      "loss": 0.2696,
+      "step": 51615
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.5253855890488023e-05,
+      "loss": 0.0878,
+      "step": 51620
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.5253387089212883e-05,
+      "loss": 0.0419,
+      "step": 51625
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.5252918287937744e-05,
+      "loss": 0.0225,
+      "step": 51630
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.5252449486662606e-05,
+      "loss": 0.0671,
+      "step": 51635
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.5251980685387466e-05,
+      "loss": 0.1203,
+      "step": 51640
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.5251511884112326e-05,
+      "loss": 0.0863,
+      "step": 51645
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.5251043082837187e-05,
+      "loss": 0.0896,
+      "step": 51650
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.5250574281562047e-05,
+      "loss": 0.179,
+      "step": 51655
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.5250105480286907e-05,
+      "loss": 0.1446,
+      "step": 51660
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.5249636679011767e-05,
+      "loss": 0.2598,
+      "step": 51665
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.5249167877736629e-05,
+      "loss": 0.0394,
+      "step": 51670
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.5248699076461489e-05,
+      "loss": 0.044,
+      "step": 51675
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.5248230275186349e-05,
+      "loss": 0.1548,
+      "step": 51680
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.524776147391121e-05,
+      "loss": 0.0878,
+      "step": 51685
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.5247292672636072e-05,
+      "loss": 0.1237,
+      "step": 51690
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.5246823871360932e-05,
+      "loss": 0.1141,
+      "step": 51695
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.5246355070085792e-05,
+      "loss": 0.0894,
+      "step": 51700
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.5245886268810652e-05,
+      "loss": 0.291,
+      "step": 51705
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.5245417467535513e-05,
+      "loss": 0.2835,
+      "step": 51710
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.5244948666260373e-05,
+      "loss": 0.3245,
+      "step": 51715
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.5244479864985233e-05,
+      "loss": 0.0313,
+      "step": 51720
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.5244011063710093e-05,
+      "loss": 0.0367,
+      "step": 51725
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.5243542262434957e-05,
+      "loss": 0.068,
+      "step": 51730
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.5243073461159816e-05,
+      "loss": 0.0796,
+      "step": 51735
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.5242604659884676e-05,
+      "loss": 0.0945,
+      "step": 51740
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.5242135858609536e-05,
+      "loss": 0.1982,
+      "step": 51745
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.5241667057334398e-05,
+      "loss": 0.1924,
+      "step": 51750
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.5241198256059258e-05,
+      "loss": 0.1331,
+      "step": 51755
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5240729454784118e-05,
+      "loss": 0.1925,
+      "step": 51760
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5240260653508978e-05,
+      "loss": 0.2428,
+      "step": 51765
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5239791852233838e-05,
+      "loss": 0.0137,
+      "step": 51770
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5239323050958701e-05,
+      "loss": 0.0859,
+      "step": 51775
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5238854249683561e-05,
+      "loss": 0.0738,
+      "step": 51780
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5238385448408421e-05,
+      "loss": 0.0828,
+      "step": 51785
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5237916647133283e-05,
+      "loss": 0.0725,
+      "step": 51790
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5237447845858142e-05,
+      "loss": 0.0752,
+      "step": 51795
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5236979044583002e-05,
+      "loss": 0.0732,
+      "step": 51800
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5236510243307862e-05,
+      "loss": 0.2306,
+      "step": 51805
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5236041442032722e-05,
+      "loss": 0.2798,
+      "step": 51810
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5235572640757584e-05,
+      "loss": 0.2878,
+      "step": 51815
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5235103839482444e-05,
+      "loss": 0.0305,
+      "step": 51820
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5234635038207305e-05,
+      "loss": 0.051,
+      "step": 51825
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5234166236932167e-05,
+      "loss": 0.0365,
+      "step": 51830
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5233697435657027e-05,
+      "loss": 0.0645,
+      "step": 51835
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5233228634381887e-05,
+      "loss": 0.1311,
+      "step": 51840
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5232759833106747e-05,
+      "loss": 0.1698,
+      "step": 51845
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5232291031831607e-05,
+      "loss": 0.0807,
+      "step": 51850
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5231822230556468e-05,
+      "loss": 0.1707,
+      "step": 51855
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5231353429281328e-05,
+      "loss": 0.192,
+      "step": 51860
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5230884628006188e-05,
+      "loss": 0.2625,
+      "step": 51865
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5230415826731052e-05,
+      "loss": 0.03,
+      "step": 51870
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5229947025455912e-05,
+      "loss": 0.0394,
+      "step": 51875
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5229478224180771e-05,
+      "loss": 0.0815,
+      "step": 51880
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5229009422905631e-05,
+      "loss": 0.0712,
+      "step": 51885
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5228540621630491e-05,
+      "loss": 0.0508,
+      "step": 51890
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5228071820355353e-05,
+      "loss": 0.1297,
+      "step": 51895
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5227603019080213e-05,
+      "loss": 0.1003,
+      "step": 51900
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5227134217805073e-05,
+      "loss": 0.2595,
+      "step": 51905
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5226665416529933e-05,
+      "loss": 0.2423,
+      "step": 51910
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5226196615254796e-05,
+      "loss": 0.3415,
+      "step": 51915
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5225727813979656e-05,
+      "loss": 0.0748,
+      "step": 51920
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5225259012704516e-05,
+      "loss": 0.0466,
+      "step": 51925
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5224790211429376e-05,
+      "loss": 0.1348,
+      "step": 51930
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5224321410154238e-05,
+      "loss": 0.0513,
+      "step": 51935
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5223852608879097e-05,
+      "loss": 0.0782,
+      "step": 51940
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5223383807603957e-05,
+      "loss": 0.1423,
+      "step": 51945
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5222915006328817e-05,
+      "loss": 0.0698,
+      "step": 51950
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5222446205053677e-05,
+      "loss": 0.1165,
+      "step": 51955
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.522197740377854e-05,
+      "loss": 0.1488,
+      "step": 51960
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.52215086025034e-05,
+      "loss": 0.2605,
+      "step": 51965
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.522103980122826e-05,
+      "loss": 0.0814,
+      "step": 51970
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5220570999953122e-05,
+      "loss": 0.0322,
+      "step": 51975
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5220102198677982e-05,
+      "loss": 0.05,
+      "step": 51980
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5219633397402842e-05,
+      "loss": 0.1171,
+      "step": 51985
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5219164596127702e-05,
+      "loss": 0.0913,
+      "step": 51990
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5218695794852562e-05,
+      "loss": 0.1017,
+      "step": 51995
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5218226993577423e-05,
+      "loss": 0.1502,
+      "step": 52000
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5217758192302283e-05,
+      "loss": 0.1181,
+      "step": 52005
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5217289391027145e-05,
+      "loss": 0.2336,
+      "step": 52010
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5216820589752007e-05,
+      "loss": 0.211,
+      "step": 52015
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5216351788476867e-05,
+      "loss": 0.0689,
+      "step": 52020
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5215882987201726e-05,
+      "loss": 0.0412,
+      "step": 52025
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5215414185926586e-05,
+      "loss": 0.0589,
+      "step": 52030
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5214945384651448e-05,
+      "loss": 0.0807,
+      "step": 52035
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5214476583376308e-05,
+      "loss": 0.1278,
+      "step": 52040
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5214007782101168e-05,
+      "loss": 0.0676,
+      "step": 52045
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5213538980826028e-05,
+      "loss": 0.0983,
+      "step": 52050
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5213070179550891e-05,
+      "loss": 0.1433,
+      "step": 52055
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5212601378275751e-05,
+      "loss": 0.2007,
+      "step": 52060
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5212132577000611e-05,
+      "loss": 0.2886,
+      "step": 52065
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5211663775725471e-05,
+      "loss": 0.034,
+      "step": 52070
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5211194974450333e-05,
+      "loss": 0.0571,
+      "step": 52075
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5210726173175193e-05,
+      "loss": 0.0658,
+      "step": 52080
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5210257371900052e-05,
+      "loss": 0.0228,
+      "step": 52085
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5209788570624912e-05,
+      "loss": 0.1038,
+      "step": 52090
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5209319769349772e-05,
+      "loss": 0.0917,
+      "step": 52095
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5208850968074636e-05,
+      "loss": 0.1264,
+      "step": 52100
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5208382166799496e-05,
+      "loss": 0.1084,
+      "step": 52105
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5207913365524356e-05,
+      "loss": 0.204,
+      "step": 52110
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5207444564249217e-05,
+      "loss": 0.3396,
+      "step": 52115
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5206975762974077e-05,
+      "loss": 0.0746,
+      "step": 52120
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5206506961698937e-05,
+      "loss": 0.0563,
+      "step": 52125
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5206038160423797e-05,
+      "loss": 0.0551,
+      "step": 52130
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5205569359148657e-05,
+      "loss": 0.0601,
+      "step": 52135
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5205100557873519e-05,
+      "loss": 0.1207,
+      "step": 52140
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5204631756598378e-05,
+      "loss": 0.1706,
+      "step": 52145
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.520416295532324e-05,
+      "loss": 0.1324,
+      "step": 52150
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5203694154048102e-05,
+      "loss": 0.1642,
+      "step": 52155
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5203225352772962e-05,
+      "loss": 0.2112,
+      "step": 52160
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5202756551497822e-05,
+      "loss": 0.2276,
+      "step": 52165
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5202287750222682e-05,
+      "loss": 0.0522,
+      "step": 52170
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5201818948947541e-05,
+      "loss": 0.0433,
+      "step": 52175
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5201350147672403e-05,
+      "loss": 0.0256,
+      "step": 52180
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.5200881346397263e-05,
+      "loss": 0.063,
+      "step": 52185
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.5200412545122123e-05,
+      "loss": 0.0976,
+      "step": 52190
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.5199943743846986e-05,
+      "loss": 0.1321,
+      "step": 52195
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.5199474942571846e-05,
+      "loss": 0.1091,
+      "step": 52200
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.5199006141296706e-05,
+      "loss": 0.1427,
+      "step": 52205
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.5198537340021566e-05,
+      "loss": 0.2313,
+      "step": 52210
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.5198068538746426e-05,
+      "loss": 0.2734,
+      "step": 52215
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.5197599737471288e-05,
+      "loss": 0.0068,
+      "step": 52220
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.5197130936196148e-05,
+      "loss": 0.0642,
+      "step": 52225
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.5196662134921007e-05,
+      "loss": 0.072,
+      "step": 52230
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.5196193333645867e-05,
+      "loss": 0.0945,
+      "step": 52235
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.519572453237073e-05,
+      "loss": 0.0911,
+      "step": 52240
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.519525573109559e-05,
+      "loss": 0.2037,
+      "step": 52245
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.519478692982045e-05,
+      "loss": 0.1951,
+      "step": 52250
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.519431812854531e-05,
+      "loss": 0.1587,
+      "step": 52255
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.5193849327270172e-05,
+      "loss": 0.2093,
+      "step": 52260
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.5193380525995032e-05,
+      "loss": 0.2805,
+      "step": 52265
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.5192911724719892e-05,
+      "loss": 0.0562,
+      "step": 52270
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.5192442923444752e-05,
+      "loss": 0.0569,
+      "step": 52275
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.5191974122169612e-05,
+      "loss": 0.1093,
+      "step": 52280
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.5191505320894475e-05,
+      "loss": 0.0803,
+      "step": 52285
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.5191036519619335e-05,
+      "loss": 0.0702,
+      "step": 52290
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.5190567718344195e-05,
+      "loss": 0.1036,
+      "step": 52295
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.5190098917069057e-05,
+      "loss": 0.157,
+      "step": 52300
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.5189630115793917e-05,
+      "loss": 0.171,
+      "step": 52305
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.5189161314518777e-05,
+      "loss": 0.3712,
+      "step": 52310
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.5188692513243637e-05,
+      "loss": 0.3013,
+      "step": 52315
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.5188223711968496e-05,
+      "loss": 0.0322,
+      "step": 52320
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.5187754910693358e-05,
+      "loss": 0.0464,
+      "step": 52325
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.5187286109418218e-05,
+      "loss": 0.0681,
+      "step": 52330
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.518681730814308e-05,
+      "loss": 0.0501,
+      "step": 52335
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.5186348506867941e-05,
+      "loss": 0.1006,
+      "step": 52340
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.5185879705592801e-05,
+      "loss": 0.1419,
+      "step": 52345
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.5185410904317661e-05,
+      "loss": 0.1124,
+      "step": 52350
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.5184942103042521e-05,
+      "loss": 0.094,
+      "step": 52355
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.5184473301767381e-05,
+      "loss": 0.2268,
+      "step": 52360
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.5184004500492243e-05,
+      "loss": 0.3197,
+      "step": 52365
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.5183535699217103e-05,
+      "loss": 0.0583,
+      "step": 52370
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.5183066897941963e-05,
+      "loss": 0.0275,
+      "step": 52375
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.5182598096666826e-05,
+      "loss": 0.0546,
+      "step": 52380
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.5182129295391686e-05,
+      "loss": 0.1263,
+      "step": 52385
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.5181660494116546e-05,
+      "loss": 0.0598,
+      "step": 52390
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.5181191692841406e-05,
+      "loss": 0.0843,
+      "step": 52395
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.5180722891566266e-05,
+      "loss": 0.1803,
+      "step": 52400
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.5180254090291127e-05,
+      "loss": 0.2682,
+      "step": 52405
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.5179785289015987e-05,
+      "loss": 0.1826,
+      "step": 52410
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.5179316487740847e-05,
+      "loss": 0.309,
+      "step": 52415
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.5178847686465707e-05,
+      "loss": 0.0619,
+      "step": 52420
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.517837888519057e-05,
+      "loss": 0.0193,
+      "step": 52425
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.517791008391543e-05,
+      "loss": 0.0525,
+      "step": 52430
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.517744128264029e-05,
+      "loss": 0.0579,
+      "step": 52435
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.517697248136515e-05,
+      "loss": 0.0894,
+      "step": 52440
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.5176503680090012e-05,
+      "loss": 0.1532,
+      "step": 52445
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.5176034878814872e-05,
+      "loss": 0.1447,
+      "step": 52450
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.5175566077539732e-05,
+      "loss": 0.1665,
+      "step": 52455
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.5175097276264592e-05,
+      "loss": 0.1983,
+      "step": 52460
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.5174628474989453e-05,
+      "loss": 0.2106,
+      "step": 52465
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.5174159673714313e-05,
+      "loss": 0.0597,
+      "step": 52470
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.5173690872439175e-05,
+      "loss": 0.0445,
+      "step": 52475
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.5173222071164035e-05,
+      "loss": 0.0628,
+      "step": 52480
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.5172753269888896e-05,
+      "loss": 0.0762,
+      "step": 52485
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.5172284468613756e-05,
+      "loss": 0.0737,
+      "step": 52490
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.5171815667338616e-05,
+      "loss": 0.0675,
+      "step": 52495
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.5171346866063476e-05,
+      "loss": 0.1033,
+      "step": 52500
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.5170878064788338e-05,
+      "loss": 0.1985,
+      "step": 52505
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.5170409263513198e-05,
+      "loss": 0.2309,
+      "step": 52510
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.5169940462238058e-05,
+      "loss": 0.2345,
+      "step": 52515
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.516947166096292e-05,
+      "loss": 0.0415,
+      "step": 52520
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.5169002859687781e-05,
+      "loss": 0.0588,
+      "step": 52525
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.516853405841264e-05,
+      "loss": 0.0386,
+      "step": 52530
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.51680652571375e-05,
+      "loss": 0.1693,
+      "step": 52535
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.516759645586236e-05,
+      "loss": 0.1481,
+      "step": 52540
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.5167127654587222e-05,
+      "loss": 0.0804,
+      "step": 52545
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.5166658853312082e-05,
+      "loss": 0.1041,
+      "step": 52550
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.5166190052036942e-05,
+      "loss": 0.1923,
+      "step": 52555
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.5165721250761802e-05,
+      "loss": 0.3416,
+      "step": 52560
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.5165252449486665e-05,
+      "loss": 0.4328,
+      "step": 52565
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.5164783648211525e-05,
+      "loss": 0.0259,
+      "step": 52570
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.5164314846936385e-05,
+      "loss": 0.0365,
+      "step": 52575
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.5163846045661245e-05,
+      "loss": 0.0338,
+      "step": 52580
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.5163377244386107e-05,
+      "loss": 0.0438,
+      "step": 52585
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.5162908443110967e-05,
+      "loss": 0.0815,
+      "step": 52590
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.5162439641835827e-05,
+      "loss": 0.1101,
+      "step": 52595
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.5161970840560687e-05,
+      "loss": 0.0867,
+      "step": 52600
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.5161502039285547e-05,
+      "loss": 0.2476,
+      "step": 52605
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.516103323801041e-05,
+      "loss": 0.2438,
+      "step": 52610
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.516056443673527e-05,
+      "loss": 0.3517,
+      "step": 52615
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.516009563546013e-05,
+      "loss": 0.075,
+      "step": 52620
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.5159626834184991e-05,
+      "loss": 0.0791,
+      "step": 52625
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.5159158032909851e-05,
+      "loss": 0.0481,
+      "step": 52630
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.5158689231634711e-05,
+      "loss": 0.129,
+      "step": 52635
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.5158220430359571e-05,
+      "loss": 0.1212,
+      "step": 52640
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.5157751629084431e-05,
+      "loss": 0.1333,
+      "step": 52645
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.5157282827809293e-05,
+      "loss": 0.1213,
+      "step": 52650
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.5156814026534153e-05,
+      "loss": 0.1455,
+      "step": 52655
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.5156345225259014e-05,
+      "loss": 0.1973,
+      "step": 52660
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.5155876423983876e-05,
+      "loss": 0.2,
+      "step": 52665
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.5155407622708736e-05,
+      "loss": 0.0601,
+      "step": 52670
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.5154938821433596e-05,
+      "loss": 0.0499,
+      "step": 52675
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.5154470020158456e-05,
+      "loss": 0.0689,
+      "step": 52680
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.5154001218883316e-05,
+      "loss": 0.0873,
+      "step": 52685
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.5153532417608177e-05,
+      "loss": 0.073,
+      "step": 52690
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.5153063616333037e-05,
+      "loss": 0.1049,
+      "step": 52695
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.5152594815057897e-05,
+      "loss": 0.128,
+      "step": 52700
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.515212601378276e-05,
+      "loss": 0.1296,
+      "step": 52705
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.515165721250762e-05,
+      "loss": 0.2877,
+      "step": 52710
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.515118841123248e-05,
+      "loss": 0.2938,
+      "step": 52715
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.515071960995734e-05,
+      "loss": 0.0638,
+      "step": 52720
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.51502508086822e-05,
+      "loss": 0.0505,
+      "step": 52725
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.5149782007407062e-05,
+      "loss": 0.057,
+      "step": 52730
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.5149313206131922e-05,
+      "loss": 0.0638,
+      "step": 52735
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.5148844404856782e-05,
+      "loss": 0.0365,
+      "step": 52740
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.5148375603581642e-05,
+      "loss": 0.0596,
+      "step": 52745
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.5147906802306505e-05,
+      "loss": 0.1375,
+      "step": 52750
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.5147438001031365e-05,
+      "loss": 0.1279,
+      "step": 52755
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.5146969199756225e-05,
+      "loss": 0.1905,
+      "step": 52760
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.5146500398481085e-05,
+      "loss": 0.2818,
+      "step": 52765
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.5146031597205946e-05,
+      "loss": 0.0624,
+      "step": 52770
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.5145562795930806e-05,
+      "loss": 0.0779,
+      "step": 52775
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.5145093994655666e-05,
+      "loss": 0.0439,
+      "step": 52780
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.5144625193380526e-05,
+      "loss": 0.0472,
+      "step": 52785
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.5144156392105386e-05,
+      "loss": 0.0525,
+      "step": 52790
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.5143687590830248e-05,
+      "loss": 0.0622,
+      "step": 52795
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.514321878955511e-05,
+      "loss": 0.1119,
+      "step": 52800
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.514274998827997e-05,
+      "loss": 0.261,
+      "step": 52805
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.5142281187004831e-05,
+      "loss": 0.3272,
+      "step": 52810
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.5141812385729691e-05,
+      "loss": 0.3198,
+      "step": 52815
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.514134358445455e-05,
+      "loss": 0.0372,
+      "step": 52820
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.514087478317941e-05,
+      "loss": 0.0189,
+      "step": 52825
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.514040598190427e-05,
+      "loss": 0.0368,
+      "step": 52830
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5139937180629132e-05,
+      "loss": 0.1019,
+      "step": 52835
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5139468379353992e-05,
+      "loss": 0.15,
+      "step": 52840
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5138999578078854e-05,
+      "loss": 0.1733,
+      "step": 52845
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5138530776803715e-05,
+      "loss": 0.0982,
+      "step": 52850
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5138061975528575e-05,
+      "loss": 0.0813,
+      "step": 52855
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5137593174253435e-05,
+      "loss": 0.247,
+      "step": 52860
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5137124372978295e-05,
+      "loss": 0.1989,
+      "step": 52865
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5136655571703155e-05,
+      "loss": 0.0529,
+      "step": 52870
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5136186770428017e-05,
+      "loss": 0.0296,
+      "step": 52875
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5135717969152877e-05,
+      "loss": 0.0404,
+      "step": 52880
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5135249167877737e-05,
+      "loss": 0.0876,
+      "step": 52885
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.51347803666026e-05,
+      "loss": 0.0909,
+      "step": 52890
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.513431156532746e-05,
+      "loss": 0.119,
+      "step": 52895
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.513384276405232e-05,
+      "loss": 0.1705,
+      "step": 52900
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.513337396277718e-05,
+      "loss": 0.1126,
+      "step": 52905
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.513290516150204e-05,
+      "loss": 0.1938,
+      "step": 52910
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5132436360226901e-05,
+      "loss": 0.3044,
+      "step": 52915
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5131967558951761e-05,
+      "loss": 0.0649,
+      "step": 52920
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5131498757676621e-05,
+      "loss": 0.0713,
+      "step": 52925
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5131029956401481e-05,
+      "loss": 0.0724,
+      "step": 52930
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5130561155126345e-05,
+      "loss": 0.0492,
+      "step": 52935
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5130092353851204e-05,
+      "loss": 0.1065,
+      "step": 52940
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5129623552576064e-05,
+      "loss": 0.1069,
+      "step": 52945
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5129154751300924e-05,
+      "loss": 0.0492,
+      "step": 52950
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5128685950025786e-05,
+      "loss": 0.159,
+      "step": 52955
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5128217148750646e-05,
+      "loss": 0.307,
+      "step": 52960
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5127748347475506e-05,
+      "loss": 0.2151,
+      "step": 52965
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5127279546200366e-05,
+      "loss": 0.0254,
+      "step": 52970
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5126810744925227e-05,
+      "loss": 0.067,
+      "step": 52975
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5126341943650087e-05,
+      "loss": 0.0801,
+      "step": 52980
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5125873142374949e-05,
+      "loss": 0.0605,
+      "step": 52985
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5125404341099809e-05,
+      "loss": 0.0755,
+      "step": 52990
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.512493553982467e-05,
+      "loss": 0.136,
+      "step": 52995
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.512446673854953e-05,
+      "loss": 0.1061,
+      "step": 53000
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.512399793727439e-05,
+      "loss": 0.1312,
+      "step": 53005
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.512352913599925e-05,
+      "loss": 0.3252,
+      "step": 53010
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5123060334724112e-05,
+      "loss": 0.3149,
+      "step": 53015
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5122591533448972e-05,
+      "loss": 0.0129,
+      "step": 53020
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5122122732173832e-05,
+      "loss": 0.0211,
+      "step": 53025
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5121653930898693e-05,
+      "loss": 0.041,
+      "step": 53030
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5121185129623555e-05,
+      "loss": 0.0847,
+      "step": 53035
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5120716328348415e-05,
+      "loss": 0.0698,
+      "step": 53040
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5120247527073275e-05,
+      "loss": 0.0816,
+      "step": 53045
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5119778725798135e-05,
+      "loss": 0.1465,
+      "step": 53050
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5119309924522996e-05,
+      "loss": 0.1993,
+      "step": 53055
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5118841123247856e-05,
+      "loss": 0.3018,
+      "step": 53060
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5118372321972716e-05,
+      "loss": 0.3261,
+      "step": 53065
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5117903520697576e-05,
+      "loss": 0.0392,
+      "step": 53070
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.511743471942244e-05,
+      "loss": 0.027,
+      "step": 53075
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.51169659181473e-05,
+      "loss": 0.0483,
+      "step": 53080
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.511649711687216e-05,
+      "loss": 0.0291,
+      "step": 53085
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.511602831559702e-05,
+      "loss": 0.1415,
+      "step": 53090
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5115559514321881e-05,
+      "loss": 0.0677,
+      "step": 53095
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5115090713046741e-05,
+      "loss": 0.1427,
+      "step": 53100
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5114621911771601e-05,
+      "loss": 0.1626,
+      "step": 53105
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5114153110496461e-05,
+      "loss": 0.2132,
+      "step": 53110
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.511368430922132e-05,
+      "loss": 0.4029,
+      "step": 53115
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5113215507946182e-05,
+      "loss": 0.0847,
+      "step": 53120
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5112746706671044e-05,
+      "loss": 0.0436,
+      "step": 53125
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5112277905395904e-05,
+      "loss": 0.0253,
+      "step": 53130
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5111809104120766e-05,
+      "loss": 0.0722,
+      "step": 53135
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5111340302845626e-05,
+      "loss": 0.117,
+      "step": 53140
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5110871501570485e-05,
+      "loss": 0.1382,
+      "step": 53145
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5110402700295345e-05,
+      "loss": 0.1069,
+      "step": 53150
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5109933899020205e-05,
+      "loss": 0.1258,
+      "step": 53155
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5109465097745067e-05,
+      "loss": 0.4169,
+      "step": 53160
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5108996296469927e-05,
+      "loss": 0.3047,
+      "step": 53165
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5108527495194789e-05,
+      "loss": 0.0683,
+      "step": 53170
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.510805869391965e-05,
+      "loss": 0.0628,
+      "step": 53175
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.510758989264451e-05,
+      "loss": 0.0626,
+      "step": 53180
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.510712109136937e-05,
+      "loss": 0.0793,
+      "step": 53185
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.510665229009423e-05,
+      "loss": 0.1138,
+      "step": 53190
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.510618348881909e-05,
+      "loss": 0.0797,
+      "step": 53195
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5105714687543952e-05,
+      "loss": 0.1331,
+      "step": 53200
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5105245886268811e-05,
+      "loss": 0.1266,
+      "step": 53205
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5104777084993671e-05,
+      "loss": 0.1997,
+      "step": 53210
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5104308283718535e-05,
+      "loss": 0.2617,
+      "step": 53215
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5103839482443395e-05,
+      "loss": 0.0412,
+      "step": 53220
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5103370681168255e-05,
+      "loss": 0.1357,
+      "step": 53225
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5102901879893115e-05,
+      "loss": 0.0339,
+      "step": 53230
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5102433078617974e-05,
+      "loss": 0.109,
+      "step": 53235
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5101964277342836e-05,
+      "loss": 0.1435,
+      "step": 53240
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5101495476067696e-05,
+      "loss": 0.1191,
+      "step": 53245
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5101026674792556e-05,
+      "loss": 0.1276,
+      "step": 53250
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5100557873517416e-05,
+      "loss": 0.2045,
+      "step": 53255
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.510008907224228e-05,
+      "loss": 0.2849,
+      "step": 53260
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5099620270967139e-05,
+      "loss": 0.3419,
+      "step": 53265
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5099151469691999e-05,
+      "loss": 0.0099,
+      "step": 53270
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5098682668416859e-05,
+      "loss": 0.0081,
+      "step": 53275
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.509821386714172e-05,
+      "loss": 0.0745,
+      "step": 53280
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.509774506586658e-05,
+      "loss": 0.1331,
+      "step": 53285
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.509727626459144e-05,
+      "loss": 0.147,
+      "step": 53290
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.50968074633163e-05,
+      "loss": 0.1196,
+      "step": 53295
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.509633866204116e-05,
+      "loss": 0.1508,
+      "step": 53300
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5095869860766022e-05,
+      "loss": 0.1784,
+      "step": 53305
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5095401059490884e-05,
+      "loss": 0.1141,
+      "step": 53310
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5094932258215744e-05,
+      "loss": 0.1613,
+      "step": 53315
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5094463456940605e-05,
+      "loss": 0.0906,
+      "step": 53320
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5093994655665465e-05,
+      "loss": 0.057,
+      "step": 53325
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5093525854390325e-05,
+      "loss": 0.0731,
+      "step": 53330
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5093057053115185e-05,
+      "loss": 0.0516,
+      "step": 53335
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5092588251840045e-05,
+      "loss": 0.0711,
+      "step": 53340
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5092119450564907e-05,
+      "loss": 0.0954,
+      "step": 53345
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5091650649289766e-05,
+      "loss": 0.0862,
+      "step": 53350
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5091181848014628e-05,
+      "loss": 0.2134,
+      "step": 53355
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.509071304673949e-05,
+      "loss": 0.2143,
+      "step": 53360
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.509024424546435e-05,
+      "loss": 0.2572,
+      "step": 53365
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.508977544418921e-05,
+      "loss": 0.039,
+      "step": 53370
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.508930664291407e-05,
+      "loss": 0.1016,
+      "step": 53375
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.508883784163893e-05,
+      "loss": 0.0638,
+      "step": 53380
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5088369040363791e-05,
+      "loss": 0.0594,
+      "step": 53385
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5087900239088651e-05,
+      "loss": 0.0642,
+      "step": 53390
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5087431437813511e-05,
+      "loss": 0.079,
+      "step": 53395
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5086962636538374e-05,
+      "loss": 0.1898,
+      "step": 53400
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5086493835263234e-05,
+      "loss": 0.178,
+      "step": 53405
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5086025033988094e-05,
+      "loss": 0.2495,
+      "step": 53410
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5085556232712954e-05,
+      "loss": 0.2828,
+      "step": 53415
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5085087431437814e-05,
+      "loss": 0.0302,
+      "step": 53420
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5084618630162676e-05,
+      "loss": 0.0251,
+      "step": 53425
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5084149828887536e-05,
+      "loss": 0.0749,
+      "step": 53430
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5083681027612395e-05,
+      "loss": 0.0388,
+      "step": 53435
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5083212226337255e-05,
+      "loss": 0.1141,
+      "step": 53440
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5082743425062117e-05,
+      "loss": 0.0893,
+      "step": 53445
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5082274623786979e-05,
+      "loss": 0.1481,
+      "step": 53450
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5081805822511839e-05,
+      "loss": 0.2765,
+      "step": 53455
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.50813370212367e-05,
+      "loss": 0.5149,
+      "step": 53460
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.508086821996156e-05,
+      "loss": 0.324,
+      "step": 53465
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.508039941868642e-05,
+      "loss": 0.0539,
+      "step": 53470
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.507993061741128e-05,
+      "loss": 0.066,
+      "step": 53475
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.507946181613614e-05,
+      "loss": 0.0415,
+      "step": 53480
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.5078993014861002e-05,
+      "loss": 0.0461,
+      "step": 53485
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.5078524213585862e-05,
+      "loss": 0.0759,
+      "step": 53490
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.5078055412310723e-05,
+      "loss": 0.1054,
+      "step": 53495
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.5077586611035585e-05,
+      "loss": 0.0651,
+      "step": 53500
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.5077117809760445e-05,
+      "loss": 0.2001,
+      "step": 53505
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.5076649008485305e-05,
+      "loss": 0.1741,
+      "step": 53510
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.5076180207210165e-05,
+      "loss": 0.1449,
+      "step": 53515
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.5075711405935025e-05,
+      "loss": 0.0408,
+      "step": 53520
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.5075242604659886e-05,
+      "loss": 0.0829,
+      "step": 53525
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.5074773803384746e-05,
+      "loss": 0.057,
+      "step": 53530
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.5074305002109606e-05,
+      "loss": 0.0364,
+      "step": 53535
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.507383620083447e-05,
+      "loss": 0.0862,
+      "step": 53540
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.507336739955933e-05,
+      "loss": 0.1627,
+      "step": 53545
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.507289859828419e-05,
+      "loss": 0.1607,
+      "step": 53550
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.5072429797009049e-05,
+      "loss": 0.2309,
+      "step": 53555
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.5071960995733909e-05,
+      "loss": 0.362,
+      "step": 53560
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.507149219445877e-05,
+      "loss": 0.257,
+      "step": 53565
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.507102339318363e-05,
+      "loss": 0.0627,
+      "step": 53570
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.507055459190849e-05,
+      "loss": 0.0498,
+      "step": 53575
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.507008579063335e-05,
+      "loss": 0.054,
+      "step": 53580
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.5069616989358214e-05,
+      "loss": 0.0815,
+      "step": 53585
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.5069148188083074e-05,
+      "loss": 0.055,
+      "step": 53590
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.5068679386807934e-05,
+      "loss": 0.1355,
+      "step": 53595
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.5068210585532794e-05,
+      "loss": 0.1415,
+      "step": 53600
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.5067741784257655e-05,
+      "loss": 0.1233,
+      "step": 53605
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.5067272982982515e-05,
+      "loss": 0.216,
+      "step": 53610
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.5066804181707375e-05,
+      "loss": 0.2266,
+      "step": 53615
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.5066335380432235e-05,
+      "loss": 0.0833,
+      "step": 53620
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.5065866579157095e-05,
+      "loss": 0.0612,
+      "step": 53625
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.5065397777881957e-05,
+      "loss": 0.0735,
+      "step": 53630
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.5064928976606818e-05,
+      "loss": 0.0573,
+      "step": 53635
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.5064460175331678e-05,
+      "loss": 0.2157,
+      "step": 53640
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.506399137405654e-05,
+      "loss": 0.126,
+      "step": 53645
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.50635225727814e-05,
+      "loss": 0.0667,
+      "step": 53650
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.506305377150626e-05,
+      "loss": 0.1685,
+      "step": 53655
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.506258497023112e-05,
+      "loss": 0.4227,
+      "step": 53660
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.506211616895598e-05,
+      "loss": 0.2431,
+      "step": 53665
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.5061647367680841e-05,
+      "loss": 0.0585,
+      "step": 53670
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.5061178566405701e-05,
+      "loss": 0.0575,
+      "step": 53675
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.5060709765130563e-05,
+      "loss": 0.0963,
+      "step": 53680
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.5060240963855424e-05,
+      "loss": 0.0607,
+      "step": 53685
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.5059772162580284e-05,
+      "loss": 0.0998,
+      "step": 53690
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.5059303361305144e-05,
+      "loss": 0.0586,
+      "step": 53695
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.5058834560030004e-05,
+      "loss": 0.1631,
+      "step": 53700
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.5058365758754864e-05,
+      "loss": 0.167,
+      "step": 53705
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.5057896957479726e-05,
+      "loss": 0.2672,
+      "step": 53710
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.5057428156204586e-05,
+      "loss": 0.2048,
+      "step": 53715
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.5056959354929446e-05,
+      "loss": 0.0415,
+      "step": 53720
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.5056490553654309e-05,
+      "loss": 0.0442,
+      "step": 53725
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.5056021752379169e-05,
+      "loss": 0.0594,
+      "step": 53730
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.5055552951104029e-05,
+      "loss": 0.1218,
+      "step": 53735
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.5055084149828889e-05,
+      "loss": 0.122,
+      "step": 53740
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.5054615348553749e-05,
+      "loss": 0.085,
+      "step": 53745
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.505414654727861e-05,
+      "loss": 0.2021,
+      "step": 53750
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.505367774600347e-05,
+      "loss": 0.1342,
+      "step": 53755
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.505320894472833e-05,
+      "loss": 0.2367,
+      "step": 53760
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.505274014345319e-05,
+      "loss": 0.3059,
+      "step": 53765
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.505227134217805e-05,
+      "loss": 0.0305,
+      "step": 53770
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.5051802540902913e-05,
+      "loss": 0.0275,
+      "step": 53775
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.5051333739627773e-05,
+      "loss": 0.035,
+      "step": 53780
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.5050864938352633e-05,
+      "loss": 0.0746,
+      "step": 53785
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.5050396137077495e-05,
+      "loss": 0.0952,
+      "step": 53790
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.5049927335802355e-05,
+      "loss": 0.1478,
+      "step": 53795
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.5049458534527215e-05,
+      "loss": 0.1021,
+      "step": 53800
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.5048989733252075e-05,
+      "loss": 0.2201,
+      "step": 53805
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.5048520931976936e-05,
+      "loss": 0.2871,
+      "step": 53810
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.5048052130701796e-05,
+      "loss": 0.2895,
+      "step": 53815
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.5047583329426658e-05,
+      "loss": 0.0503,
+      "step": 53820
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.5047114528151518e-05,
+      "loss": 0.0398,
+      "step": 53825
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.504664572687638e-05,
+      "loss": 0.069,
+      "step": 53830
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.504617692560124e-05,
+      "loss": 0.0884,
+      "step": 53835
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.50457081243261e-05,
+      "loss": 0.07,
+      "step": 53840
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.504523932305096e-05,
+      "loss": 0.1193,
+      "step": 53845
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.504477052177582e-05,
+      "loss": 0.1512,
+      "step": 53850
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.504430172050068e-05,
+      "loss": 0.1932,
+      "step": 53855
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.504383291922554e-05,
+      "loss": 0.2692,
+      "step": 53860
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.5043364117950402e-05,
+      "loss": 0.3341,
+      "step": 53865
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.5042895316675264e-05,
+      "loss": 0.0146,
+      "step": 53870
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.5042426515400124e-05,
+      "loss": 0.0379,
+      "step": 53875
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.5041957714124984e-05,
+      "loss": 0.0244,
+      "step": 53880
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.5041488912849844e-05,
+      "loss": 0.0709,
+      "step": 53885
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.5041020111574705e-05,
+      "loss": 0.1115,
+      "step": 53890
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.5040551310299565e-05,
+      "loss": 0.0665,
+      "step": 53895
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.5040082509024425e-05,
+      "loss": 0.0744,
+      "step": 53900
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.5039613707749285e-05,
+      "loss": 0.1328,
+      "step": 53905
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.5039144906474148e-05,
+      "loss": 0.2599,
+      "step": 53910
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.5038676105199008e-05,
+      "loss": 0.2966,
+      "step": 53915
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.5038207303923868e-05,
+      "loss": 0.0359,
+      "step": 53920
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.5037738502648728e-05,
+      "loss": 0.0412,
+      "step": 53925
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.503726970137359e-05,
+      "loss": 0.0839,
+      "step": 53930
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.503680090009845e-05,
+      "loss": 0.048,
+      "step": 53935
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.503633209882331e-05,
+      "loss": 0.0786,
+      "step": 53940
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.503586329754817e-05,
+      "loss": 0.1743,
+      "step": 53945
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.503539449627303e-05,
+      "loss": 0.0831,
+      "step": 53950
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.5034925694997891e-05,
+      "loss": 0.2851,
+      "step": 53955
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.5034456893722753e-05,
+      "loss": 0.3397,
+      "step": 53960
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.5033988092447613e-05,
+      "loss": 0.3113,
+      "step": 53965
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.5033519291172474e-05,
+      "loss": 0.0512,
+      "step": 53970
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.5033050489897334e-05,
+      "loss": 0.0442,
+      "step": 53975
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.5032581688622194e-05,
+      "loss": 0.0382,
+      "step": 53980
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.5032112887347054e-05,
+      "loss": 0.0607,
+      "step": 53985
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.5031644086071914e-05,
+      "loss": 0.1256,
+      "step": 53990
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.5031175284796776e-05,
+      "loss": 0.1335,
+      "step": 53995
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.5030706483521636e-05,
+      "loss": 0.1469,
+      "step": 54000
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.5030237682246497e-05,
+      "loss": 0.1371,
+      "step": 54005
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.5029768880971359e-05,
+      "loss": 0.1676,
+      "step": 54010
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.5029300079696219e-05,
+      "loss": 0.1999,
+      "step": 54015
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.5028831278421079e-05,
+      "loss": 0.0053,
+      "step": 54020
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.5028362477145939e-05,
+      "loss": 0.0534,
+      "step": 54025
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.5027893675870799e-05,
+      "loss": 0.0322,
+      "step": 54030
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.502742487459566e-05,
+      "loss": 0.1084,
+      "step": 54035
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.502695607332052e-05,
+      "loss": 0.1186,
+      "step": 54040
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.502648727204538e-05,
+      "loss": 0.0961,
+      "step": 54045
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.5026018470770244e-05,
+      "loss": 0.1099,
+      "step": 54050
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.5025549669495104e-05,
+      "loss": 0.2092,
+      "step": 54055
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.5025080868219963e-05,
+      "loss": 0.4047,
+      "step": 54060
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.5024612066944823e-05,
+      "loss": 0.2458,
+      "step": 54065
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.5024143265669683e-05,
+      "loss": 0.0388,
+      "step": 54070
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.5023674464394545e-05,
+      "loss": 0.0411,
+      "step": 54075
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.5023205663119405e-05,
+      "loss": 0.051,
+      "step": 54080
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.5022736861844265e-05,
+      "loss": 0.0488,
+      "step": 54085
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.5022268060569125e-05,
+      "loss": 0.0987,
+      "step": 54090
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.5021799259293985e-05,
+      "loss": 0.0812,
+      "step": 54095
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.5021330458018848e-05,
+      "loss": 0.142,
+      "step": 54100
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.5020861656743708e-05,
+      "loss": 0.1601,
+      "step": 54105
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.5020392855468568e-05,
+      "loss": 0.2332,
+      "step": 54110
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.501992405419343e-05,
+      "loss": 0.2225,
+      "step": 54115
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.501945525291829e-05,
+      "loss": 0.0234,
+      "step": 54120
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.501898645164315e-05,
+      "loss": 0.0471,
+      "step": 54125
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.501851765036801e-05,
+      "loss": 0.0303,
+      "step": 54130
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.501804884909287e-05,
+      "loss": 0.0284,
+      "step": 54135
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.501758004781773e-05,
+      "loss": 0.0837,
+      "step": 54140
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.5017111246542592e-05,
+      "loss": 0.1419,
+      "step": 54145
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.5016642445267452e-05,
+      "loss": 0.1755,
+      "step": 54150
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.5016173643992314e-05,
+      "loss": 0.1494,
+      "step": 54155
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.5015704842717174e-05,
+      "loss": 0.1468,
+      "step": 54160
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.5015236041442034e-05,
+      "loss": 0.2734,
+      "step": 54165
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.5014767240166894e-05,
+      "loss": 0.0356,
+      "step": 54170
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.5014298438891754e-05,
+      "loss": 0.0384,
+      "step": 54175
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.5013829637616615e-05,
+      "loss": 0.0356,
+      "step": 54180
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.5013360836341475e-05,
+      "loss": 0.0438,
+      "step": 54185
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.5012892035066337e-05,
+      "loss": 0.1356,
+      "step": 54190
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.5012423233791199e-05,
+      "loss": 0.0935,
+      "step": 54195
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.5011954432516059e-05,
+      "loss": 0.1811,
+      "step": 54200
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.5011485631240918e-05,
+      "loss": 0.1958,
+      "step": 54205
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.5011016829965778e-05,
+      "loss": 0.1376,
+      "step": 54210
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.5010548028690638e-05,
+      "loss": 0.229,
+      "step": 54215
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.50100792274155e-05,
+      "loss": 0.0747,
+      "step": 54220
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.500961042614036e-05,
+      "loss": 0.0206,
+      "step": 54225
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.500914162486522e-05,
+      "loss": 0.0575,
+      "step": 54230
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.5008672823590083e-05,
+      "loss": 0.063,
+      "step": 54235
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.5008204022314943e-05,
+      "loss": 0.0903,
+      "step": 54240
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.5007735221039803e-05,
+      "loss": 0.2531,
+      "step": 54245
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.5007266419764663e-05,
+      "loss": 0.1318,
+      "step": 54250
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.5006797618489523e-05,
+      "loss": 0.2507,
+      "step": 54255
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.5006328817214384e-05,
+      "loss": 0.2718,
+      "step": 54260
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.5005860015939244e-05,
+      "loss": 0.24,
+      "step": 54265
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.5005391214664104e-05,
+      "loss": 0.0388,
+      "step": 54270
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.5004922413388964e-05,
+      "loss": 0.0633,
+      "step": 54275
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.5004453612113826e-05,
+      "loss": 0.0199,
+      "step": 54280
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.5003984810838688e-05,
+      "loss": 0.0784,
+      "step": 54285
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.5003516009563547e-05,
+      "loss": 0.0515,
+      "step": 54290
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.5003047208288407e-05,
+      "loss": 0.197,
+      "step": 54295
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.5002578407013269e-05,
+      "loss": 0.146,
+      "step": 54300
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.5002109605738129e-05,
+      "loss": 0.1319,
+      "step": 54305
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.5001640804462989e-05,
+      "loss": 0.3043,
+      "step": 54310
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.5001172003187849e-05,
+      "loss": 0.3923,
+      "step": 54315
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.500070320191271e-05,
+      "loss": 0.053,
+      "step": 54320
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.500023440063757e-05,
+      "loss": 0.0313,
+      "step": 54325
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.4999765599362432e-05,
+      "loss": 0.066,
+      "step": 54330
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.4999296798087292e-05,
+      "loss": 0.0837,
+      "step": 54335
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.4998827996812154e-05,
+      "loss": 0.06,
+      "step": 54340
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.4998359195537014e-05,
+      "loss": 0.1305,
+      "step": 54345
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.4997890394261873e-05,
+      "loss": 0.1331,
+      "step": 54350
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.4997421592986733e-05,
+      "loss": 0.1286,
+      "step": 54355
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.4996952791711595e-05,
+      "loss": 0.2388,
+      "step": 54360
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.4996483990436455e-05,
+      "loss": 0.419,
+      "step": 54365
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.4996015189161315e-05,
+      "loss": 0.0119,
+      "step": 54370
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.4995546387886177e-05,
+      "loss": 0.0559,
+      "step": 54375
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.4995077586611038e-05,
+      "loss": 0.0893,
+      "step": 54380
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.4994608785335898e-05,
+      "loss": 0.0824,
+      "step": 54385
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.4994139984060758e-05,
+      "loss": 0.0721,
+      "step": 54390
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.4993671182785618e-05,
+      "loss": 0.1021,
+      "step": 54395
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.499320238151048e-05,
+      "loss": 0.1121,
+      "step": 54400
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.499273358023534e-05,
+      "loss": 0.1667,
+      "step": 54405
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.49922647789602e-05,
+      "loss": 0.2541,
+      "step": 54410
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.499179597768506e-05,
+      "loss": 0.3177,
+      "step": 54415
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.499132717640992e-05,
+      "loss": 0.0115,
+      "step": 54420
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.4990858375134783e-05,
+      "loss": 0.0391,
+      "step": 54425
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.4990389573859643e-05,
+      "loss": 0.0732,
+      "step": 54430
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.4989920772584503e-05,
+      "loss": 0.1121,
+      "step": 54435
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.4989451971309364e-05,
+      "loss": 0.1034,
+      "step": 54440
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.4988983170034224e-05,
+      "loss": 0.1104,
+      "step": 54445
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.4988514368759084e-05,
+      "loss": 0.0782,
+      "step": 54450
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.4988045567483944e-05,
+      "loss": 0.125,
+      "step": 54455
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.4987576766208804e-05,
+      "loss": 0.251,
+      "step": 54460
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.4987107964933665e-05,
+      "loss": 0.2208,
+      "step": 54465
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.4986639163658527e-05,
+      "loss": 0.0434,
+      "step": 54470
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.4986170362383387e-05,
+      "loss": 0.047,
+      "step": 54475
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.4985701561108249e-05,
+      "loss": 0.0701,
+      "step": 54480
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.4985232759833109e-05,
+      "loss": 0.0824,
+      "step": 54485
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.4984763958557969e-05,
+      "loss": 0.062,
+      "step": 54490
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.4984295157282828e-05,
+      "loss": 0.1074,
+      "step": 54495
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.4983826356007688e-05,
+      "loss": 0.2118,
+      "step": 54500
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.498335755473255e-05,
+      "loss": 0.1498,
+      "step": 54505
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.498288875345741e-05,
+      "loss": 0.1589,
+      "step": 54510
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.4982419952182272e-05,
+      "loss": 0.265,
+      "step": 54515
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.4981951150907133e-05,
+      "loss": 0.0316,
+      "step": 54520
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.4981482349631993e-05,
+      "loss": 0.0771,
+      "step": 54525
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.4981013548356853e-05,
+      "loss": 0.0136,
+      "step": 54530
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.4980544747081713e-05,
+      "loss": 0.0744,
+      "step": 54535
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.4980075945806573e-05,
+      "loss": 0.1031,
+      "step": 54540
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.4979607144531435e-05,
+      "loss": 0.0737,
+      "step": 54545
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.4979138343256295e-05,
+      "loss": 0.0639,
+      "step": 54550
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.4978669541981154e-05,
+      "loss": 0.143,
+      "step": 54555
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.4978200740706018e-05,
+      "loss": 0.2828,
+      "step": 54560
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.4977731939430878e-05,
+      "loss": 0.3214,
+      "step": 54565
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.4977263138155738e-05,
+      "loss": 0.0628,
+      "step": 54570
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.4976794336880598e-05,
+      "loss": 0.028,
+      "step": 54575
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.4976325535605458e-05,
+      "loss": 0.0796,
+      "step": 54580
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.4975856734330319e-05,
+      "loss": 0.086,
+      "step": 54585
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.4975387933055179e-05,
+      "loss": 0.0964,
+      "step": 54590
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.4974919131780039e-05,
+      "loss": 0.0622,
+      "step": 54595
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.4974450330504899e-05,
+      "loss": 0.1113,
+      "step": 54600
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.4973981529229759e-05,
+      "loss": 0.1587,
+      "step": 54605
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.4973512727954622e-05,
+      "loss": 0.215,
+      "step": 54610
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.4973043926679482e-05,
+      "loss": 0.2566,
+      "step": 54615
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.4972575125404342e-05,
+      "loss": 0.0598,
+      "step": 54620
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.4972106324129204e-05,
+      "loss": 0.0519,
+      "step": 54625
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.4971637522854064e-05,
+      "loss": 0.0432,
+      "step": 54630
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.4971168721578924e-05,
+      "loss": 0.0895,
+      "step": 54635
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.4970699920303784e-05,
+      "loss": 0.0863,
+      "step": 54640
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.4970231119028643e-05,
+      "loss": 0.2322,
+      "step": 54645
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.4969762317753505e-05,
+      "loss": 0.079,
+      "step": 54650
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.4969293516478367e-05,
+      "loss": 0.1671,
+      "step": 54655
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.4968824715203227e-05,
+      "loss": 0.1948,
+      "step": 54660
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.4968355913928088e-05,
+      "loss": 0.2194,
+      "step": 54665
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.4967887112652948e-05,
+      "loss": 0.0337,
+      "step": 54670
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.4967418311377808e-05,
+      "loss": 0.0521,
+      "step": 54675
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.4966949510102668e-05,
+      "loss": 0.0716,
+      "step": 54680
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.4966480708827528e-05,
+      "loss": 0.074,
+      "step": 54685
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.496601190755239e-05,
+      "loss": 0.0648,
+      "step": 54690
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.496554310627725e-05,
+      "loss": 0.0953,
+      "step": 54695
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.4965074305002111e-05,
+      "loss": 0.1457,
+      "step": 54700
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.4964605503726973e-05,
+      "loss": 0.0831,
+      "step": 54705
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.4964136702451833e-05,
+      "loss": 0.222,
+      "step": 54710
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.4963667901176693e-05,
+      "loss": 0.2152,
+      "step": 54715
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.4963199099901553e-05,
+      "loss": 0.0534,
+      "step": 54720
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.4962730298626413e-05,
+      "loss": 0.0514,
+      "step": 54725
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.4962261497351274e-05,
+      "loss": 0.0556,
+      "step": 54730
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.4961792696076134e-05,
+      "loss": 0.1183,
+      "step": 54735
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.4961323894800994e-05,
+      "loss": 0.0392,
+      "step": 54740
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.4960855093525854e-05,
+      "loss": 0.1177,
+      "step": 54745
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.4960386292250717e-05,
+      "loss": 0.2041,
+      "step": 54750
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.4959917490975577e-05,
+      "loss": 0.116,
+      "step": 54755
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.4959448689700437e-05,
+      "loss": 0.225,
+      "step": 54760
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.4958979888425297e-05,
+      "loss": 0.3145,
+      "step": 54765
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.4958511087150159e-05,
+      "loss": 0.1032,
+      "step": 54770
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.4958042285875019e-05,
+      "loss": 0.0883,
+      "step": 54775
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.4957573484599879e-05,
+      "loss": 0.0607,
+      "step": 54780
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.4957104683324739e-05,
+      "loss": 0.1201,
+      "step": 54785
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.49566358820496e-05,
+      "loss": 0.064,
+      "step": 54790
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.4956167080774462e-05,
+      "loss": 0.2079,
+      "step": 54795
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.4955698279499322e-05,
+      "loss": 0.143,
+      "step": 54800
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.4955229478224182e-05,
+      "loss": 0.1583,
+      "step": 54805
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.4954760676949043e-05,
+      "loss": 0.277,
+      "step": 54810
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.4954291875673903e-05,
+      "loss": 0.2445,
+      "step": 54815
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.4953823074398763e-05,
+      "loss": 0.068,
+      "step": 54820
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.4953354273123623e-05,
+      "loss": 0.0589,
+      "step": 54825
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.4952885471848485e-05,
+      "loss": 0.038,
+      "step": 54830
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.4952416670573345e-05,
+      "loss": 0.0265,
+      "step": 54835
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.4951947869298206e-05,
+      "loss": 0.1087,
+      "step": 54840
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.4951479068023066e-05,
+      "loss": 0.1027,
+      "step": 54845
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.4951010266747928e-05,
+      "loss": 0.1725,
+      "step": 54850
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.4950541465472788e-05,
+      "loss": 0.1339,
+      "step": 54855
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.4950072664197648e-05,
+      "loss": 0.33,
+      "step": 54860
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.4949603862922508e-05,
+      "loss": 0.2939,
+      "step": 54865
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.494913506164737e-05,
+      "loss": 0.0765,
+      "step": 54870
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.494866626037223e-05,
+      "loss": 0.0295,
+      "step": 54875
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.4948197459097089e-05,
+      "loss": 0.0551,
+      "step": 54880
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.4947728657821952e-05,
+      "loss": 0.0445,
+      "step": 54885
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.4947259856546812e-05,
+      "loss": 0.0828,
+      "step": 54890
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.4946791055271672e-05,
+      "loss": 0.0708,
+      "step": 54895
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.4946322253996532e-05,
+      "loss": 0.1411,
+      "step": 54900
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.4945853452721392e-05,
+      "loss": 0.0814,
+      "step": 54905
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.4945384651446254e-05,
+      "loss": 0.3506,
+      "step": 54910
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.4944915850171114e-05,
+      "loss": 0.265,
+      "step": 54915
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.4944447048895974e-05,
+      "loss": 0.0974,
+      "step": 54920
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.4943978247620834e-05,
+      "loss": 0.0545,
+      "step": 54925
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.4943509446345694e-05,
+      "loss": 0.0914,
+      "step": 54930
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.4943040645070557e-05,
+      "loss": 0.0748,
+      "step": 54935
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.4942571843795417e-05,
+      "loss": 0.0855,
+      "step": 54940
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.4942103042520277e-05,
+      "loss": 0.1053,
+      "step": 54945
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.4941634241245138e-05,
+      "loss": 0.0821,
+      "step": 54950
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.4941165439969998e-05,
+      "loss": 0.2162,
+      "step": 54955
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.4940696638694858e-05,
+      "loss": 0.1372,
+      "step": 54960
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.4940227837419718e-05,
+      "loss": 0.333,
+      "step": 54965
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.4939759036144578e-05,
+      "loss": 0.0484,
+      "step": 54970
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.493929023486944e-05,
+      "loss": 0.0849,
+      "step": 54975
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.4938821433594301e-05,
+      "loss": 0.0428,
+      "step": 54980
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.4938352632319161e-05,
+      "loss": 0.0314,
+      "step": 54985
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.4937883831044023e-05,
+      "loss": 0.0623,
+      "step": 54990
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.4937415029768883e-05,
+      "loss": 0.2235,
+      "step": 54995
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.4936946228493743e-05,
+      "loss": 0.1594,
+      "step": 55000
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.4936477427218603e-05,
+      "loss": 0.1248,
+      "step": 55005
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.4936008625943463e-05,
+      "loss": 0.2323,
+      "step": 55010
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.4935539824668324e-05,
+      "loss": 0.1702,
+      "step": 55015
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.4935071023393184e-05,
+      "loss": 0.0247,
+      "step": 55020
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.4934602222118046e-05,
+      "loss": 0.0557,
+      "step": 55025
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.4934133420842907e-05,
+      "loss": 0.0506,
+      "step": 55030
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.4933664619567767e-05,
+      "loss": 0.1097,
+      "step": 55035
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.4933195818292627e-05,
+      "loss": 0.0586,
+      "step": 55040
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.4932727017017487e-05,
+      "loss": 0.1202,
+      "step": 55045
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.4932258215742347e-05,
+      "loss": 0.141,
+      "step": 55050
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.4931789414467209e-05,
+      "loss": 0.1917,
+      "step": 55055
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.4931320613192069e-05,
+      "loss": 0.2892,
+      "step": 55060
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.4930851811916929e-05,
+      "loss": 0.1868,
+      "step": 55065
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.4930383010641789e-05,
+      "loss": 0.0403,
+      "step": 55070
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.4929914209366652e-05,
+      "loss": 0.051,
+      "step": 55075
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.4929445408091512e-05,
+      "loss": 0.0712,
+      "step": 55080
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.4928976606816372e-05,
+      "loss": 0.1299,
+      "step": 55085
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.4928507805541232e-05,
+      "loss": 0.1003,
+      "step": 55090
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.4928039004266093e-05,
+      "loss": 0.0991,
+      "step": 55095
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.4927570202990953e-05,
+      "loss": 0.0788,
+      "step": 55100
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.4927101401715813e-05,
+      "loss": 0.1653,
+      "step": 55105
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.4926632600440673e-05,
+      "loss": 0.2166,
+      "step": 55110
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.4926163799165533e-05,
+      "loss": 0.3222,
+      "step": 55115
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.4925694997890396e-05,
+      "loss": 0.0577,
+      "step": 55120
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.4925226196615256e-05,
+      "loss": 0.0373,
+      "step": 55125
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.4924757395340116e-05,
+      "loss": 0.0392,
+      "step": 55130
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.4924288594064978e-05,
+      "loss": 0.0796,
+      "step": 55135
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.4923819792789838e-05,
+      "loss": 0.0793,
+      "step": 55140
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.4923350991514698e-05,
+      "loss": 0.1843,
+      "step": 55145
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.4922882190239558e-05,
+      "loss": 0.1432,
+      "step": 55150
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.4922413388964418e-05,
+      "loss": 0.146,
+      "step": 55155
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.492194458768928e-05,
+      "loss": 0.2226,
+      "step": 55160
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.4921475786414141e-05,
+      "loss": 0.3435,
+      "step": 55165
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.4921006985139e-05,
+      "loss": 0.0359,
+      "step": 55170
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.4920538183863862e-05,
+      "loss": 0.0193,
+      "step": 55175
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.4920069382588722e-05,
+      "loss": 0.0916,
+      "step": 55180
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.4919600581313582e-05,
+      "loss": 0.0995,
+      "step": 55185
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.4919131780038442e-05,
+      "loss": 0.1484,
+      "step": 55190
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.4918662978763302e-05,
+      "loss": 0.0871,
+      "step": 55195
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.4918194177488164e-05,
+      "loss": 0.14,
+      "step": 55200
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.4917725376213024e-05,
+      "loss": 0.1125,
+      "step": 55205
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.4917256574937885e-05,
+      "loss": 0.2089,
+      "step": 55210
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.4916787773662747e-05,
+      "loss": 0.2097,
+      "step": 55215
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.4916318972387607e-05,
+      "loss": 0.0522,
+      "step": 55220
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.4915850171112467e-05,
+      "loss": 0.0457,
+      "step": 55225
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.4915381369837327e-05,
+      "loss": 0.0382,
+      "step": 55230
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.4914912568562188e-05,
+      "loss": 0.0416,
+      "step": 55235
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.4914443767287048e-05,
+      "loss": 0.1091,
+      "step": 55240
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.4913974966011908e-05,
+      "loss": 0.1649,
+      "step": 55245
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.4913506164736768e-05,
+      "loss": 0.1362,
+      "step": 55250
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.4913037363461628e-05,
+      "loss": 0.1366,
+      "step": 55255
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.4912568562186492e-05,
+      "loss": 0.2839,
+      "step": 55260
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.4912099760911351e-05,
+      "loss": 0.2657,
+      "step": 55265
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.4911630959636211e-05,
+      "loss": 0.0139,
+      "step": 55270
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.4911162158361073e-05,
+      "loss": 0.057,
+      "step": 55275
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.4910693357085933e-05,
+      "loss": 0.0532,
+      "step": 55280
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.4910224555810793e-05,
+      "loss": 0.0633,
+      "step": 55285
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.4909755754535653e-05,
+      "loss": 0.0604,
+      "step": 55290
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.4909286953260513e-05,
+      "loss": 0.1417,
+      "step": 55295
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.4908818151985374e-05,
+      "loss": 0.1108,
+      "step": 55300
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.4908349350710236e-05,
+      "loss": 0.158,
+      "step": 55305
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.4907880549435096e-05,
+      "loss": 0.2174,
+      "step": 55310
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.4907411748159958e-05,
+      "loss": 0.2398,
+      "step": 55315
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.4906942946884817e-05,
+      "loss": 0.0446,
+      "step": 55320
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.4906474145609677e-05,
+      "loss": 0.0344,
+      "step": 55325
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.4906005344334537e-05,
+      "loss": 0.0328,
+      "step": 55330
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.4905536543059397e-05,
+      "loss": 0.1214,
+      "step": 55335
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.4905067741784259e-05,
+      "loss": 0.0773,
+      "step": 55340
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.4904598940509119e-05,
+      "loss": 0.0828,
+      "step": 55345
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.490413013923398e-05,
+      "loss": 0.1193,
+      "step": 55350
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.4903661337958842e-05,
+      "loss": 0.1879,
+      "step": 55355
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.4903192536683702e-05,
+      "loss": 0.2597,
+      "step": 55360
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.4902723735408562e-05,
+      "loss": 0.243,
+      "step": 55365
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.4902254934133422e-05,
+      "loss": 0.1397,
+      "step": 55370
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.4901786132858282e-05,
+      "loss": 0.0523,
+      "step": 55375
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.4901317331583143e-05,
+      "loss": 0.0604,
+      "step": 55380
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.4900848530308003e-05,
+      "loss": 0.0544,
+      "step": 55385
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.4900379729032863e-05,
+      "loss": 0.067,
+      "step": 55390
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.4899910927757723e-05,
+      "loss": 0.1031,
+      "step": 55395
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.4899442126482587e-05,
+      "loss": 0.1228,
+      "step": 55400
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.4898973325207447e-05,
+      "loss": 0.1742,
+      "step": 55405
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.4898504523932306e-05,
+      "loss": 0.1916,
+      "step": 55410
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.4898035722657166e-05,
+      "loss": 0.2808,
+      "step": 55415
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.4897566921382028e-05,
+      "loss": 0.0512,
+      "step": 55420
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.4897098120106888e-05,
+      "loss": 0.0845,
+      "step": 55425
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.4896629318831748e-05,
+      "loss": 0.0709,
+      "step": 55430
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.4896160517556608e-05,
+      "loss": 0.0582,
+      "step": 55435
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.4895691716281468e-05,
+      "loss": 0.0797,
+      "step": 55440
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.4895222915006331e-05,
+      "loss": 0.0405,
+      "step": 55445
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.4894754113731191e-05,
+      "loss": 0.139,
+      "step": 55450
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.4894285312456051e-05,
+      "loss": 0.1857,
+      "step": 55455
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.4893816511180913e-05,
+      "loss": 0.2593,
+      "step": 55460
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.4893347709905773e-05,
+      "loss": 0.354,
+      "step": 55465
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.4892878908630632e-05,
+      "loss": 0.0158,
+      "step": 55470
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.4892410107355492e-05,
+      "loss": 0.0485,
+      "step": 55475
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.4891941306080352e-05,
+      "loss": 0.0488,
+      "step": 55480
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.4891472504805214e-05,
+      "loss": 0.0494,
+      "step": 55485
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.4891003703530076e-05,
+      "loss": 0.141,
+      "step": 55490
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.4890534902254935e-05,
+      "loss": 0.0894,
+      "step": 55495
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.4890066100979797e-05,
+      "loss": 0.1028,
+      "step": 55500
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.4889597299704657e-05,
+      "loss": 0.135,
+      "step": 55505
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.4889128498429517e-05,
+      "loss": 0.1806,
+      "step": 55510
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.4888659697154377e-05,
+      "loss": 0.2292,
+      "step": 55515
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.4888190895879237e-05,
+      "loss": 0.0453,
+      "step": 55520
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.4887722094604098e-05,
+      "loss": 0.0331,
+      "step": 55525
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.4887253293328958e-05,
+      "loss": 0.0538,
+      "step": 55530
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.488678449205382e-05,
+      "loss": 0.0925,
+      "step": 55535
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.4886315690778682e-05,
+      "loss": 0.0954,
+      "step": 55540
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.4885846889503542e-05,
+      "loss": 0.0866,
+      "step": 55545
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.4885378088228402e-05,
+      "loss": 0.1247,
+      "step": 55550
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.4884909286953261e-05,
+      "loss": 0.2373,
+      "step": 55555
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.4884440485678121e-05,
+      "loss": 0.2644,
+      "step": 55560
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.4883971684402983e-05,
+      "loss": 0.2973,
+      "step": 55565
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.4883502883127843e-05,
+      "loss": 0.0809,
+      "step": 55570
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.4883034081852703e-05,
+      "loss": 0.0438,
+      "step": 55575
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.4882565280577563e-05,
+      "loss": 0.0554,
+      "step": 55580
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.4882096479302426e-05,
+      "loss": 0.0847,
+      "step": 55585
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.4881627678027286e-05,
+      "loss": 0.1215,
+      "step": 55590
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.4881158876752146e-05,
+      "loss": 0.0465,
+      "step": 55595
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.4880690075477006e-05,
+      "loss": 0.1264,
+      "step": 55600
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.4880221274201868e-05,
+      "loss": 0.1865,
+      "step": 55605
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.4879752472926728e-05,
+      "loss": 0.2029,
+      "step": 55610
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4879283671651587e-05,
+      "loss": 0.2115,
+      "step": 55615
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4878814870376447e-05,
+      "loss": 0.0313,
+      "step": 55620
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4878346069101309e-05,
+      "loss": 0.0566,
+      "step": 55625
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.487787726782617e-05,
+      "loss": 0.0854,
+      "step": 55630
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.487740846655103e-05,
+      "loss": 0.0547,
+      "step": 55635
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.487693966527589e-05,
+      "loss": 0.0987,
+      "step": 55640
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4876470864000752e-05,
+      "loss": 0.0807,
+      "step": 55645
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4876002062725612e-05,
+      "loss": 0.1574,
+      "step": 55650
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4875533261450472e-05,
+      "loss": 0.1102,
+      "step": 55655
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4875064460175332e-05,
+      "loss": 0.3842,
+      "step": 55660
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4874595658900194e-05,
+      "loss": 0.2638,
+      "step": 55665
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4874126857625053e-05,
+      "loss": 0.0447,
+      "step": 55670
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4873658056349915e-05,
+      "loss": 0.0429,
+      "step": 55675
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4873189255074775e-05,
+      "loss": 0.056,
+      "step": 55680
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4872720453799637e-05,
+      "loss": 0.0471,
+      "step": 55685
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4872251652524497e-05,
+      "loss": 0.1042,
+      "step": 55690
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4871782851249357e-05,
+      "loss": 0.1077,
+      "step": 55695
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4871314049974216e-05,
+      "loss": 0.117,
+      "step": 55700
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4870845248699078e-05,
+      "loss": 0.1148,
+      "step": 55705
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4870376447423938e-05,
+      "loss": 0.2782,
+      "step": 55710
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4869907646148798e-05,
+      "loss": 0.2159,
+      "step": 55715
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4869438844873658e-05,
+      "loss": 0.0824,
+      "step": 55720
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4868970043598521e-05,
+      "loss": 0.0114,
+      "step": 55725
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4868501242323381e-05,
+      "loss": 0.0292,
+      "step": 55730
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4868032441048241e-05,
+      "loss": 0.0636,
+      "step": 55735
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4867563639773101e-05,
+      "loss": 0.0672,
+      "step": 55740
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4867094838497963e-05,
+      "loss": 0.1228,
+      "step": 55745
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4866626037222823e-05,
+      "loss": 0.1111,
+      "step": 55750
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4866157235947683e-05,
+      "loss": 0.1702,
+      "step": 55755
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4865688434672542e-05,
+      "loss": 0.1814,
+      "step": 55760
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4865219633397402e-05,
+      "loss": 0.2145,
+      "step": 55765
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4864750832122266e-05,
+      "loss": 0.0433,
+      "step": 55770
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4864282030847126e-05,
+      "loss": 0.0851,
+      "step": 55775
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4863813229571986e-05,
+      "loss": 0.0285,
+      "step": 55780
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4863344428296847e-05,
+      "loss": 0.084,
+      "step": 55785
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4862875627021707e-05,
+      "loss": 0.0675,
+      "step": 55790
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4862406825746567e-05,
+      "loss": 0.0788,
+      "step": 55795
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4861938024471427e-05,
+      "loss": 0.1191,
+      "step": 55800
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4861469223196287e-05,
+      "loss": 0.2021,
+      "step": 55805
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4861000421921149e-05,
+      "loss": 0.343,
+      "step": 55810
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.486053162064601e-05,
+      "loss": 0.2186,
+      "step": 55815
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.486006281937087e-05,
+      "loss": 0.0562,
+      "step": 55820
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4859594018095732e-05,
+      "loss": 0.0369,
+      "step": 55825
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4859125216820592e-05,
+      "loss": 0.0856,
+      "step": 55830
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4858656415545452e-05,
+      "loss": 0.1225,
+      "step": 55835
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4858187614270312e-05,
+      "loss": 0.124,
+      "step": 55840
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4857718812995172e-05,
+      "loss": 0.1764,
+      "step": 55845
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4857250011720033e-05,
+      "loss": 0.15,
+      "step": 55850
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4856781210444893e-05,
+      "loss": 0.1934,
+      "step": 55855
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4856312409169755e-05,
+      "loss": 0.3649,
+      "step": 55860
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4855843607894616e-05,
+      "loss": 0.3696,
+      "step": 55865
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4855374806619476e-05,
+      "loss": 0.0612,
+      "step": 55870
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4854906005344336e-05,
+      "loss": 0.0421,
+      "step": 55875
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4854437204069196e-05,
+      "loss": 0.0431,
+      "step": 55880
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4853968402794056e-05,
+      "loss": 0.0768,
+      "step": 55885
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4853499601518918e-05,
+      "loss": 0.0669,
+      "step": 55890
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4853030800243778e-05,
+      "loss": 0.1025,
+      "step": 55895
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4852561998968638e-05,
+      "loss": 0.1342,
+      "step": 55900
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4852093197693497e-05,
+      "loss": 0.1633,
+      "step": 55905
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.485162439641836e-05,
+      "loss": 0.2885,
+      "step": 55910
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.485115559514322e-05,
+      "loss": 0.2572,
+      "step": 55915
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.485068679386808e-05,
+      "loss": 0.0396,
+      "step": 55920
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.485021799259294e-05,
+      "loss": 0.1123,
+      "step": 55925
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4849749191317802e-05,
+      "loss": 0.0515,
+      "step": 55930
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4849280390042662e-05,
+      "loss": 0.0417,
+      "step": 55935
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4848811588767522e-05,
+      "loss": 0.0986,
+      "step": 55940
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4848342787492382e-05,
+      "loss": 0.0709,
+      "step": 55945
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4847873986217242e-05,
+      "loss": 0.1535,
+      "step": 55950
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4847405184942105e-05,
+      "loss": 0.1495,
+      "step": 55955
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4846936383666965e-05,
+      "loss": 0.2546,
+      "step": 55960
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4846467582391825e-05,
+      "loss": 0.3005,
+      "step": 55965
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4845998781116687e-05,
+      "loss": 0.0819,
+      "step": 55970
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4845529979841547e-05,
+      "loss": 0.06,
+      "step": 55975
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4845061178566407e-05,
+      "loss": 0.0428,
+      "step": 55980
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4844592377291267e-05,
+      "loss": 0.099,
+      "step": 55985
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4844123576016127e-05,
+      "loss": 0.0624,
+      "step": 55990
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4843654774740988e-05,
+      "loss": 0.0641,
+      "step": 55995
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.484318597346585e-05,
+      "loss": 0.1507,
+      "step": 56000
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.484271717219071e-05,
+      "loss": 0.128,
+      "step": 56005
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4842248370915571e-05,
+      "loss": 0.2025,
+      "step": 56010
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4841779569640431e-05,
+      "loss": 0.2703,
+      "step": 56015
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4841310768365291e-05,
+      "loss": 0.0541,
+      "step": 56020
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4840841967090151e-05,
+      "loss": 0.0346,
+      "step": 56025
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4840373165815011e-05,
+      "loss": 0.0098,
+      "step": 56030
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4839904364539873e-05,
+      "loss": 0.064,
+      "step": 56035
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4839435563264733e-05,
+      "loss": 0.1539,
+      "step": 56040
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4838966761989593e-05,
+      "loss": 0.141,
+      "step": 56045
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4838497960714456e-05,
+      "loss": 0.1373,
+      "step": 56050
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4838029159439316e-05,
+      "loss": 0.2292,
+      "step": 56055
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4837560358164176e-05,
+      "loss": 0.2279,
+      "step": 56060
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4837091556889036e-05,
+      "loss": 0.381,
+      "step": 56065
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4836622755613896e-05,
+      "loss": 0.063,
+      "step": 56070
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4836153954338757e-05,
+      "loss": 0.0523,
+      "step": 56075
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4835685153063617e-05,
+      "loss": 0.1318,
+      "step": 56080
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4835216351788477e-05,
+      "loss": 0.0529,
+      "step": 56085
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4834747550513337e-05,
+      "loss": 0.0833,
+      "step": 56090
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.48342787492382e-05,
+      "loss": 0.2288,
+      "step": 56095
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.483380994796306e-05,
+      "loss": 0.137,
+      "step": 56100
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.483334114668792e-05,
+      "loss": 0.1722,
+      "step": 56105
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.483287234541278e-05,
+      "loss": 0.3163,
+      "step": 56110
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4832403544137642e-05,
+      "loss": 0.4034,
+      "step": 56115
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4831934742862502e-05,
+      "loss": 0.0912,
+      "step": 56120
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4831465941587362e-05,
+      "loss": 0.0487,
+      "step": 56125
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4830997140312222e-05,
+      "loss": 0.0635,
+      "step": 56130
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4830528339037083e-05,
+      "loss": 0.065,
+      "step": 56135
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4830059537761945e-05,
+      "loss": 0.0522,
+      "step": 56140
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4829590736486805e-05,
+      "loss": 0.1437,
+      "step": 56145
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4829121935211665e-05,
+      "loss": 0.1129,
+      "step": 56150
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4828653133936526e-05,
+      "loss": 0.1928,
+      "step": 56155
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4828184332661386e-05,
+      "loss": 0.2616,
+      "step": 56160
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4827715531386246e-05,
+      "loss": 0.2916,
+      "step": 56165
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4827246730111106e-05,
+      "loss": 0.0677,
+      "step": 56170
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4826777928835968e-05,
+      "loss": 0.0552,
+      "step": 56175
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4826309127560828e-05,
+      "loss": 0.0957,
+      "step": 56180
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.482584032628569e-05,
+      "loss": 0.0683,
+      "step": 56185
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.482537152501055e-05,
+      "loss": 0.1212,
+      "step": 56190
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4824902723735411e-05,
+      "loss": 0.0701,
+      "step": 56195
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.482443392246027e-05,
+      "loss": 0.1366,
+      "step": 56200
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.482396512118513e-05,
+      "loss": 0.184,
+      "step": 56205
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.482349631990999e-05,
+      "loss": 0.1445,
+      "step": 56210
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4823027518634852e-05,
+      "loss": 0.2239,
+      "step": 56215
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4822558717359712e-05,
+      "loss": 0.0696,
+      "step": 56220
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4822089916084572e-05,
+      "loss": 0.0632,
+      "step": 56225
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4821621114809432e-05,
+      "loss": 0.0973,
+      "step": 56230
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4821152313534295e-05,
+      "loss": 0.0835,
+      "step": 56235
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4820683512259155e-05,
+      "loss": 0.0855,
+      "step": 56240
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4820214710984015e-05,
+      "loss": 0.065,
+      "step": 56245
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4819745909708875e-05,
+      "loss": 0.1209,
+      "step": 56250
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4819277108433737e-05,
+      "loss": 0.1289,
+      "step": 56255
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.4818808307158597e-05,
+      "loss": 0.1962,
+      "step": 56260
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.4818339505883457e-05,
+      "loss": 0.2531,
+      "step": 56265
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.4817870704608317e-05,
+      "loss": 0.0775,
+      "step": 56270
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.4817401903333177e-05,
+      "loss": 0.0633,
+      "step": 56275
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.481693310205804e-05,
+      "loss": 0.0361,
+      "step": 56280
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.48164643007829e-05,
+      "loss": 0.0948,
+      "step": 56285
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.481599549950776e-05,
+      "loss": 0.402,
+      "step": 56290
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.4815526698232621e-05,
+      "loss": 0.0857,
+      "step": 56295
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.4815057896957481e-05,
+      "loss": 0.0479,
+      "step": 56300
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.4814589095682341e-05,
+      "loss": 0.1922,
+      "step": 56305
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.4814120294407201e-05,
+      "loss": 0.2842,
+      "step": 56310
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.4813651493132061e-05,
+      "loss": 0.2176,
+      "step": 56315
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.4813182691856923e-05,
+      "loss": 0.0288,
+      "step": 56320
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.4812713890581784e-05,
+      "loss": 0.053,
+      "step": 56325
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.4812245089306644e-05,
+      "loss": 0.0439,
+      "step": 56330
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.4811776288031506e-05,
+      "loss": 0.0837,
+      "step": 56335
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.4811307486756366e-05,
+      "loss": 0.0933,
+      "step": 56340
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.4810838685481226e-05,
+      "loss": 0.0983,
+      "step": 56345
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.4810369884206086e-05,
+      "loss": 0.1216,
+      "step": 56350
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.4809901082930946e-05,
+      "loss": 0.2602,
+      "step": 56355
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.4809432281655807e-05,
+      "loss": 0.2467,
+      "step": 56360
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.4808963480380667e-05,
+      "loss": 0.2412,
+      "step": 56365
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.4808494679105527e-05,
+      "loss": 0.1279,
+      "step": 56370
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.480802587783039e-05,
+      "loss": 0.0306,
+      "step": 56375
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.480755707655525e-05,
+      "loss": 0.0568,
+      "step": 56380
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.480708827528011e-05,
+      "loss": 0.1136,
+      "step": 56385
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.480661947400497e-05,
+      "loss": 0.0458,
+      "step": 56390
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.480615067272983e-05,
+      "loss": 0.1037,
+      "step": 56395
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.4805681871454692e-05,
+      "loss": 0.1402,
+      "step": 56400
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.4805213070179552e-05,
+      "loss": 0.1404,
+      "step": 56405
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.4804744268904412e-05,
+      "loss": 0.3928,
+      "step": 56410
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.4804275467629272e-05,
+      "loss": 0.2752,
+      "step": 56415
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.4803806666354135e-05,
+      "loss": 0.0587,
+      "step": 56420
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.4803337865078995e-05,
+      "loss": 0.0282,
+      "step": 56425
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.4802869063803855e-05,
+      "loss": 0.0494,
+      "step": 56430
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.4802400262528715e-05,
+      "loss": 0.1203,
+      "step": 56435
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.4801931461253576e-05,
+      "loss": 0.0802,
+      "step": 56440
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.4801462659978436e-05,
+      "loss": 0.1469,
+      "step": 56445
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.4800993858703296e-05,
+      "loss": 0.1011,
+      "step": 56450
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.4800525057428156e-05,
+      "loss": 0.1856,
+      "step": 56455
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.4800056256153016e-05,
+      "loss": 0.2412,
+      "step": 56460
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.479958745487788e-05,
+      "loss": 0.2946,
+      "step": 56465
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.479911865360274e-05,
+      "loss": 0.0999,
+      "step": 56470
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.47986498523276e-05,
+      "loss": 0.0356,
+      "step": 56475
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.4798181051052461e-05,
+      "loss": 0.0473,
+      "step": 56480
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.4797712249777321e-05,
+      "loss": 0.0465,
+      "step": 56485
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.4797243448502181e-05,
+      "loss": 0.1163,
+      "step": 56490
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.479677464722704e-05,
+      "loss": 0.084,
+      "step": 56495
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.47963058459519e-05,
+      "loss": 0.123,
+      "step": 56500
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.4795837044676762e-05,
+      "loss": 0.2126,
+      "step": 56505
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.4795368243401624e-05,
+      "loss": 0.2022,
+      "step": 56510
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.4794899442126484e-05,
+      "loss": 0.2718,
+      "step": 56515
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.4794430640851346e-05,
+      "loss": 0.0692,
+      "step": 56520
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.4793961839576205e-05,
+      "loss": 0.0684,
+      "step": 56525
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.4793493038301065e-05,
+      "loss": 0.0628,
+      "step": 56530
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.4793024237025925e-05,
+      "loss": 0.0465,
+      "step": 56535
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.4792555435750785e-05,
+      "loss": 0.1655,
+      "step": 56540
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.4792086634475647e-05,
+      "loss": 0.1165,
+      "step": 56545
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.4791617833200507e-05,
+      "loss": 0.1698,
+      "step": 56550
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.4791149031925367e-05,
+      "loss": 0.0829,
+      "step": 56555
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.479068023065023e-05,
+      "loss": 0.1631,
+      "step": 56560
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.479021142937509e-05,
+      "loss": 0.2569,
+      "step": 56565
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.478974262809995e-05,
+      "loss": 0.0661,
+      "step": 56570
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.478927382682481e-05,
+      "loss": 0.0431,
+      "step": 56575
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.478880502554967e-05,
+      "loss": 0.0998,
+      "step": 56580
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.4788336224274531e-05,
+      "loss": 0.067,
+      "step": 56585
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.4787867422999391e-05,
+      "loss": 0.1911,
+      "step": 56590
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.4787398621724251e-05,
+      "loss": 0.076,
+      "step": 56595
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.4786929820449111e-05,
+      "loss": 0.095,
+      "step": 56600
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.4786461019173975e-05,
+      "loss": 0.1959,
+      "step": 56605
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.4785992217898835e-05,
+      "loss": 0.1887,
+      "step": 56610
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.4785523416623694e-05,
+      "loss": 0.3195,
+      "step": 56615
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.4785054615348554e-05,
+      "loss": 0.0381,
+      "step": 56620
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.4784585814073416e-05,
+      "loss": 0.0168,
+      "step": 56625
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.4784117012798276e-05,
+      "loss": 0.0497,
+      "step": 56630
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.4783648211523136e-05,
+      "loss": 0.0663,
+      "step": 56635
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.4783179410247996e-05,
+      "loss": 0.153,
+      "step": 56640
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.4782710608972857e-05,
+      "loss": 0.0553,
+      "step": 56645
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.4782241807697719e-05,
+      "loss": 0.0652,
+      "step": 56650
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.4781773006422579e-05,
+      "loss": 0.1328,
+      "step": 56655
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.478130420514744e-05,
+      "loss": 0.3094,
+      "step": 56660
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.47808354038723e-05,
+      "loss": 0.3157,
+      "step": 56665
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.478036660259716e-05,
+      "loss": 0.0119,
+      "step": 56670
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.477989780132202e-05,
+      "loss": 0.05,
+      "step": 56675
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.477942900004688e-05,
+      "loss": 0.0535,
+      "step": 56680
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4778960198771742e-05,
+      "loss": 0.0421,
+      "step": 56685
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4778491397496602e-05,
+      "loss": 0.0757,
+      "step": 56690
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4778022596221462e-05,
+      "loss": 0.0902,
+      "step": 56695
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4777553794946325e-05,
+      "loss": 0.1036,
+      "step": 56700
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4777084993671185e-05,
+      "loss": 0.0464,
+      "step": 56705
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4776616192396045e-05,
+      "loss": 0.3221,
+      "step": 56710
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4776147391120905e-05,
+      "loss": 0.2555,
+      "step": 56715
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4775678589845765e-05,
+      "loss": 0.0529,
+      "step": 56720
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4775209788570627e-05,
+      "loss": 0.0143,
+      "step": 56725
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4774740987295486e-05,
+      "loss": 0.0219,
+      "step": 56730
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4774272186020346e-05,
+      "loss": 0.0693,
+      "step": 56735
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4773803384745206e-05,
+      "loss": 0.0585,
+      "step": 56740
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.477333458347007e-05,
+      "loss": 0.1523,
+      "step": 56745
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.477286578219493e-05,
+      "loss": 0.1027,
+      "step": 56750
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.477239698091979e-05,
+      "loss": 0.1413,
+      "step": 56755
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.477192817964465e-05,
+      "loss": 0.3343,
+      "step": 56760
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4771459378369511e-05,
+      "loss": 0.2061,
+      "step": 56765
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4770990577094371e-05,
+      "loss": 0.0353,
+      "step": 56770
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4770521775819231e-05,
+      "loss": 0.0424,
+      "step": 56775
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4770052974544091e-05,
+      "loss": 0.0619,
+      "step": 56780
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.476958417326895e-05,
+      "loss": 0.0789,
+      "step": 56785
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4769115371993814e-05,
+      "loss": 0.1295,
+      "step": 56790
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4768646570718674e-05,
+      "loss": 0.0659,
+      "step": 56795
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4768177769443534e-05,
+      "loss": 0.133,
+      "step": 56800
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4767708968168396e-05,
+      "loss": 0.2318,
+      "step": 56805
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4767240166893256e-05,
+      "loss": 0.2537,
+      "step": 56810
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4766771365618116e-05,
+      "loss": 0.306,
+      "step": 56815
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4766302564342975e-05,
+      "loss": 0.0372,
+      "step": 56820
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4765833763067835e-05,
+      "loss": 0.0766,
+      "step": 56825
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4765364961792697e-05,
+      "loss": 0.0259,
+      "step": 56830
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4764896160517559e-05,
+      "loss": 0.0714,
+      "step": 56835
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4764427359242419e-05,
+      "loss": 0.0551,
+      "step": 56840
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.476395855796728e-05,
+      "loss": 0.158,
+      "step": 56845
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.476348975669214e-05,
+      "loss": 0.1384,
+      "step": 56850
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4763020955417e-05,
+      "loss": 0.2766,
+      "step": 56855
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.476255215414186e-05,
+      "loss": 0.3387,
+      "step": 56860
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.476208335286672e-05,
+      "loss": 0.2891,
+      "step": 56865
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4761614551591582e-05,
+      "loss": 0.0331,
+      "step": 56870
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4761145750316442e-05,
+      "loss": 0.0907,
+      "step": 56875
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4760676949041301e-05,
+      "loss": 0.0488,
+      "step": 56880
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4760208147766165e-05,
+      "loss": 0.062,
+      "step": 56885
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4759739346491025e-05,
+      "loss": 0.0947,
+      "step": 56890
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4759270545215885e-05,
+      "loss": 0.0958,
+      "step": 56895
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4758801743940745e-05,
+      "loss": 0.1018,
+      "step": 56900
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4758332942665604e-05,
+      "loss": 0.1602,
+      "step": 56905
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4757864141390466e-05,
+      "loss": 0.2821,
+      "step": 56910
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4757395340115326e-05,
+      "loss": 0.2309,
+      "step": 56915
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4756926538840186e-05,
+      "loss": 0.0593,
+      "step": 56920
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4756457737565046e-05,
+      "loss": 0.0374,
+      "step": 56925
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.475598893628991e-05,
+      "loss": 0.0436,
+      "step": 56930
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.475552013501477e-05,
+      "loss": 0.0978,
+      "step": 56935
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4755051333739629e-05,
+      "loss": 0.0841,
+      "step": 56940
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4754582532464489e-05,
+      "loss": 0.124,
+      "step": 56945
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.475411373118935e-05,
+      "loss": 0.1214,
+      "step": 56950
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.475364492991421e-05,
+      "loss": 0.11,
+      "step": 56955
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.475317612863907e-05,
+      "loss": 0.2251,
+      "step": 56960
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.475270732736393e-05,
+      "loss": 0.3817,
+      "step": 56965
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4752238526088792e-05,
+      "loss": 0.0172,
+      "step": 56970
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4751769724813654e-05,
+      "loss": 0.0308,
+      "step": 56975
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4751300923538514e-05,
+      "loss": 0.0851,
+      "step": 56980
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4750832122263374e-05,
+      "loss": 0.0944,
+      "step": 56985
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4750363320988235e-05,
+      "loss": 0.0882,
+      "step": 56990
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4749894519713095e-05,
+      "loss": 0.1397,
+      "step": 56995
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4749425718437955e-05,
+      "loss": 0.1041,
+      "step": 57000
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4748956917162815e-05,
+      "loss": 0.1524,
+      "step": 57005
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4748488115887677e-05,
+      "loss": 0.2294,
+      "step": 57010
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4748019314612537e-05,
+      "loss": 0.3339,
+      "step": 57015
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4747550513337397e-05,
+      "loss": 0.0251,
+      "step": 57020
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4747081712062258e-05,
+      "loss": 0.0389,
+      "step": 57025
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.474661291078712e-05,
+      "loss": 0.0649,
+      "step": 57030
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.474614410951198e-05,
+      "loss": 0.1636,
+      "step": 57035
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.474567530823684e-05,
+      "loss": 0.0941,
+      "step": 57040
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.47452065069617e-05,
+      "loss": 0.1066,
+      "step": 57045
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4744737705686561e-05,
+      "loss": 0.155,
+      "step": 57050
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4744268904411421e-05,
+      "loss": 0.196,
+      "step": 57055
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4743800103136281e-05,
+      "loss": 0.2468,
+      "step": 57060
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4743331301861141e-05,
+      "loss": 0.2952,
+      "step": 57065
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4742862500586004e-05,
+      "loss": 0.0832,
+      "step": 57070
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4742393699310864e-05,
+      "loss": 0.0328,
+      "step": 57075
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4741924898035724e-05,
+      "loss": 0.075,
+      "step": 57080
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4741456096760584e-05,
+      "loss": 0.0648,
+      "step": 57085
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4740987295485446e-05,
+      "loss": 0.0571,
+      "step": 57090
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4740518494210306e-05,
+      "loss": 0.0371,
+      "step": 57095
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4740049692935166e-05,
+      "loss": 0.0917,
+      "step": 57100
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4739580891660026e-05,
+      "loss": 0.1858,
+      "step": 57105
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4739112090384885e-05,
+      "loss": 0.2585,
+      "step": 57110
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4738643289109749e-05,
+      "loss": 0.2586,
+      "step": 57115
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4738174487834609e-05,
+      "loss": 0.0449,
+      "step": 57120
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4737705686559469e-05,
+      "loss": 0.0409,
+      "step": 57125
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.473723688528433e-05,
+      "loss": 0.0844,
+      "step": 57130
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.473676808400919e-05,
+      "loss": 0.1158,
+      "step": 57135
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.473629928273405e-05,
+      "loss": 0.0533,
+      "step": 57140
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.473583048145891e-05,
+      "loss": 0.0643,
+      "step": 57145
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.473536168018377e-05,
+      "loss": 0.1587,
+      "step": 57150
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4734892878908632e-05,
+      "loss": 0.1664,
+      "step": 57155
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4734424077633493e-05,
+      "loss": 0.3299,
+      "step": 57160
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4733955276358353e-05,
+      "loss": 0.2293,
+      "step": 57165
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4733486475083215e-05,
+      "loss": 0.057,
+      "step": 57170
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4733017673808075e-05,
+      "loss": 0.0548,
+      "step": 57175
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4732548872532935e-05,
+      "loss": 0.0323,
+      "step": 57180
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4732080071257795e-05,
+      "loss": 0.0656,
+      "step": 57185
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4731611269982655e-05,
+      "loss": 0.0821,
+      "step": 57190
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4731142468707516e-05,
+      "loss": 0.1533,
+      "step": 57195
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4730673667432376e-05,
+      "loss": 0.1175,
+      "step": 57200
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4730204866157236e-05,
+      "loss": 0.1819,
+      "step": 57205
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.47297360648821e-05,
+      "loss": 0.2317,
+      "step": 57210
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.472926726360696e-05,
+      "loss": 0.194,
+      "step": 57215
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.472879846233182e-05,
+      "loss": 0.0501,
+      "step": 57220
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.472832966105668e-05,
+      "loss": 0.061,
+      "step": 57225
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4727860859781539e-05,
+      "loss": 0.0412,
+      "step": 57230
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.47273920585064e-05,
+      "loss": 0.027,
+      "step": 57235
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.472692325723126e-05,
+      "loss": 0.1291,
+      "step": 57240
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.472645445595612e-05,
+      "loss": 0.0757,
+      "step": 57245
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.472598565468098e-05,
+      "loss": 0.1663,
+      "step": 57250
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4725516853405844e-05,
+      "loss": 0.2098,
+      "step": 57255
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4725048052130704e-05,
+      "loss": 0.2877,
+      "step": 57260
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4724579250855564e-05,
+      "loss": 0.3111,
+      "step": 57265
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4724110449580424e-05,
+      "loss": 0.1081,
+      "step": 57270
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4723641648305285e-05,
+      "loss": 0.04,
+      "step": 57275
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4723172847030145e-05,
+      "loss": 0.0321,
+      "step": 57280
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4722704045755005e-05,
+      "loss": 0.0997,
+      "step": 57285
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4722235244479865e-05,
+      "loss": 0.0482,
+      "step": 57290
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4721766443204725e-05,
+      "loss": 0.122,
+      "step": 57295
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4721297641929588e-05,
+      "loss": 0.0914,
+      "step": 57300
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4720828840654448e-05,
+      "loss": 0.1392,
+      "step": 57305
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4720360039379308e-05,
+      "loss": 0.2636,
+      "step": 57310
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.471989123810417e-05,
+      "loss": 0.401,
+      "step": 57315
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.471942243682903e-05,
+      "loss": 0.0342,
+      "step": 57320
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.471895363555389e-05,
+      "loss": 0.0521,
+      "step": 57325
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.471848483427875e-05,
+      "loss": 0.0773,
+      "step": 57330
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.471801603300361e-05,
+      "loss": 0.0623,
+      "step": 57335
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.4717547231728471e-05,
+      "loss": 0.0676,
+      "step": 57340
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.4717078430453331e-05,
+      "loss": 0.1184,
+      "step": 57345
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.4716609629178193e-05,
+      "loss": 0.1047,
+      "step": 57350
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.4716140827903054e-05,
+      "loss": 0.1898,
+      "step": 57355
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.4715672026627914e-05,
+      "loss": 0.3378,
+      "step": 57360
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.4715203225352774e-05,
+      "loss": 0.1942,
+      "step": 57365
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.4714734424077634e-05,
+      "loss": 0.0085,
+      "step": 57370
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.4714265622802494e-05,
+      "loss": 0.0643,
+      "step": 57375
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.4713796821527356e-05,
+      "loss": 0.0491,
+      "step": 57380
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.4713328020252216e-05,
+      "loss": 0.1337,
+      "step": 57385
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.4712859218977076e-05,
+      "loss": 0.099,
+      "step": 57390
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.4712390417701939e-05,
+      "loss": 0.1491,
+      "step": 57395
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.4711921616426799e-05,
+      "loss": 0.1278,
+      "step": 57400
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.4711452815151659e-05,
+      "loss": 0.2245,
+      "step": 57405
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.4710984013876519e-05,
+      "loss": 0.2875,
+      "step": 57410
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.4710515212601379e-05,
+      "loss": 0.2976,
+      "step": 57415
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.471004641132624e-05,
+      "loss": 0.0304,
+      "step": 57420
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.47095776100511e-05,
+      "loss": 0.0545,
+      "step": 57425
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.470910880877596e-05,
+      "loss": 0.061,
+      "step": 57430
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.470864000750082e-05,
+      "loss": 0.0573,
+      "step": 57435
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.4708171206225683e-05,
+      "loss": 0.1313,
+      "step": 57440
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.4707702404950543e-05,
+      "loss": 0.1579,
+      "step": 57445
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.4707233603675403e-05,
+      "loss": 0.067,
+      "step": 57450
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.4706764802400263e-05,
+      "loss": 0.1581,
+      "step": 57455
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.4706296001125125e-05,
+      "loss": 0.2488,
+      "step": 57460
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.4705827199849985e-05,
+      "loss": 0.2114,
+      "step": 57465
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.4705358398574845e-05,
+      "loss": 0.0693,
+      "step": 57470
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.4704889597299705e-05,
+      "loss": 0.031,
+      "step": 57475
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.4704420796024566e-05,
+      "loss": 0.0395,
+      "step": 57480
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.4703951994749428e-05,
+      "loss": 0.0701,
+      "step": 57485
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.4703483193474288e-05,
+      "loss": 0.0973,
+      "step": 57490
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.4703014392199148e-05,
+      "loss": 0.0761,
+      "step": 57495
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.470254559092401e-05,
+      "loss": 0.0594,
+      "step": 57500
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.470207678964887e-05,
+      "loss": 0.156,
+      "step": 57505
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.470160798837373e-05,
+      "loss": 0.2419,
+      "step": 57510
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.470113918709859e-05,
+      "loss": 0.2983,
+      "step": 57515
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.4700670385823451e-05,
+      "loss": 0.0448,
+      "step": 57520
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.470020158454831e-05,
+      "loss": 0.0478,
+      "step": 57525
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.469973278327317e-05,
+      "loss": 0.1021,
+      "step": 57530
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.4699263981998032e-05,
+      "loss": 0.0631,
+      "step": 57535
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.4698795180722894e-05,
+      "loss": 0.0271,
+      "step": 57540
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.4698326379447754e-05,
+      "loss": 0.1192,
+      "step": 57545
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.4697857578172614e-05,
+      "loss": 0.0927,
+      "step": 57550
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.4697388776897474e-05,
+      "loss": 0.2041,
+      "step": 57555
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.4696919975622335e-05,
+      "loss": 0.2522,
+      "step": 57560
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.4696451174347195e-05,
+      "loss": 0.2796,
+      "step": 57565
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.4695982373072055e-05,
+      "loss": 0.0444,
+      "step": 57570
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.4695513571796915e-05,
+      "loss": 0.0531,
+      "step": 57575
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.4695044770521779e-05,
+      "loss": 0.1158,
+      "step": 57580
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.4694575969246638e-05,
+      "loss": 0.0572,
+      "step": 57585
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.4694107167971498e-05,
+      "loss": 0.1109,
+      "step": 57590
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.4693638366696358e-05,
+      "loss": 0.0798,
+      "step": 57595
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.469316956542122e-05,
+      "loss": 0.1928,
+      "step": 57600
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.469270076414608e-05,
+      "loss": 0.2423,
+      "step": 57605
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.469223196287094e-05,
+      "loss": 0.2319,
+      "step": 57610
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.46917631615958e-05,
+      "loss": 0.2285,
+      "step": 57615
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.469129436032066e-05,
+      "loss": 0.0387,
+      "step": 57620
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.4690825559045523e-05,
+      "loss": 0.0568,
+      "step": 57625
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.4690356757770383e-05,
+      "loss": 0.0665,
+      "step": 57630
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.4689887956495243e-05,
+      "loss": 0.1037,
+      "step": 57635
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.4689419155220105e-05,
+      "loss": 0.0905,
+      "step": 57640
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.4688950353944964e-05,
+      "loss": 0.0775,
+      "step": 57645
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.4688481552669824e-05,
+      "loss": 0.1319,
+      "step": 57650
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.4688012751394684e-05,
+      "loss": 0.1594,
+      "step": 57655
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.4687543950119544e-05,
+      "loss": 0.3679,
+      "step": 57660
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.4687075148844406e-05,
+      "loss": 0.6374,
+      "step": 57665
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.4686606347569266e-05,
+      "loss": 0.0202,
+      "step": 57670
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.4686137546294127e-05,
+      "loss": 0.0683,
+      "step": 57675
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.4685668745018989e-05,
+      "loss": 0.0405,
+      "step": 57680
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.4685199943743849e-05,
+      "loss": 0.0402,
+      "step": 57685
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.4684731142468709e-05,
+      "loss": 0.1176,
+      "step": 57690
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.4684262341193569e-05,
+      "loss": 0.0638,
+      "step": 57695
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.4683793539918429e-05,
+      "loss": 0.1016,
+      "step": 57700
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.468332473864329e-05,
+      "loss": 0.0693,
+      "step": 57705
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.468285593736815e-05,
+      "loss": 0.1265,
+      "step": 57710
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.468238713609301e-05,
+      "loss": 0.2031,
+      "step": 57715
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.4681918334817874e-05,
+      "loss": 0.0728,
+      "step": 57720
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.4681449533542734e-05,
+      "loss": 0.0588,
+      "step": 57725
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.4680980732267593e-05,
+      "loss": 0.0832,
+      "step": 57730
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.4680511930992453e-05,
+      "loss": 0.177,
+      "step": 57735
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.4680043129717313e-05,
+      "loss": 0.1171,
+      "step": 57740
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.4679574328442175e-05,
+      "loss": 0.1549,
+      "step": 57745
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.4679105527167035e-05,
+      "loss": 0.1312,
+      "step": 57750
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.4678636725891895e-05,
+      "loss": 0.2211,
+      "step": 57755
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.4678167924616755e-05,
+      "loss": 0.3696,
+      "step": 57760
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.4677699123341618e-05,
+      "loss": 0.2266,
+      "step": 57765
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.4677230322066478e-05,
+      "loss": 0.0231,
+      "step": 57770
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.4676761520791338e-05,
+      "loss": 0.029,
+      "step": 57775
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.4676292719516198e-05,
+      "loss": 0.0498,
+      "step": 57780
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.467582391824106e-05,
+      "loss": 0.0852,
+      "step": 57785
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.467535511696592e-05,
+      "loss": 0.0683,
+      "step": 57790
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.467488631569078e-05,
+      "loss": 0.0996,
+      "step": 57795
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.467441751441564e-05,
+      "loss": 0.1235,
+      "step": 57800
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.46739487131405e-05,
+      "loss": 0.1227,
+      "step": 57805
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.4673479911865363e-05,
+      "loss": 0.1879,
+      "step": 57810
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.4673011110590223e-05,
+      "loss": 0.2279,
+      "step": 57815
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.4672542309315082e-05,
+      "loss": 0.0929,
+      "step": 57820
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.4672073508039944e-05,
+      "loss": 0.0051,
+      "step": 57825
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.4671604706764804e-05,
+      "loss": 0.035,
+      "step": 57830
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.4671135905489664e-05,
+      "loss": 0.0929,
+      "step": 57835
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.4670667104214524e-05,
+      "loss": 0.0664,
+      "step": 57840
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.4670198302939384e-05,
+      "loss": 0.0527,
+      "step": 57845
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.4669729501664245e-05,
+      "loss": 0.1453,
+      "step": 57850
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.4669260700389105e-05,
+      "loss": 0.1784,
+      "step": 57855
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.4668791899113967e-05,
+      "loss": 0.1884,
+      "step": 57860
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.4668323097838829e-05,
+      "loss": 0.3005,
+      "step": 57865
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.4667854296563689e-05,
+      "loss": 0.0338,
+      "step": 57870
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.4667385495288549e-05,
+      "loss": 0.0088,
+      "step": 57875
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.4666916694013408e-05,
+      "loss": 0.0496,
+      "step": 57880
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.4666447892738268e-05,
+      "loss": 0.0948,
+      "step": 57885
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.466597909146313e-05,
+      "loss": 0.1013,
+      "step": 57890
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.466551029018799e-05,
+      "loss": 0.0931,
+      "step": 57895
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.466504148891285e-05,
+      "loss": 0.0853,
+      "step": 57900
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.4664572687637713e-05,
+      "loss": 0.1581,
+      "step": 57905
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.4664103886362573e-05,
+      "loss": 0.1908,
+      "step": 57910
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.4663635085087433e-05,
+      "loss": 0.203,
+      "step": 57915
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.4663166283812293e-05,
+      "loss": 0.0477,
+      "step": 57920
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.4662697482537153e-05,
+      "loss": 0.0474,
+      "step": 57925
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.4662228681262015e-05,
+      "loss": 0.0508,
+      "step": 57930
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.4661759879986874e-05,
+      "loss": 0.063,
+      "step": 57935
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.4661291078711734e-05,
+      "loss": 0.0944,
+      "step": 57940
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.4660822277436594e-05,
+      "loss": 0.1145,
+      "step": 57945
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.4660353476161458e-05,
+      "loss": 0.0749,
+      "step": 57950
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.4659884674886318e-05,
+      "loss": 0.1389,
+      "step": 57955
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.4659415873611178e-05,
+      "loss": 0.938,
+      "step": 57960
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.4658947072336037e-05,
+      "loss": 0.3698,
+      "step": 57965
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.4658478271060899e-05,
+      "loss": 0.0518,
+      "step": 57970
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.4658009469785759e-05,
+      "loss": 0.0445,
+      "step": 57975
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.4657540668510619e-05,
+      "loss": 0.041,
+      "step": 57980
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.4657071867235479e-05,
+      "loss": 0.1005,
+      "step": 57985
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.465660306596034e-05,
+      "loss": 0.1254,
+      "step": 57990
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.46561342646852e-05,
+      "loss": 0.076,
+      "step": 57995
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.4655665463410062e-05,
+      "loss": 0.0907,
+      "step": 58000
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.4655196662134922e-05,
+      "loss": 0.1793,
+      "step": 58005
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.4654727860859784e-05,
+      "loss": 0.2108,
+      "step": 58010
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.4654259059584644e-05,
+      "loss": 0.2919,
+      "step": 58015
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.4653790258309504e-05,
+      "loss": 0.0323,
+      "step": 58020
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.4653321457034363e-05,
+      "loss": 0.0621,
+      "step": 58025
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.4652852655759225e-05,
+      "loss": 0.0744,
+      "step": 58030
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.4652383854484085e-05,
+      "loss": 0.0495,
+      "step": 58035
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.4651915053208945e-05,
+      "loss": 0.0437,
+      "step": 58040
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.4651446251933807e-05,
+      "loss": 0.1159,
+      "step": 58045
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.4650977450658668e-05,
+      "loss": 0.1174,
+      "step": 58050
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.4650508649383528e-05,
+      "loss": 0.1162,
+      "step": 58055
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.4650039848108388e-05,
+      "loss": 0.3202,
+      "step": 58060
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.4649571046833248e-05,
+      "loss": 0.2968,
+      "step": 58065
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.464910224555811e-05,
+      "loss": 0.0346,
+      "step": 58070
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.464863344428297e-05,
+      "loss": 0.0612,
+      "step": 58075
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.464816464300783e-05,
+      "loss": 0.0309,
+      "step": 58080
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.464769584173269e-05,
+      "loss": 0.0749,
+      "step": 58085
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.4647227040457553e-05,
+      "loss": 0.0842,
+      "step": 58090
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.4646758239182413e-05,
+      "loss": 0.0822,
+      "step": 58095
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.4646289437907273e-05,
+      "loss": 0.1735,
+      "step": 58100
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.4645820636632133e-05,
+      "loss": 0.1367,
+      "step": 58105
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.4645351835356994e-05,
+      "loss": 0.2274,
+      "step": 58110
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.4644883034081854e-05,
+      "loss": 0.2721,
+      "step": 58115
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.4644414232806714e-05,
+      "loss": 0.0713,
+      "step": 58120
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.4643945431531574e-05,
+      "loss": 0.0516,
+      "step": 58125
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.4643476630256434e-05,
+      "loss": 0.0327,
+      "step": 58130
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.4643007828981297e-05,
+      "loss": 0.0882,
+      "step": 58135
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.4642539027706157e-05,
+      "loss": 0.1025,
+      "step": 58140
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.4642070226431017e-05,
+      "loss": 0.0793,
+      "step": 58145
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.4641601425155879e-05,
+      "loss": 0.2249,
+      "step": 58150
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.4641132623880739e-05,
+      "loss": 0.187,
+      "step": 58155
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.4640663822605599e-05,
+      "loss": 0.2013,
+      "step": 58160
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.4640195021330459e-05,
+      "loss": 0.3344,
+      "step": 58165
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.4639726220055318e-05,
+      "loss": 0.0645,
+      "step": 58170
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.463925741878018e-05,
+      "loss": 0.0234,
+      "step": 58175
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.463878861750504e-05,
+      "loss": 0.0384,
+      "step": 58180
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.4638319816229902e-05,
+      "loss": 0.0991,
+      "step": 58185
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.4637851014954763e-05,
+      "loss": 0.064,
+      "step": 58190
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.4637382213679623e-05,
+      "loss": 0.1129,
+      "step": 58195
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.4636913412404483e-05,
+      "loss": 0.1446,
+      "step": 58200
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.4636444611129343e-05,
+      "loss": 0.1168,
+      "step": 58205
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.4635975809854203e-05,
+      "loss": 0.2535,
+      "step": 58210
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.4635507008579065e-05,
+      "loss": 0.2867,
+      "step": 58215
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.4635038207303925e-05,
+      "loss": 0.0554,
+      "step": 58220
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.4634569406028785e-05,
+      "loss": 0.0347,
+      "step": 58225
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.4634100604753648e-05,
+      "loss": 0.0444,
+      "step": 58230
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.4633631803478508e-05,
+      "loss": 0.0649,
+      "step": 58235
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.4633163002203368e-05,
+      "loss": 0.0463,
+      "step": 58240
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.4632694200928228e-05,
+      "loss": 0.0279,
+      "step": 58245
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.4632225399653088e-05,
+      "loss": 0.0854,
+      "step": 58250
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.463175659837795e-05,
+      "loss": 0.2262,
+      "step": 58255
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.4631287797102809e-05,
+      "loss": 0.2178,
+      "step": 58260
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.4630818995827669e-05,
+      "loss": 0.4489,
+      "step": 58265
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.4630350194552529e-05,
+      "loss": 0.0468,
+      "step": 58270
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.4629881393277392e-05,
+      "loss": 0.0298,
+      "step": 58275
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.4629412592002252e-05,
+      "loss": 0.0473,
+      "step": 58280
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.4628943790727112e-05,
+      "loss": 0.0977,
+      "step": 58285
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.4628474989451972e-05,
+      "loss": 0.0873,
+      "step": 58290
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.4628006188176834e-05,
+      "loss": 0.155,
+      "step": 58295
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.4627537386901694e-05,
+      "loss": 0.1499,
+      "step": 58300
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.4627068585626554e-05,
+      "loss": 0.1295,
+      "step": 58305
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.4626599784351414e-05,
+      "loss": 0.2148,
+      "step": 58310
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.4626130983076273e-05,
+      "loss": 0.2469,
+      "step": 58315
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.4625662181801135e-05,
+      "loss": 0.0731,
+      "step": 58320
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.4625193380525997e-05,
+      "loss": 0.051,
+      "step": 58325
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.4624724579250857e-05,
+      "loss": 0.037,
+      "step": 58330
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.4624255777975718e-05,
+      "loss": 0.081,
+      "step": 58335
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.4623786976700578e-05,
+      "loss": 0.1175,
+      "step": 58340
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.4623318175425438e-05,
+      "loss": 0.1785,
+      "step": 58345
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.4622849374150298e-05,
+      "loss": 0.1697,
+      "step": 58350
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.4622380572875158e-05,
+      "loss": 0.2097,
+      "step": 58355
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.462191177160002e-05,
+      "loss": 0.1784,
+      "step": 58360
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.462144297032488e-05,
+      "loss": 0.2476,
+      "step": 58365
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.4620974169049741e-05,
+      "loss": 0.0426,
+      "step": 58370
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.4620505367774603e-05,
+      "loss": 0.0245,
+      "step": 58375
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.4620036566499463e-05,
+      "loss": 0.0399,
+      "step": 58380
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.4619567765224323e-05,
+      "loss": 0.0598,
+      "step": 58385
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.4619098963949183e-05,
+      "loss": 0.0551,
+      "step": 58390
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.4618630162674044e-05,
+      "loss": 0.0788,
+      "step": 58395
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.4618161361398904e-05,
+      "loss": 0.1213,
+      "step": 58400
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.4617692560123764e-05,
+      "loss": 0.1959,
+      "step": 58405
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.4617223758848624e-05,
+      "loss": 0.279,
+      "step": 58410
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.4616754957573487e-05,
+      "loss": 0.449,
+      "step": 58415
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.4616286156298347e-05,
+      "loss": 0.0205,
+      "step": 58420
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.4615817355023207e-05,
+      "loss": 0.0736,
+      "step": 58425
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.4615348553748067e-05,
+      "loss": 0.059,
+      "step": 58430
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.4614879752472929e-05,
+      "loss": 0.0961,
+      "step": 58435
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.4614410951197789e-05,
+      "loss": 0.0493,
+      "step": 58440
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.4613942149922649e-05,
+      "loss": 0.1004,
+      "step": 58445
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.4613473348647509e-05,
+      "loss": 0.1301,
+      "step": 58450
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.4613004547372369e-05,
+      "loss": 0.1581,
+      "step": 58455
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.4612535746097232e-05,
+      "loss": 0.2438,
+      "step": 58460
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.4612066944822092e-05,
+      "loss": 0.3437,
+      "step": 58465
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.4611598143546952e-05,
+      "loss": 0.0362,
+      "step": 58470
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.4611129342271813e-05,
+      "loss": 0.0567,
+      "step": 58475
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.4610660540996673e-05,
+      "loss": 0.025,
+      "step": 58480
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.4610191739721533e-05,
+      "loss": 0.0793,
+      "step": 58485
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.4609722938446393e-05,
+      "loss": 0.197,
+      "step": 58490
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.4609254137171253e-05,
+      "loss": 0.0953,
+      "step": 58495
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.4608785335896115e-05,
+      "loss": 0.1387,
+      "step": 58500
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.4608316534620975e-05,
+      "loss": 0.1528,
+      "step": 58505
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.4607847733345836e-05,
+      "loss": 0.2204,
+      "step": 58510
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.4607378932070698e-05,
+      "loss": 0.2327,
+      "step": 58515
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.4606910130795558e-05,
+      "loss": 0.0646,
+      "step": 58520
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.4606441329520418e-05,
+      "loss": 0.0443,
+      "step": 58525
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.4605972528245278e-05,
+      "loss": 0.0464,
+      "step": 58530
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.4605503726970138e-05,
+      "loss": 0.0746,
+      "step": 58535
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.4605034925695e-05,
+      "loss": 0.0805,
+      "step": 58540
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.460456612441986e-05,
+      "loss": 0.0536,
+      "step": 58545
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.460409732314472e-05,
+      "loss": 0.2158,
+      "step": 58550
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.4603628521869582e-05,
+      "loss": 0.1918,
+      "step": 58555
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.4603159720594442e-05,
+      "loss": 0.1836,
+      "step": 58560
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.4602690919319302e-05,
+      "loss": 0.3381,
+      "step": 58565
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.4602222118044162e-05,
+      "loss": 0.1033,
+      "step": 58570
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.4601753316769022e-05,
+      "loss": 0.0329,
+      "step": 58575
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.4601284515493884e-05,
+      "loss": 0.0704,
+      "step": 58580
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.4600815714218744e-05,
+      "loss": 0.0311,
+      "step": 58585
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.4600346912943604e-05,
+      "loss": 0.0826,
+      "step": 58590
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.4599878111668464e-05,
+      "loss": 0.1267,
+      "step": 58595
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.4599409310393327e-05,
+      "loss": 0.1818,
+      "step": 58600
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.4598940509118187e-05,
+      "loss": 0.1993,
+      "step": 58605
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.4598471707843047e-05,
+      "loss": 0.2257,
+      "step": 58610
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.4598002906567907e-05,
+      "loss": 0.2809,
+      "step": 58615
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.4597534105292768e-05,
+      "loss": 0.0743,
+      "step": 58620
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.4597065304017628e-05,
+      "loss": 0.0514,
+      "step": 58625
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.4596596502742488e-05,
+      "loss": 0.087,
+      "step": 58630
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.4596127701467348e-05,
+      "loss": 0.0579,
+      "step": 58635
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.4595658900192208e-05,
+      "loss": 0.1159,
+      "step": 58640
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.459519009891707e-05,
+      "loss": 0.0654,
+      "step": 58645
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.4594721297641931e-05,
+      "loss": 0.0656,
+      "step": 58650
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.4594252496366791e-05,
+      "loss": 0.0926,
+      "step": 58655
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.4593783695091653e-05,
+      "loss": 0.1706,
+      "step": 58660
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.4593314893816513e-05,
+      "loss": 0.1598,
+      "step": 58665
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.4592846092541373e-05,
+      "loss": 0.027,
+      "step": 58670
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.4592377291266233e-05,
+      "loss": 0.0427,
+      "step": 58675
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.4591908489991093e-05,
+      "loss": 0.0539,
+      "step": 58680
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.4591439688715954e-05,
+      "loss": 0.084,
+      "step": 58685
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.4590970887440814e-05,
+      "loss": 0.0856,
+      "step": 58690
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.4590502086165676e-05,
+      "loss": 0.0986,
+      "step": 58695
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.4590033284890538e-05,
+      "loss": 0.1526,
+      "step": 58700
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.4589564483615397e-05,
+      "loss": 0.1722,
+      "step": 58705
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.4589095682340257e-05,
+      "loss": 0.2371,
+      "step": 58710
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.4588626881065117e-05,
+      "loss": 0.2128,
+      "step": 58715
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.4588158079789977e-05,
+      "loss": 0.0469,
+      "step": 58720
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.4587689278514839e-05,
+      "loss": 0.0508,
+      "step": 58725
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.4587220477239699e-05,
+      "loss": 0.0676,
+      "step": 58730
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.4586751675964559e-05,
+      "loss": 0.0685,
+      "step": 58735
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.4586282874689422e-05,
+      "loss": 0.0737,
+      "step": 58740
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.4585814073414282e-05,
+      "loss": 0.062,
+      "step": 58745
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.4585345272139142e-05,
+      "loss": 0.0714,
+      "step": 58750
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.4584876470864002e-05,
+      "loss": 0.1646,
+      "step": 58755
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.4584407669588862e-05,
+      "loss": 0.2577,
+      "step": 58760
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.4583938868313723e-05,
+      "loss": 0.2857,
+      "step": 58765
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.4583470067038583e-05,
+      "loss": 0.0321,
+      "step": 58770
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.4583001265763443e-05,
+      "loss": 0.0556,
+      "step": 58775
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.4582532464488303e-05,
+      "loss": 0.1188,
+      "step": 58780
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.4582063663213165e-05,
+      "loss": 0.0495,
+      "step": 58785
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.4581594861938026e-05,
+      "loss": 0.1317,
+      "step": 58790
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.4581126060662886e-05,
+      "loss": 0.1131,
+      "step": 58795
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.4580657259387746e-05,
+      "loss": 0.1266,
+      "step": 58800
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.4580188458112608e-05,
+      "loss": 0.1261,
+      "step": 58805
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.4579719656837468e-05,
+      "loss": 0.17,
+      "step": 58810
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.4579250855562328e-05,
+      "loss": 0.2425,
+      "step": 58815
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.4578782054287188e-05,
+      "loss": 0.0795,
+      "step": 58820
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.457831325301205e-05,
+      "loss": 0.0429,
+      "step": 58825
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.457784445173691e-05,
+      "loss": 0.0437,
+      "step": 58830
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.4577375650461771e-05,
+      "loss": 0.0756,
+      "step": 58835
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.4576906849186631e-05,
+      "loss": 0.0911,
+      "step": 58840
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.4576438047911493e-05,
+      "loss": 0.1158,
+      "step": 58845
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.4575969246636352e-05,
+      "loss": 0.1984,
+      "step": 58850
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.4575500445361212e-05,
+      "loss": 0.1924,
+      "step": 58855
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.4575031644086072e-05,
+      "loss": 0.2444,
+      "step": 58860
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.4574562842810934e-05,
+      "loss": 0.2247,
+      "step": 58865
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.4574094041535794e-05,
+      "loss": 0.0268,
+      "step": 58870
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.4573625240260654e-05,
+      "loss": 0.033,
+      "step": 58875
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.4573156438985515e-05,
+      "loss": 0.0568,
+      "step": 58880
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.4572687637710377e-05,
+      "loss": 0.0308,
+      "step": 58885
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.4572218836435237e-05,
+      "loss": 0.065,
+      "step": 58890
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.4571750035160097e-05,
+      "loss": 0.1286,
+      "step": 58895
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.4571281233884957e-05,
+      "loss": 0.1551,
+      "step": 58900
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.4570812432609819e-05,
+      "loss": 0.3151,
+      "step": 58905
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.4570343631334678e-05,
+      "loss": 0.2196,
+      "step": 58910
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.4569874830059538e-05,
+      "loss": 0.3267,
+      "step": 58915
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.4569406028784398e-05,
+      "loss": 0.0439,
+      "step": 58920
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.4568937227509262e-05,
+      "loss": 0.0192,
+      "step": 58925
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.4568468426234122e-05,
+      "loss": 0.058,
+      "step": 58930
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.4567999624958981e-05,
+      "loss": 0.096,
+      "step": 58935
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.4567530823683841e-05,
+      "loss": 0.1104,
+      "step": 58940
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.4567062022408703e-05,
+      "loss": 0.1205,
+      "step": 58945
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.4566593221133563e-05,
+      "loss": 0.0622,
+      "step": 58950
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.4566124419858423e-05,
+      "loss": 0.2004,
+      "step": 58955
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.4565655618583283e-05,
+      "loss": 0.2747,
+      "step": 58960
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.4565186817308143e-05,
+      "loss": 0.2782,
+      "step": 58965
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.4564718016033004e-05,
+      "loss": 0.0895,
+      "step": 58970
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.4564249214757866e-05,
+      "loss": 0.1072,
+      "step": 58975
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.4563780413482726e-05,
+      "loss": 0.0672,
+      "step": 58980
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.4563311612207588e-05,
+      "loss": 0.0578,
+      "step": 58985
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.4562842810932448e-05,
+      "loss": 0.0943,
+      "step": 58990
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.4562374009657307e-05,
+      "loss": 0.1205,
+      "step": 58995
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.4561905208382167e-05,
+      "loss": 0.2066,
+      "step": 59000
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.4561436407107027e-05,
+      "loss": 0.2212,
+      "step": 59005
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.4560967605831889e-05,
+      "loss": 0.3231,
+      "step": 59010
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.4560498804556749e-05,
+      "loss": 0.18,
+      "step": 59015
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.456003000328161e-05,
+      "loss": 0.025,
+      "step": 59020
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.4559561202006472e-05,
+      "loss": 0.0556,
+      "step": 59025
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.4559092400731332e-05,
+      "loss": 0.0676,
+      "step": 59030
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.4558623599456192e-05,
+      "loss": 0.0609,
+      "step": 59035
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.4558154798181052e-05,
+      "loss": 0.0417,
+      "step": 59040
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.4557685996905912e-05,
+      "loss": 0.0902,
+      "step": 59045
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.4557217195630774e-05,
+      "loss": 0.0511,
+      "step": 59050
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.4556748394355633e-05,
+      "loss": 0.1862,
+      "step": 59055
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.4556279593080493e-05,
+      "loss": 0.2465,
+      "step": 59060
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.4555810791805357e-05,
+      "loss": 0.4655,
+      "step": 59065
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.4555341990530217e-05,
+      "loss": 0.0271,
+      "step": 59070
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.4554873189255077e-05,
+      "loss": 0.0468,
+      "step": 59075
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.4554404387979937e-05,
+      "loss": 0.0496,
+      "step": 59080
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.4553935586704796e-05,
+      "loss": 0.0626,
+      "step": 59085
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.4553466785429658e-05,
+      "loss": 0.0464,
+      "step": 59090
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.4552997984154518e-05,
+      "loss": 0.1007,
+      "step": 59095
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.4552529182879378e-05,
+      "loss": 0.0917,
+      "step": 59100
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.4552060381604238e-05,
+      "loss": 0.1745,
+      "step": 59105
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.4551591580329098e-05,
+      "loss": 0.3145,
+      "step": 59110
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.4551122779053961e-05,
+      "loss": 0.2732,
+      "step": 59115
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.4550653977778821e-05,
+      "loss": 0.0603,
+      "step": 59120
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.4550185176503681e-05,
+      "loss": 0.0799,
+      "step": 59125
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.4549716375228543e-05,
+      "loss": 0.0561,
+      "step": 59130
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.4549247573953403e-05,
+      "loss": 0.1065,
+      "step": 59135
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.4548778772678262e-05,
+      "loss": 0.0601,
+      "step": 59140
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.4548309971403122e-05,
+      "loss": 0.0949,
+      "step": 59145
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.4547841170127982e-05,
+      "loss": 0.1082,
+      "step": 59150
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.4547372368852844e-05,
+      "loss": 0.1364,
+      "step": 59155
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.4546903567577706e-05,
+      "loss": 0.1918,
+      "step": 59160
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.4546434766302566e-05,
+      "loss": 0.1907,
+      "step": 59165
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.4545965965027427e-05,
+      "loss": 0.0355,
+      "step": 59170
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.4545497163752287e-05,
+      "loss": 0.0816,
+      "step": 59175
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.4545028362477147e-05,
+      "loss": 0.0342,
+      "step": 59180
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.4544559561202007e-05,
+      "loss": 0.04,
+      "step": 59185
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.4544090759926867e-05,
+      "loss": 0.0471,
+      "step": 59190
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.4543621958651729e-05,
+      "loss": 0.1639,
+      "step": 59195
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.4543153157376588e-05,
+      "loss": 0.0879,
+      "step": 59200
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.454268435610145e-05,
+      "loss": 0.1821,
+      "step": 59205
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.4542215554826312e-05,
+      "loss": 0.188,
+      "step": 59210
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.4541746753551172e-05,
+      "loss": 0.2354,
+      "step": 59215
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.4541277952276032e-05,
+      "loss": 0.0471,
+      "step": 59220
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.4540809151000892e-05,
+      "loss": 0.0739,
+      "step": 59225
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.4540340349725751e-05,
+      "loss": 0.0486,
+      "step": 59230
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.4539871548450613e-05,
+      "loss": 0.0612,
+      "step": 59235
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.4539402747175473e-05,
+      "loss": 0.0693,
+      "step": 59240
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.4538933945900333e-05,
+      "loss": 0.0583,
+      "step": 59245
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.4538465144625196e-05,
+      "loss": 0.1747,
+      "step": 59250
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.4537996343350056e-05,
+      "loss": 0.2391,
+      "step": 59255
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4537527542074916e-05,
+      "loss": 0.2129,
+      "step": 59260
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4537058740799776e-05,
+      "loss": 0.3122,
+      "step": 59265
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4536589939524636e-05,
+      "loss": 0.039,
+      "step": 59270
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4536121138249498e-05,
+      "loss": 0.0314,
+      "step": 59275
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4535652336974358e-05,
+      "loss": 0.0421,
+      "step": 59280
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4535183535699218e-05,
+      "loss": 0.0855,
+      "step": 59285
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4534714734424077e-05,
+      "loss": 0.0655,
+      "step": 59290
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4534245933148939e-05,
+      "loss": 0.0452,
+      "step": 59295
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.45337771318738e-05,
+      "loss": 0.1428,
+      "step": 59300
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.453330833059866e-05,
+      "loss": 0.1613,
+      "step": 59305
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.453283952932352e-05,
+      "loss": 0.2401,
+      "step": 59310
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4532370728048382e-05,
+      "loss": 0.26,
+      "step": 59315
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4531901926773242e-05,
+      "loss": 0.0539,
+      "step": 59320
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4531433125498102e-05,
+      "loss": 0.0868,
+      "step": 59325
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4530964324222962e-05,
+      "loss": 0.052,
+      "step": 59330
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4530495522947824e-05,
+      "loss": 0.1254,
+      "step": 59335
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4530026721672684e-05,
+      "loss": 0.0738,
+      "step": 59340
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4529557920397545e-05,
+      "loss": 0.0963,
+      "step": 59345
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4529089119122405e-05,
+      "loss": 0.1563,
+      "step": 59350
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4528620317847267e-05,
+      "loss": 0.1559,
+      "step": 59355
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4528151516572127e-05,
+      "loss": 0.2358,
+      "step": 59360
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4527682715296987e-05,
+      "loss": 0.2168,
+      "step": 59365
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4527213914021847e-05,
+      "loss": 0.0453,
+      "step": 59370
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4526745112746708e-05,
+      "loss": 0.0242,
+      "step": 59375
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4526276311471568e-05,
+      "loss": 0.0305,
+      "step": 59380
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4525807510196428e-05,
+      "loss": 0.0608,
+      "step": 59385
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.452533870892129e-05,
+      "loss": 0.095,
+      "step": 59390
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4524869907646151e-05,
+      "loss": 0.1336,
+      "step": 59395
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4524401106371011e-05,
+      "loss": 0.0679,
+      "step": 59400
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4523932305095871e-05,
+      "loss": 0.1726,
+      "step": 59405
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4523463503820731e-05,
+      "loss": 0.1142,
+      "step": 59410
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4522994702545593e-05,
+      "loss": 0.2268,
+      "step": 59415
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4522525901270453e-05,
+      "loss": 0.0513,
+      "step": 59420
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4522057099995313e-05,
+      "loss": 0.0354,
+      "step": 59425
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4521588298720173e-05,
+      "loss": 0.1059,
+      "step": 59430
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4521119497445032e-05,
+      "loss": 0.1179,
+      "step": 59435
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4520650696169896e-05,
+      "loss": 0.0819,
+      "step": 59440
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4520181894894756e-05,
+      "loss": 0.1253,
+      "step": 59445
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4519713093619616e-05,
+      "loss": 0.1265,
+      "step": 59450
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4519244292344477e-05,
+      "loss": 0.1525,
+      "step": 59455
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4518775491069337e-05,
+      "loss": 0.1325,
+      "step": 59460
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4518306689794197e-05,
+      "loss": 0.2705,
+      "step": 59465
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4517837888519057e-05,
+      "loss": 0.0597,
+      "step": 59470
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4517369087243917e-05,
+      "loss": 0.0441,
+      "step": 59475
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4516900285968779e-05,
+      "loss": 0.0538,
+      "step": 59480
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.451643148469364e-05,
+      "loss": 0.0849,
+      "step": 59485
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.45159626834185e-05,
+      "loss": 0.0892,
+      "step": 59490
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4515493882143362e-05,
+      "loss": 0.0865,
+      "step": 59495
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4515025080868222e-05,
+      "loss": 0.0963,
+      "step": 59500
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4514556279593082e-05,
+      "loss": 0.1407,
+      "step": 59505
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4514087478317942e-05,
+      "loss": 0.3032,
+      "step": 59510
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4513618677042802e-05,
+      "loss": 0.3026,
+      "step": 59515
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4513149875767663e-05,
+      "loss": 0.0481,
+      "step": 59520
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4512681074492523e-05,
+      "loss": 0.0437,
+      "step": 59525
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4512212273217385e-05,
+      "loss": 0.0677,
+      "step": 59530
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4511743471942246e-05,
+      "loss": 0.0582,
+      "step": 59535
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4511274670667106e-05,
+      "loss": 0.0841,
+      "step": 59540
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4510805869391966e-05,
+      "loss": 0.0991,
+      "step": 59545
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4510337068116826e-05,
+      "loss": 0.0962,
+      "step": 59550
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4509868266841686e-05,
+      "loss": 0.1709,
+      "step": 59555
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4509399465566548e-05,
+      "loss": 0.2061,
+      "step": 59560
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4508930664291408e-05,
+      "loss": 0.2598,
+      "step": 59565
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4508461863016268e-05,
+      "loss": 0.0427,
+      "step": 59570
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4507993061741131e-05,
+      "loss": 0.0679,
+      "step": 59575
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4507524260465991e-05,
+      "loss": 0.0677,
+      "step": 59580
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.450705545919085e-05,
+      "loss": 0.0553,
+      "step": 59585
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.450658665791571e-05,
+      "loss": 0.0714,
+      "step": 59590
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.450611785664057e-05,
+      "loss": 0.0563,
+      "step": 59595
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4505649055365432e-05,
+      "loss": 0.0597,
+      "step": 59600
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4505180254090292e-05,
+      "loss": 0.1147,
+      "step": 59605
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4504711452815152e-05,
+      "loss": 0.1732,
+      "step": 59610
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4504242651540012e-05,
+      "loss": 0.2322,
+      "step": 59615
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4503773850264872e-05,
+      "loss": 0.0126,
+      "step": 59620
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4503305048989735e-05,
+      "loss": 0.0667,
+      "step": 59625
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4502836247714595e-05,
+      "loss": 0.0183,
+      "step": 59630
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4502367446439455e-05,
+      "loss": 0.1245,
+      "step": 59635
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4501898645164317e-05,
+      "loss": 0.211,
+      "step": 59640
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4501429843889177e-05,
+      "loss": 0.0844,
+      "step": 59645
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4500961042614037e-05,
+      "loss": 0.1949,
+      "step": 59650
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4500492241338897e-05,
+      "loss": 0.1545,
+      "step": 59655
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4500023440063757e-05,
+      "loss": 0.3161,
+      "step": 59660
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4499554638788618e-05,
+      "loss": 0.2175,
+      "step": 59665
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.449908583751348e-05,
+      "loss": 0.0381,
+      "step": 59670
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.449861703623834e-05,
+      "loss": 0.0196,
+      "step": 59675
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4498148234963201e-05,
+      "loss": 0.1143,
+      "step": 59680
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4497679433688061e-05,
+      "loss": 0.0837,
+      "step": 59685
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4497210632412921e-05,
+      "loss": 0.129,
+      "step": 59690
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4496741831137781e-05,
+      "loss": 0.0908,
+      "step": 59695
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4496273029862641e-05,
+      "loss": 0.1649,
+      "step": 59700
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4495804228587503e-05,
+      "loss": 0.1539,
+      "step": 59705
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4495335427312363e-05,
+      "loss": 0.3572,
+      "step": 59710
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4494866626037224e-05,
+      "loss": 0.3036,
+      "step": 59715
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4494397824762086e-05,
+      "loss": 0.0154,
+      "step": 59720
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4493929023486946e-05,
+      "loss": 0.0211,
+      "step": 59725
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4493460222211806e-05,
+      "loss": 0.0294,
+      "step": 59730
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4492991420936666e-05,
+      "loss": 0.0781,
+      "step": 59735
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4492522619661526e-05,
+      "loss": 0.0934,
+      "step": 59740
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4492053818386387e-05,
+      "loss": 0.0954,
+      "step": 59745
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4491585017111247e-05,
+      "loss": 0.1425,
+      "step": 59750
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4491116215836107e-05,
+      "loss": 0.1863,
+      "step": 59755
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4490647414560967e-05,
+      "loss": 0.1952,
+      "step": 59760
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.449017861328583e-05,
+      "loss": 0.1966,
+      "step": 59765
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.448970981201069e-05,
+      "loss": 0.0379,
+      "step": 59770
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.448924101073555e-05,
+      "loss": 0.0309,
+      "step": 59775
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.448877220946041e-05,
+      "loss": 0.0834,
+      "step": 59780
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4488303408185272e-05,
+      "loss": 0.064,
+      "step": 59785
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4487834606910132e-05,
+      "loss": 0.0708,
+      "step": 59790
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4487365805634992e-05,
+      "loss": 0.1656,
+      "step": 59795
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4486897004359852e-05,
+      "loss": 0.133,
+      "step": 59800
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4486428203084713e-05,
+      "loss": 0.1545,
+      "step": 59805
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4485959401809575e-05,
+      "loss": 0.1484,
+      "step": 59810
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4485490600534435e-05,
+      "loss": 0.2154,
+      "step": 59815
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4485021799259296e-05,
+      "loss": 0.0304,
+      "step": 59820
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4484552997984156e-05,
+      "loss": 0.2941,
+      "step": 59825
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4484084196709016e-05,
+      "loss": 0.099,
+      "step": 59830
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4483615395433876e-05,
+      "loss": 0.0803,
+      "step": 59835
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4483146594158736e-05,
+      "loss": 0.0795,
+      "step": 59840
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4482677792883598e-05,
+      "loss": 0.112,
+      "step": 59845
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4482208991608458e-05,
+      "loss": 0.1533,
+      "step": 59850
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.448174019033332e-05,
+      "loss": 0.1552,
+      "step": 59855
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4481271389058181e-05,
+      "loss": 0.2634,
+      "step": 59860
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4480802587783041e-05,
+      "loss": 0.2807,
+      "step": 59865
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4480333786507901e-05,
+      "loss": 0.0613,
+      "step": 59870
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.447986498523276e-05,
+      "loss": 0.0644,
+      "step": 59875
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.447939618395762e-05,
+      "loss": 0.0322,
+      "step": 59880
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4478927382682482e-05,
+      "loss": 0.0303,
+      "step": 59885
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4478458581407342e-05,
+      "loss": 0.0804,
+      "step": 59890
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4477989780132202e-05,
+      "loss": 0.136,
+      "step": 59895
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4477520978857066e-05,
+      "loss": 0.1515,
+      "step": 59900
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4477052177581926e-05,
+      "loss": 0.2001,
+      "step": 59905
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4476583376306785e-05,
+      "loss": 0.2547,
+      "step": 59910
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4476114575031645e-05,
+      "loss": 0.315,
+      "step": 59915
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4475645773756505e-05,
+      "loss": 0.082,
+      "step": 59920
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4475176972481367e-05,
+      "loss": 0.0578,
+      "step": 59925
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4474708171206227e-05,
+      "loss": 0.0752,
+      "step": 59930
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4474239369931087e-05,
+      "loss": 0.0437,
+      "step": 59935
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4473770568655947e-05,
+      "loss": 0.1486,
+      "step": 59940
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4473301767380807e-05,
+      "loss": 0.1381,
+      "step": 59945
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.447283296610567e-05,
+      "loss": 0.0976,
+      "step": 59950
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.447236416483053e-05,
+      "loss": 0.0948,
+      "step": 59955
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.447189536355539e-05,
+      "loss": 0.1661,
+      "step": 59960
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4471426562280251e-05,
+      "loss": 0.1997,
+      "step": 59965
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4470957761005111e-05,
+      "loss": 0.0974,
+      "step": 59970
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4470488959729971e-05,
+      "loss": 0.0479,
+      "step": 59975
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4470020158454831e-05,
+      "loss": 0.0315,
+      "step": 59980
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4469551357179691e-05,
+      "loss": 0.0788,
+      "step": 59985
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4469082555904553e-05,
+      "loss": 0.0727,
+      "step": 59990
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4468613754629414e-05,
+      "loss": 0.1111,
+      "step": 59995
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4468144953354274e-05,
+      "loss": 0.0669,
+      "step": 60000
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4467676152079136e-05,
+      "loss": 0.1314,
+      "step": 60005
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4467207350803996e-05,
+      "loss": 0.2948,
+      "step": 60010
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4466738549528856e-05,
+      "loss": 0.2877,
+      "step": 60015
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4466269748253716e-05,
+      "loss": 0.0344,
+      "step": 60020
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4465800946978576e-05,
+      "loss": 0.0666,
+      "step": 60025
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4465332145703437e-05,
+      "loss": 0.053,
+      "step": 60030
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4464863344428297e-05,
+      "loss": 0.0348,
+      "step": 60035
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4464394543153159e-05,
+      "loss": 0.0455,
+      "step": 60040
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.446392574187802e-05,
+      "loss": 0.0948,
+      "step": 60045
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.446345694060288e-05,
+      "loss": 0.1343,
+      "step": 60050
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.446298813932774e-05,
+      "loss": 0.1746,
+      "step": 60055
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.44625193380526e-05,
+      "loss": 0.2422,
+      "step": 60060
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.446205053677746e-05,
+      "loss": 0.4788,
+      "step": 60065
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4461581735502322e-05,
+      "loss": 0.0414,
+      "step": 60070
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4461112934227182e-05,
+      "loss": 0.0488,
+      "step": 60075
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4460644132952042e-05,
+      "loss": 0.057,
+      "step": 60080
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4460175331676902e-05,
+      "loss": 0.0404,
+      "step": 60085
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4459706530401765e-05,
+      "loss": 0.1053,
+      "step": 60090
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4459237729126625e-05,
+      "loss": 0.1003,
+      "step": 60095
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4458768927851485e-05,
+      "loss": 0.042,
+      "step": 60100
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4458300126576345e-05,
+      "loss": 0.1399,
+      "step": 60105
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4457831325301207e-05,
+      "loss": 0.1524,
+      "step": 60110
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4457362524026066e-05,
+      "loss": 0.2414,
+      "step": 60115
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4456893722750926e-05,
+      "loss": 0.0433,
+      "step": 60120
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4456424921475786e-05,
+      "loss": 0.0522,
+      "step": 60125
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4455956120200646e-05,
+      "loss": 0.0466,
+      "step": 60130
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.445548731892551e-05,
+      "loss": 0.1194,
+      "step": 60135
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.445501851765037e-05,
+      "loss": 0.0274,
+      "step": 60140
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.445454971637523e-05,
+      "loss": 0.0872,
+      "step": 60145
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4454080915100091e-05,
+      "loss": 0.3045,
+      "step": 60150
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4453612113824951e-05,
+      "loss": 0.1004,
+      "step": 60155
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4453143312549811e-05,
+      "loss": 0.2428,
+      "step": 60160
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4452674511274671e-05,
+      "loss": 0.3085,
+      "step": 60165
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4452205709999532e-05,
+      "loss": 0.0418,
+      "step": 60170
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4451736908724392e-05,
+      "loss": 0.0232,
+      "step": 60175
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4451268107449254e-05,
+      "loss": 0.054,
+      "step": 60180
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4450799306174114e-05,
+      "loss": 0.0278,
+      "step": 60185
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4450330504898976e-05,
+      "loss": 0.0537,
+      "step": 60190
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4449861703623836e-05,
+      "loss": 0.0563,
+      "step": 60195
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4449392902348695e-05,
+      "loss": 0.215,
+      "step": 60200
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4448924101073555e-05,
+      "loss": 0.2211,
+      "step": 60205
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4448455299798417e-05,
+      "loss": 0.2244,
+      "step": 60210
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4447986498523277e-05,
+      "loss": 0.304,
+      "step": 60215
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4447517697248137e-05,
+      "loss": 0.0084,
+      "step": 60220
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4447048895972999e-05,
+      "loss": 0.0444,
+      "step": 60225
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.444658009469786e-05,
+      "loss": 0.0909,
+      "step": 60230
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.444611129342272e-05,
+      "loss": 0.0759,
+      "step": 60235
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.444564249214758e-05,
+      "loss": 0.1385,
+      "step": 60240
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.444517369087244e-05,
+      "loss": 0.1458,
+      "step": 60245
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4444704889597302e-05,
+      "loss": 0.0788,
+      "step": 60250
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4444236088322162e-05,
+      "loss": 0.1923,
+      "step": 60255
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4443767287047021e-05,
+      "loss": 0.3621,
+      "step": 60260
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4443298485771881e-05,
+      "loss": 0.2793,
+      "step": 60265
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4442829684496741e-05,
+      "loss": 0.0878,
+      "step": 60270
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4442360883221605e-05,
+      "loss": 0.0328,
+      "step": 60275
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4441892081946465e-05,
+      "loss": 0.0712,
+      "step": 60280
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4441423280671325e-05,
+      "loss": 0.0215,
+      "step": 60285
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4440954479396186e-05,
+      "loss": 0.0537,
+      "step": 60290
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4440485678121046e-05,
+      "loss": 0.0864,
+      "step": 60295
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4440016876845906e-05,
+      "loss": 0.1054,
+      "step": 60300
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4439548075570766e-05,
+      "loss": 0.1377,
+      "step": 60305
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4439079274295626e-05,
+      "loss": 0.2142,
+      "step": 60310
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4438610473020488e-05,
+      "loss": 0.2317,
+      "step": 60315
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4438141671745349e-05,
+      "loss": 0.0255,
+      "step": 60320
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4437672870470209e-05,
+      "loss": 0.0471,
+      "step": 60325
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.443720406919507e-05,
+      "loss": 0.056,
+      "step": 60330
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.443673526791993e-05,
+      "loss": 0.1309,
+      "step": 60335
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.443626646664479e-05,
+      "loss": 0.0999,
+      "step": 60340
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.443579766536965e-05,
+      "loss": 0.1236,
+      "step": 60345
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.443532886409451e-05,
+      "loss": 0.1663,
+      "step": 60350
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.4434860062819372e-05,
+      "loss": 0.2222,
+      "step": 60355
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.4434391261544232e-05,
+      "loss": 0.2062,
+      "step": 60360
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.4433922460269094e-05,
+      "loss": 0.1832,
+      "step": 60365
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.4433453658993955e-05,
+      "loss": 0.0224,
+      "step": 60370
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.4432984857718815e-05,
+      "loss": 0.0739,
+      "step": 60375
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.4432516056443675e-05,
+      "loss": 0.045,
+      "step": 60380
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.4432047255168535e-05,
+      "loss": 0.1194,
+      "step": 60385
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.4431578453893395e-05,
+      "loss": 0.1009,
+      "step": 60390
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.4431109652618257e-05,
+      "loss": 0.0623,
+      "step": 60395
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.4430640851343117e-05,
+      "loss": 0.095,
+      "step": 60400
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.4430172050067976e-05,
+      "loss": 0.2098,
+      "step": 60405
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.4429703248792836e-05,
+      "loss": 0.2175,
+      "step": 60410
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.44292344475177e-05,
+      "loss": 0.2349,
+      "step": 60415
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.442876564624256e-05,
+      "loss": 0.0606,
+      "step": 60420
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.442829684496742e-05,
+      "loss": 0.0261,
+      "step": 60425
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.442782804369228e-05,
+      "loss": 0.0415,
+      "step": 60430
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.4427359242417141e-05,
+      "loss": 0.0852,
+      "step": 60435
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.4426890441142001e-05,
+      "loss": 0.0841,
+      "step": 60440
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.4426421639866861e-05,
+      "loss": 0.0815,
+      "step": 60445
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.4425952838591721e-05,
+      "loss": 0.0924,
+      "step": 60450
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.4425484037316581e-05,
+      "loss": 0.1154,
+      "step": 60455
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.4425015236041444e-05,
+      "loss": 0.2005,
+      "step": 60460
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.4424546434766304e-05,
+      "loss": 0.2531,
+      "step": 60465
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.4424077633491164e-05,
+      "loss": 0.0243,
+      "step": 60470
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.4423608832216026e-05,
+      "loss": 0.0358,
+      "step": 60475
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.4423140030940886e-05,
+      "loss": 0.0395,
+      "step": 60480
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.4422671229665746e-05,
+      "loss": 0.0441,
+      "step": 60485
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.4422202428390606e-05,
+      "loss": 0.0726,
+      "step": 60490
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.4421733627115465e-05,
+      "loss": 0.1384,
+      "step": 60495
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.4421264825840327e-05,
+      "loss": 0.1046,
+      "step": 60500
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.4420796024565189e-05,
+      "loss": 0.1901,
+      "step": 60505
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.4420327223290049e-05,
+      "loss": 0.3416,
+      "step": 60510
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.441985842201491e-05,
+      "loss": 0.2348,
+      "step": 60515
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.441938962073977e-05,
+      "loss": 0.0289,
+      "step": 60520
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.441892081946463e-05,
+      "loss": 0.0462,
+      "step": 60525
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.441845201818949e-05,
+      "loss": 0.0298,
+      "step": 60530
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.441798321691435e-05,
+      "loss": 0.0507,
+      "step": 60535
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.4417514415639212e-05,
+      "loss": 0.0689,
+      "step": 60540
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4417045614364072e-05,
+      "loss": 0.0854,
+      "step": 60545
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4416576813088933e-05,
+      "loss": 0.1096,
+      "step": 60550
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4416108011813795e-05,
+      "loss": 0.1565,
+      "step": 60555
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4415639210538655e-05,
+      "loss": 0.2587,
+      "step": 60560
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4415170409263515e-05,
+      "loss": 0.3275,
+      "step": 60565
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4414701607988375e-05,
+      "loss": 0.0329,
+      "step": 60570
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4414232806713235e-05,
+      "loss": 0.0129,
+      "step": 60575
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4413764005438096e-05,
+      "loss": 0.0224,
+      "step": 60580
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4413295204162956e-05,
+      "loss": 0.0823,
+      "step": 60585
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4412826402887816e-05,
+      "loss": 0.1211,
+      "step": 60590
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4412357601612676e-05,
+      "loss": 0.0709,
+      "step": 60595
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.441188880033754e-05,
+      "loss": 0.1255,
+      "step": 60600
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.44114199990624e-05,
+      "loss": 0.1404,
+      "step": 60605
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.441095119778726e-05,
+      "loss": 0.1606,
+      "step": 60610
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4410482396512119e-05,
+      "loss": 0.1291,
+      "step": 60615
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.441001359523698e-05,
+      "loss": 0.0979,
+      "step": 60620
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.440954479396184e-05,
+      "loss": 0.0407,
+      "step": 60625
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.44090759926867e-05,
+      "loss": 0.025,
+      "step": 60630
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.440860719141156e-05,
+      "loss": 0.0598,
+      "step": 60635
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4408138390136422e-05,
+      "loss": 0.0943,
+      "step": 60640
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4407669588861284e-05,
+      "loss": 0.1123,
+      "step": 60645
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4407200787586144e-05,
+      "loss": 0.133,
+      "step": 60650
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4406731986311004e-05,
+      "loss": 0.2083,
+      "step": 60655
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4406263185035865e-05,
+      "loss": 0.2574,
+      "step": 60660
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4405794383760725e-05,
+      "loss": 0.241,
+      "step": 60665
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4405325582485585e-05,
+      "loss": 0.0453,
+      "step": 60670
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4404856781210445e-05,
+      "loss": 0.0483,
+      "step": 60675
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4404387979935307e-05,
+      "loss": 0.0555,
+      "step": 60680
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4403919178660167e-05,
+      "loss": 0.0621,
+      "step": 60685
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4403450377385028e-05,
+      "loss": 0.0828,
+      "step": 60690
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4402981576109888e-05,
+      "loss": 0.0364,
+      "step": 60695
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.440251277483475e-05,
+      "loss": 0.0896,
+      "step": 60700
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.440204397355961e-05,
+      "loss": 0.2197,
+      "step": 60705
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.440157517228447e-05,
+      "loss": 0.2638,
+      "step": 60710
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.440110637100933e-05,
+      "loss": 0.1872,
+      "step": 60715
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4400637569734191e-05,
+      "loss": 0.0346,
+      "step": 60720
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4400168768459051e-05,
+      "loss": 0.0323,
+      "step": 60725
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4399699967183911e-05,
+      "loss": 0.0482,
+      "step": 60730
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4399231165908771e-05,
+      "loss": 0.0482,
+      "step": 60735
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4398762364633634e-05,
+      "loss": 0.0727,
+      "step": 60740
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4398293563358494e-05,
+      "loss": 0.0952,
+      "step": 60745
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4397824762083354e-05,
+      "loss": 0.1124,
+      "step": 60750
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4397355960808214e-05,
+      "loss": 0.0838,
+      "step": 60755
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4396887159533076e-05,
+      "loss": 0.1974,
+      "step": 60760
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4396418358257936e-05,
+      "loss": 0.2214,
+      "step": 60765
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4395949556982796e-05,
+      "loss": 0.0473,
+      "step": 60770
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4395480755707656e-05,
+      "loss": 0.0588,
+      "step": 60775
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4395011954432516e-05,
+      "loss": 0.1081,
+      "step": 60780
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4394543153157379e-05,
+      "loss": 0.0593,
+      "step": 60785
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4394074351882239e-05,
+      "loss": 0.0878,
+      "step": 60790
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4393605550607099e-05,
+      "loss": 0.135,
+      "step": 60795
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.439313674933196e-05,
+      "loss": 0.0865,
+      "step": 60800
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.439266794805682e-05,
+      "loss": 0.1214,
+      "step": 60805
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.439219914678168e-05,
+      "loss": 0.2504,
+      "step": 60810
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.439173034550654e-05,
+      "loss": 0.4163,
+      "step": 60815
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.43912615442314e-05,
+      "loss": 0.0759,
+      "step": 60820
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4390792742956262e-05,
+      "loss": 0.0468,
+      "step": 60825
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4390323941681123e-05,
+      "loss": 0.0518,
+      "step": 60830
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4389855140405983e-05,
+      "loss": 0.053,
+      "step": 60835
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4389386339130845e-05,
+      "loss": 0.0998,
+      "step": 60840
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4388917537855705e-05,
+      "loss": 0.111,
+      "step": 60845
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4388448736580565e-05,
+      "loss": 0.1585,
+      "step": 60850
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4387979935305425e-05,
+      "loss": 0.1454,
+      "step": 60855
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4387511134030285e-05,
+      "loss": 0.2594,
+      "step": 60860
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4387042332755146e-05,
+      "loss": 0.1849,
+      "step": 60865
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4386573531480006e-05,
+      "loss": 0.019,
+      "step": 60870
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4386104730204868e-05,
+      "loss": 0.0433,
+      "step": 60875
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.438563592892973e-05,
+      "loss": 0.0268,
+      "step": 60880
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.438516712765459e-05,
+      "loss": 0.0626,
+      "step": 60885
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.438469832637945e-05,
+      "loss": 0.0737,
+      "step": 60890
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.438422952510431e-05,
+      "loss": 0.076,
+      "step": 60895
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.438376072382917e-05,
+      "loss": 0.2067,
+      "step": 60900
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.438329192255403e-05,
+      "loss": 0.1042,
+      "step": 60905
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.438282312127889e-05,
+      "loss": 0.2718,
+      "step": 60910
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.438235432000375e-05,
+      "loss": 0.3147,
+      "step": 60915
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.438188551872861e-05,
+      "loss": 0.0184,
+      "step": 60920
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4381416717453474e-05,
+      "loss": 0.0486,
+      "step": 60925
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4380947916178334e-05,
+      "loss": 0.0481,
+      "step": 60930
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4380479114903194e-05,
+      "loss": 0.0807,
+      "step": 60935
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4380010313628054e-05,
+      "loss": 0.0744,
+      "step": 60940
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4379541512352915e-05,
+      "loss": 0.0737,
+      "step": 60945
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4379072711077775e-05,
+      "loss": 0.1207,
+      "step": 60950
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4378603909802635e-05,
+      "loss": 0.1458,
+      "step": 60955
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4378135108527495e-05,
+      "loss": 0.2767,
+      "step": 60960
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4377666307252355e-05,
+      "loss": 0.2507,
+      "step": 60965
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4377197505977218e-05,
+      "loss": 0.064,
+      "step": 60970
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4376728704702078e-05,
+      "loss": 0.0188,
+      "step": 60975
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4376259903426938e-05,
+      "loss": 0.0876,
+      "step": 60980
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.43757911021518e-05,
+      "loss": 0.0304,
+      "step": 60985
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.437532230087666e-05,
+      "loss": 0.1134,
+      "step": 60990
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.437485349960152e-05,
+      "loss": 0.1214,
+      "step": 60995
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.437438469832638e-05,
+      "loss": 0.1877,
+      "step": 61000
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.437391589705124e-05,
+      "loss": 0.1703,
+      "step": 61005
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4373447095776101e-05,
+      "loss": 0.2195,
+      "step": 61010
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4372978294500963e-05,
+      "loss": 0.354,
+      "step": 61015
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4372509493225823e-05,
+      "loss": 0.0335,
+      "step": 61020
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4372040691950684e-05,
+      "loss": 0.0675,
+      "step": 61025
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4371571890675544e-05,
+      "loss": 0.0821,
+      "step": 61030
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4371103089400404e-05,
+      "loss": 0.0535,
+      "step": 61035
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4370634288125264e-05,
+      "loss": 0.1184,
+      "step": 61040
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4370165486850124e-05,
+      "loss": 0.0816,
+      "step": 61045
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4369696685574986e-05,
+      "loss": 0.0924,
+      "step": 61050
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4369227884299846e-05,
+      "loss": 0.1121,
+      "step": 61055
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4368759083024706e-05,
+      "loss": 0.2305,
+      "step": 61060
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4368290281749569e-05,
+      "loss": 0.413,
+      "step": 61065
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4367821480474429e-05,
+      "loss": 0.0227,
+      "step": 61070
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4367352679199289e-05,
+      "loss": 0.0205,
+      "step": 61075
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4366883877924149e-05,
+      "loss": 0.0373,
+      "step": 61080
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4366415076649009e-05,
+      "loss": 0.1037,
+      "step": 61085
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.436594627537387e-05,
+      "loss": 0.0472,
+      "step": 61090
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.436547747409873e-05,
+      "loss": 0.1188,
+      "step": 61095
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.436500867282359e-05,
+      "loss": 0.0784,
+      "step": 61100
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.436453987154845e-05,
+      "loss": 0.158,
+      "step": 61105
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4364071070273314e-05,
+      "loss": 0.192,
+      "step": 61110
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4363602268998173e-05,
+      "loss": 0.2495,
+      "step": 61115
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4363133467723033e-05,
+      "loss": 0.077,
+      "step": 61120
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4362664666447893e-05,
+      "loss": 0.063,
+      "step": 61125
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4362195865172755e-05,
+      "loss": 0.037,
+      "step": 61130
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4361727063897615e-05,
+      "loss": 0.0352,
+      "step": 61135
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4361258262622475e-05,
+      "loss": 0.0903,
+      "step": 61140
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4360789461347335e-05,
+      "loss": 0.0915,
+      "step": 61145
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4360320660072196e-05,
+      "loss": 0.1707,
+      "step": 61150
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4359851858797058e-05,
+      "loss": 0.1028,
+      "step": 61155
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4359383057521918e-05,
+      "loss": 0.2532,
+      "step": 61160
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4358914256246778e-05,
+      "loss": 0.307,
+      "step": 61165
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.435844545497164e-05,
+      "loss": 0.0508,
+      "step": 61170
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.43579766536965e-05,
+      "loss": 0.0748,
+      "step": 61175
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.435750785242136e-05,
+      "loss": 0.0633,
+      "step": 61180
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.435703905114622e-05,
+      "loss": 0.0856,
+      "step": 61185
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4356570249871081e-05,
+      "loss": 0.0617,
+      "step": 61190
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4356101448595941e-05,
+      "loss": 0.1655,
+      "step": 61195
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4355632647320802e-05,
+      "loss": 0.2037,
+      "step": 61200
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4355163846045662e-05,
+      "loss": 0.2147,
+      "step": 61205
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4354695044770524e-05,
+      "loss": 0.2051,
+      "step": 61210
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4354226243495384e-05,
+      "loss": 0.4263,
+      "step": 61215
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4353757442220244e-05,
+      "loss": 0.0426,
+      "step": 61220
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4353288640945104e-05,
+      "loss": 0.0636,
+      "step": 61225
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4352819839669965e-05,
+      "loss": 0.0451,
+      "step": 61230
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4352351038394825e-05,
+      "loss": 0.0919,
+      "step": 61235
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4351882237119685e-05,
+      "loss": 0.069,
+      "step": 61240
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4351413435844545e-05,
+      "loss": 0.0685,
+      "step": 61245
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4350944634569409e-05,
+      "loss": 0.0752,
+      "step": 61250
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4350475833294269e-05,
+      "loss": 0.1631,
+      "step": 61255
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4350007032019128e-05,
+      "loss": 0.2122,
+      "step": 61260
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4349538230743988e-05,
+      "loss": 0.251,
+      "step": 61265
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.434906942946885e-05,
+      "loss": 0.07,
+      "step": 61270
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.434860062819371e-05,
+      "loss": 0.0372,
+      "step": 61275
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.434813182691857e-05,
+      "loss": 0.035,
+      "step": 61280
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.434766302564343e-05,
+      "loss": 0.0704,
+      "step": 61285
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.434719422436829e-05,
+      "loss": 0.0953,
+      "step": 61290
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4346725423093153e-05,
+      "loss": 0.0836,
+      "step": 61295
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4346256621818013e-05,
+      "loss": 0.0872,
+      "step": 61300
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4345787820542873e-05,
+      "loss": 0.1247,
+      "step": 61305
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4345319019267735e-05,
+      "loss": 0.1821,
+      "step": 61310
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4344850217992595e-05,
+      "loss": 0.2211,
+      "step": 61315
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4344381416717454e-05,
+      "loss": 0.0347,
+      "step": 61320
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4343912615442314e-05,
+      "loss": 0.059,
+      "step": 61325
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4343443814167174e-05,
+      "loss": 0.0936,
+      "step": 61330
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4342975012892036e-05,
+      "loss": 0.0936,
+      "step": 61335
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4342506211616898e-05,
+      "loss": 0.0566,
+      "step": 61340
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4342037410341758e-05,
+      "loss": 0.1761,
+      "step": 61345
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4341568609066619e-05,
+      "loss": 0.1749,
+      "step": 61350
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4341099807791479e-05,
+      "loss": 0.1305,
+      "step": 61355
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4340631006516339e-05,
+      "loss": 0.2353,
+      "step": 61360
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4340162205241199e-05,
+      "loss": 0.2172,
+      "step": 61365
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4339693403966059e-05,
+      "loss": 0.0978,
+      "step": 61370
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.433922460269092e-05,
+      "loss": 0.06,
+      "step": 61375
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.433875580141578e-05,
+      "loss": 0.0539,
+      "step": 61380
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.433828700014064e-05,
+      "loss": 0.0406,
+      "step": 61385
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4337818198865504e-05,
+      "loss": 0.0755,
+      "step": 61390
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4337349397590364e-05,
+      "loss": 0.119,
+      "step": 61395
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4336880596315224e-05,
+      "loss": 0.1196,
+      "step": 61400
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4336411795040083e-05,
+      "loss": 0.0995,
+      "step": 61405
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4335942993764943e-05,
+      "loss": 0.1492,
+      "step": 61410
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4335474192489805e-05,
+      "loss": 0.2199,
+      "step": 61415
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4335005391214665e-05,
+      "loss": 0.0173,
+      "step": 61420
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4334536589939525e-05,
+      "loss": 0.034,
+      "step": 61425
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4334067788664385e-05,
+      "loss": 0.0801,
+      "step": 61430
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4333598987389248e-05,
+      "loss": 0.1636,
+      "step": 61435
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4333130186114108e-05,
+      "loss": 0.0911,
+      "step": 61440
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4332661384838968e-05,
+      "loss": 0.0715,
+      "step": 61445
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4332192583563828e-05,
+      "loss": 0.1519,
+      "step": 61450
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.433172378228869e-05,
+      "loss": 0.1616,
+      "step": 61455
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.433125498101355e-05,
+      "loss": 0.2893,
+      "step": 61460
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.433078617973841e-05,
+      "loss": 0.2141,
+      "step": 61465
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.433031737846327e-05,
+      "loss": 0.0291,
+      "step": 61470
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.432984857718813e-05,
+      "loss": 0.0565,
+      "step": 61475
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4329379775912993e-05,
+      "loss": 0.0623,
+      "step": 61480
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4328910974637853e-05,
+      "loss": 0.1416,
+      "step": 61485
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4328442173362713e-05,
+      "loss": 0.0452,
+      "step": 61490
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4327973372087574e-05,
+      "loss": 0.0532,
+      "step": 61495
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4327504570812434e-05,
+      "loss": 0.1677,
+      "step": 61500
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4327035769537294e-05,
+      "loss": 0.1528,
+      "step": 61505
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4326566968262154e-05,
+      "loss": 0.2106,
+      "step": 61510
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4326098166987014e-05,
+      "loss": 0.2591,
+      "step": 61515
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4325629365711876e-05,
+      "loss": 0.085,
+      "step": 61520
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4325160564436737e-05,
+      "loss": 0.0254,
+      "step": 61525
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4324691763161597e-05,
+      "loss": 0.0941,
+      "step": 61530
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4324222961886459e-05,
+      "loss": 0.051,
+      "step": 61535
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4323754160611319e-05,
+      "loss": 0.0773,
+      "step": 61540
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4323285359336179e-05,
+      "loss": 0.1572,
+      "step": 61545
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4322816558061039e-05,
+      "loss": 0.1354,
+      "step": 61550
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4322347756785898e-05,
+      "loss": 0.1235,
+      "step": 61555
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.432187895551076e-05,
+      "loss": 0.3151,
+      "step": 61560
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.432141015423562e-05,
+      "loss": 0.1998,
+      "step": 61565
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.432094135296048e-05,
+      "loss": 0.0537,
+      "step": 61570
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4320472551685343e-05,
+      "loss": 0.0522,
+      "step": 61575
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4320003750410203e-05,
+      "loss": 0.0644,
+      "step": 61580
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4319534949135063e-05,
+      "loss": 0.0724,
+      "step": 61585
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4319066147859923e-05,
+      "loss": 0.0751,
+      "step": 61590
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4318597346584785e-05,
+      "loss": 0.0635,
+      "step": 61595
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4318128545309645e-05,
+      "loss": 0.0984,
+      "step": 61600
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4317659744034505e-05,
+      "loss": 0.2148,
+      "step": 61605
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4317190942759364e-05,
+      "loss": 0.2069,
+      "step": 61610
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4316722141484224e-05,
+      "loss": 0.1928,
+      "step": 61615
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4316253340209088e-05,
+      "loss": 0.0248,
+      "step": 61620
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4315784538933948e-05,
+      "loss": 0.0543,
+      "step": 61625
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4315315737658808e-05,
+      "loss": 0.0697,
+      "step": 61630
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.431484693638367e-05,
+      "loss": 0.067,
+      "step": 61635
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.431437813510853e-05,
+      "loss": 0.0684,
+      "step": 61640
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4313909333833389e-05,
+      "loss": 0.0666,
+      "step": 61645
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4313440532558249e-05,
+      "loss": 0.1475,
+      "step": 61650
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4312971731283109e-05,
+      "loss": 0.0815,
+      "step": 61655
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.431250293000797e-05,
+      "loss": 0.432,
+      "step": 61660
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4312034128732832e-05,
+      "loss": 0.1714,
+      "step": 61665
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4311565327457692e-05,
+      "loss": 0.0722,
+      "step": 61670
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4311096526182554e-05,
+      "loss": 0.1003,
+      "step": 61675
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4310627724907414e-05,
+      "loss": 0.0671,
+      "step": 61680
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4310158923632274e-05,
+      "loss": 0.0371,
+      "step": 61685
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4309690122357134e-05,
+      "loss": 0.0837,
+      "step": 61690
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4309221321081994e-05,
+      "loss": 0.1399,
+      "step": 61695
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4308752519806855e-05,
+      "loss": 0.0965,
+      "step": 61700
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4308283718531715e-05,
+      "loss": 0.1974,
+      "step": 61705
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4307814917256575e-05,
+      "loss": 0.1911,
+      "step": 61710
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4307346115981438e-05,
+      "loss": 0.2249,
+      "step": 61715
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4306877314706298e-05,
+      "loss": 0.043,
+      "step": 61720
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4306408513431158e-05,
+      "loss": 0.0711,
+      "step": 61725
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4305939712156018e-05,
+      "loss": 0.0502,
+      "step": 61730
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4305470910880878e-05,
+      "loss": 0.0629,
+      "step": 61735
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.430500210960574e-05,
+      "loss": 0.0486,
+      "step": 61740
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.43045333083306e-05,
+      "loss": 0.111,
+      "step": 61745
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.430406450705546e-05,
+      "loss": 0.0744,
+      "step": 61750
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.430359570578032e-05,
+      "loss": 0.1504,
+      "step": 61755
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4303126904505183e-05,
+      "loss": 0.165,
+      "step": 61760
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4302658103230043e-05,
+      "loss": 0.3405,
+      "step": 61765
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4302189301954903e-05,
+      "loss": 0.0162,
+      "step": 61770
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4301720500679763e-05,
+      "loss": 0.0862,
+      "step": 61775
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4301251699404624e-05,
+      "loss": 0.0322,
+      "step": 61780
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4300782898129484e-05,
+      "loss": 0.048,
+      "step": 61785
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4300314096854344e-05,
+      "loss": 0.0931,
+      "step": 61790
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4299845295579204e-05,
+      "loss": 0.0647,
+      "step": 61795
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4299376494304064e-05,
+      "loss": 0.0786,
+      "step": 61800
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4298907693028927e-05,
+      "loss": 0.2143,
+      "step": 61805
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4298438891753787e-05,
+      "loss": 0.1282,
+      "step": 61810
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4297970090478647e-05,
+      "loss": 0.3249,
+      "step": 61815
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4297501289203509e-05,
+      "loss": 0.0299,
+      "step": 61820
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4297032487928369e-05,
+      "loss": 0.0226,
+      "step": 61825
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4296563686653229e-05,
+      "loss": 0.0279,
+      "step": 61830
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4296094885378089e-05,
+      "loss": 0.0813,
+      "step": 61835
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4295626084102949e-05,
+      "loss": 0.0733,
+      "step": 61840
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.429515728282781e-05,
+      "loss": 0.0613,
+      "step": 61845
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4294688481552672e-05,
+      "loss": 0.172,
+      "step": 61850
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4294219680277532e-05,
+      "loss": 0.2214,
+      "step": 61855
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4293750879002393e-05,
+      "loss": 0.1672,
+      "step": 61860
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4293282077727253e-05,
+      "loss": 0.207,
+      "step": 61865
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4292813276452113e-05,
+      "loss": 0.0749,
+      "step": 61870
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4292344475176973e-05,
+      "loss": 0.0837,
+      "step": 61875
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4291875673901833e-05,
+      "loss": 0.0423,
+      "step": 61880
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4291406872626695e-05,
+      "loss": 0.0607,
+      "step": 61885
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4290938071351555e-05,
+      "loss": 0.073,
+      "step": 61890
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4290469270076415e-05,
+      "loss": 0.1784,
+      "step": 61895
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4290000468801278e-05,
+      "loss": 0.1983,
+      "step": 61900
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4289531667526138e-05,
+      "loss": 0.1165,
+      "step": 61905
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4289062866250998e-05,
+      "loss": 0.246,
+      "step": 61910
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4288594064975858e-05,
+      "loss": 0.4182,
+      "step": 61915
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4288125263700718e-05,
+      "loss": 0.0554,
+      "step": 61920
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.428765646242558e-05,
+      "loss": 0.0374,
+      "step": 61925
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.428718766115044e-05,
+      "loss": 0.0909,
+      "step": 61930
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4286718859875299e-05,
+      "loss": 0.0459,
+      "step": 61935
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4286250058600159e-05,
+      "loss": 0.0981,
+      "step": 61940
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4285781257325022e-05,
+      "loss": 0.173,
+      "step": 61945
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4285312456049882e-05,
+      "loss": 0.1568,
+      "step": 61950
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4284843654774742e-05,
+      "loss": 0.1952,
+      "step": 61955
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4284374853499602e-05,
+      "loss": 0.2491,
+      "step": 61960
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4283906052224464e-05,
+      "loss": 0.3357,
+      "step": 61965
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4283437250949324e-05,
+      "loss": 0.0521,
+      "step": 61970
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4282968449674184e-05,
+      "loss": 0.0331,
+      "step": 61975
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4282499648399044e-05,
+      "loss": 0.0295,
+      "step": 61980
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4282030847123905e-05,
+      "loss": 0.0363,
+      "step": 61985
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4281562045848767e-05,
+      "loss": 0.1385,
+      "step": 61990
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4281093244573627e-05,
+      "loss": 0.0938,
+      "step": 61995
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4280624443298487e-05,
+      "loss": 0.041,
+      "step": 62000
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4280155642023348e-05,
+      "loss": 0.1099,
+      "step": 62005
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4279686840748208e-05,
+      "loss": 0.2192,
+      "step": 62010
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4279218039473068e-05,
+      "loss": 0.203,
+      "step": 62015
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4278749238197928e-05,
+      "loss": 0.0887,
+      "step": 62020
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.427828043692279e-05,
+      "loss": 0.0572,
+      "step": 62025
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.427781163564765e-05,
+      "loss": 0.0779,
+      "step": 62030
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.427734283437251e-05,
+      "loss": 0.0652,
+      "step": 62035
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4276874033097371e-05,
+      "loss": 0.096,
+      "step": 62040
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4276405231822233e-05,
+      "loss": 0.0918,
+      "step": 62045
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4275936430547093e-05,
+      "loss": 0.1491,
+      "step": 62050
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4275467629271953e-05,
+      "loss": 0.0714,
+      "step": 62055
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4274998827996813e-05,
+      "loss": 0.2989,
+      "step": 62060
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4274530026721674e-05,
+      "loss": 0.151,
+      "step": 62065
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4274061225446534e-05,
+      "loss": 0.3291,
+      "step": 62070
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4273592424171394e-05,
+      "loss": 0.0823,
+      "step": 62075
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4273123622896254e-05,
+      "loss": 0.0368,
+      "step": 62080
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4272654821621117e-05,
+      "loss": 0.0819,
+      "step": 62085
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4272186020345977e-05,
+      "loss": 0.1533,
+      "step": 62090
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4271717219070837e-05,
+      "loss": 0.0694,
+      "step": 62095
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4271248417795697e-05,
+      "loss": 0.1272,
+      "step": 62100
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4270779616520559e-05,
+      "loss": 0.1179,
+      "step": 62105
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4270310815245419e-05,
+      "loss": 0.1788,
+      "step": 62110
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4269842013970279e-05,
+      "loss": 0.2079,
+      "step": 62115
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4269373212695139e-05,
+      "loss": 0.0381,
+      "step": 62120
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4268904411419999e-05,
+      "loss": 0.0527,
+      "step": 62125
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4268435610144862e-05,
+      "loss": 0.0433,
+      "step": 62130
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4267966808869722e-05,
+      "loss": 0.0685,
+      "step": 62135
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4267498007594582e-05,
+      "loss": 0.0537,
+      "step": 62140
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4267029206319443e-05,
+      "loss": 0.0927,
+      "step": 62145
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4266560405044303e-05,
+      "loss": 0.1351,
+      "step": 62150
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4266091603769163e-05,
+      "loss": 0.0956,
+      "step": 62155
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4265622802494023e-05,
+      "loss": 0.1689,
+      "step": 62160
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4265154001218883e-05,
+      "loss": 0.2929,
+      "step": 62165
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4264685199943745e-05,
+      "loss": 0.0412,
+      "step": 62170
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4264216398668606e-05,
+      "loss": 0.0285,
+      "step": 62175
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4263747597393466e-05,
+      "loss": 0.0957,
+      "step": 62180
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4263278796118328e-05,
+      "loss": 0.0749,
+      "step": 62185
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4262809994843188e-05,
+      "loss": 0.11,
+      "step": 62190
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4262341193568048e-05,
+      "loss": 0.0803,
+      "step": 62195
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4261872392292908e-05,
+      "loss": 0.0942,
+      "step": 62200
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4261403591017768e-05,
+      "loss": 0.1192,
+      "step": 62205
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.426093478974263e-05,
+      "loss": 0.2655,
+      "step": 62210
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.426046598846749e-05,
+      "loss": 0.2694,
+      "step": 62215
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.425999718719235e-05,
+      "loss": 0.0398,
+      "step": 62220
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4259528385917213e-05,
+      "loss": 0.0649,
+      "step": 62225
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4259059584642072e-05,
+      "loss": 0.0913,
+      "step": 62230
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4258590783366932e-05,
+      "loss": 0.1374,
+      "step": 62235
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4258121982091792e-05,
+      "loss": 0.0531,
+      "step": 62240
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4257653180816652e-05,
+      "loss": 0.0716,
+      "step": 62245
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4257184379541514e-05,
+      "loss": 0.1004,
+      "step": 62250
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4256715578266374e-05,
+      "loss": 0.137,
+      "step": 62255
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4256246776991234e-05,
+      "loss": 0.262,
+      "step": 62260
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4255777975716094e-05,
+      "loss": 0.2588,
+      "step": 62265
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4255309174440957e-05,
+      "loss": 0.022,
+      "step": 62270
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4254840373165817e-05,
+      "loss": 0.0912,
+      "step": 62275
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4254371571890677e-05,
+      "loss": 0.0856,
+      "step": 62280
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4253902770615537e-05,
+      "loss": 0.0125,
+      "step": 62285
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4253433969340398e-05,
+      "loss": 0.0782,
+      "step": 62290
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4252965168065258e-05,
+      "loss": 0.1191,
+      "step": 62295
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4252496366790118e-05,
+      "loss": 0.1224,
+      "step": 62300
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4252027565514978e-05,
+      "loss": 0.1303,
+      "step": 62305
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4251558764239838e-05,
+      "loss": 0.2055,
+      "step": 62310
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4251089962964702e-05,
+      "loss": 0.2296,
+      "step": 62315
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4250621161689561e-05,
+      "loss": 0.0625,
+      "step": 62320
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4250152360414421e-05,
+      "loss": 0.0877,
+      "step": 62325
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4249683559139283e-05,
+      "loss": 0.0508,
+      "step": 62330
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4249214757864143e-05,
+      "loss": 0.0208,
+      "step": 62335
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4248745956589003e-05,
+      "loss": 0.066,
+      "step": 62340
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4248277155313863e-05,
+      "loss": 0.1559,
+      "step": 62345
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4247808354038723e-05,
+      "loss": 0.1327,
+      "step": 62350
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4247339552763584e-05,
+      "loss": 0.1594,
+      "step": 62355
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4246870751488444e-05,
+      "loss": 0.2158,
+      "step": 62360
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4246401950213306e-05,
+      "loss": 0.4769,
+      "step": 62365
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4245933148938168e-05,
+      "loss": 0.0163,
+      "step": 62370
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4245464347663028e-05,
+      "loss": 0.0447,
+      "step": 62375
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4244995546387887e-05,
+      "loss": 0.0663,
+      "step": 62380
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4244526745112747e-05,
+      "loss": 0.0934,
+      "step": 62385
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4244057943837607e-05,
+      "loss": 0.0503,
+      "step": 62390
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4243589142562469e-05,
+      "loss": 0.1216,
+      "step": 62395
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4243120341287329e-05,
+      "loss": 0.1094,
+      "step": 62400
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4242651540012189e-05,
+      "loss": 0.1975,
+      "step": 62405
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4242182738737052e-05,
+      "loss": 0.2689,
+      "step": 62410
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4241713937461912e-05,
+      "loss": 0.338,
+      "step": 62415
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4241245136186772e-05,
+      "loss": 0.0438,
+      "step": 62420
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4240776334911632e-05,
+      "loss": 0.0389,
+      "step": 62425
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4240307533636492e-05,
+      "loss": 0.0787,
+      "step": 62430
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4239838732361353e-05,
+      "loss": 0.0889,
+      "step": 62435
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4239369931086213e-05,
+      "loss": 0.0315,
+      "step": 62440
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4238901129811073e-05,
+      "loss": 0.0751,
+      "step": 62445
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4238432328535933e-05,
+      "loss": 0.116,
+      "step": 62450
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4237963527260797e-05,
+      "loss": 0.1544,
+      "step": 62455
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4237494725985657e-05,
+      "loss": 0.1478,
+      "step": 62460
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4237025924710516e-05,
+      "loss": 0.2487,
+      "step": 62465
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4236557123435376e-05,
+      "loss": 0.0402,
+      "step": 62470
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4236088322160238e-05,
+      "loss": 0.0696,
+      "step": 62475
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4235619520885098e-05,
+      "loss": 0.0629,
+      "step": 62480
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4235150719609958e-05,
+      "loss": 0.0973,
+      "step": 62485
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4234681918334818e-05,
+      "loss": 0.0445,
+      "step": 62490
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.423421311705968e-05,
+      "loss": 0.1361,
+      "step": 62495
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4233744315784541e-05,
+      "loss": 0.1161,
+      "step": 62500
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4233275514509401e-05,
+      "loss": 0.185,
+      "step": 62505
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4232806713234261e-05,
+      "loss": 0.4291,
+      "step": 62510
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4232337911959123e-05,
+      "loss": 0.2839,
+      "step": 62515
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4231869110683983e-05,
+      "loss": 0.0522,
+      "step": 62520
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4231400309408842e-05,
+      "loss": 0.0422,
+      "step": 62525
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4230931508133702e-05,
+      "loss": 0.115,
+      "step": 62530
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4230462706858564e-05,
+      "loss": 0.0269,
+      "step": 62535
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4229993905583424e-05,
+      "loss": 0.0985,
+      "step": 62540
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4229525104308284e-05,
+      "loss": 0.0298,
+      "step": 62545
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4229056303033146e-05,
+      "loss": 0.0855,
+      "step": 62550
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4228587501758007e-05,
+      "loss": 0.1481,
+      "step": 62555
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4228118700482867e-05,
+      "loss": 0.0833,
+      "step": 62560
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4227649899207727e-05,
+      "loss": 0.2215,
+      "step": 62565
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4227181097932587e-05,
+      "loss": 0.0178,
+      "step": 62570
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4226712296657449e-05,
+      "loss": 0.0319,
+      "step": 62575
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4226243495382308e-05,
+      "loss": 0.085,
+      "step": 62580
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4225774694107168e-05,
+      "loss": 0.0879,
+      "step": 62585
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4225305892832028e-05,
+      "loss": 0.0685,
+      "step": 62590
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4224837091556892e-05,
+      "loss": 0.0989,
+      "step": 62595
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4224368290281752e-05,
+      "loss": 0.0491,
+      "step": 62600
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4223899489006612e-05,
+      "loss": 0.1561,
+      "step": 62605
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4223430687731471e-05,
+      "loss": 0.2575,
+      "step": 62610
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4222961886456333e-05,
+      "loss": 0.1737,
+      "step": 62615
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4222493085181193e-05,
+      "loss": 0.0793,
+      "step": 62620
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4222024283906053e-05,
+      "loss": 0.0559,
+      "step": 62625
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4221555482630913e-05,
+      "loss": 0.0621,
+      "step": 62630
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4221086681355773e-05,
+      "loss": 0.1113,
+      "step": 62635
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4220617880080636e-05,
+      "loss": 0.0519,
+      "step": 62640
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4220149078805496e-05,
+      "loss": 0.1117,
+      "step": 62645
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4219680277530356e-05,
+      "loss": 0.1568,
+      "step": 62650
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4219211476255218e-05,
+      "loss": 0.1619,
+      "step": 62655
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4218742674980078e-05,
+      "loss": 0.2119,
+      "step": 62660
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4218273873704938e-05,
+      "loss": 0.1848,
+      "step": 62665
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4217805072429797e-05,
+      "loss": 0.0474,
+      "step": 62670
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4217336271154657e-05,
+      "loss": 0.0365,
+      "step": 62675
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4216867469879519e-05,
+      "loss": 0.0216,
+      "step": 62680
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4216398668604379e-05,
+      "loss": 0.053,
+      "step": 62685
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.421592986732924e-05,
+      "loss": 0.1091,
+      "step": 62690
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4215461066054102e-05,
+      "loss": 0.1586,
+      "step": 62695
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4214992264778962e-05,
+      "loss": 0.0618,
+      "step": 62700
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4214523463503822e-05,
+      "loss": 0.2021,
+      "step": 62705
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4214054662228682e-05,
+      "loss": 0.2014,
+      "step": 62710
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4213585860953542e-05,
+      "loss": 0.24,
+      "step": 62715
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4213117059678404e-05,
+      "loss": 0.0265,
+      "step": 62720
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4212648258403264e-05,
+      "loss": 0.0587,
+      "step": 62725
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4212179457128123e-05,
+      "loss": 0.0479,
+      "step": 62730
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4211710655852987e-05,
+      "loss": 0.1283,
+      "step": 62735
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4211241854577847e-05,
+      "loss": 0.0703,
+      "step": 62740
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4210773053302707e-05,
+      "loss": 0.0843,
+      "step": 62745
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4210304252027567e-05,
+      "loss": 0.0996,
+      "step": 62750
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4209835450752427e-05,
+      "loss": 0.2006,
+      "step": 62755
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4209366649477288e-05,
+      "loss": 0.3242,
+      "step": 62760
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4208897848202148e-05,
+      "loss": 0.391,
+      "step": 62765
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4208429046927008e-05,
+      "loss": 0.0666,
+      "step": 62770
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4207960245651868e-05,
+      "loss": 0.0281,
+      "step": 62775
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4207491444376731e-05,
+      "loss": 0.1492,
+      "step": 62780
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4207022643101591e-05,
+      "loss": 0.0293,
+      "step": 62785
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4206553841826451e-05,
+      "loss": 0.0691,
+      "step": 62790
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4206085040551311e-05,
+      "loss": 0.0777,
+      "step": 62795
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4205616239276173e-05,
+      "loss": 0.2008,
+      "step": 62800
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4205147438001033e-05,
+      "loss": 0.1873,
+      "step": 62805
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4204678636725893e-05,
+      "loss": 0.2477,
+      "step": 62810
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4204209835450752e-05,
+      "loss": 0.2355,
+      "step": 62815
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4203741034175612e-05,
+      "loss": 0.0399,
+      "step": 62820
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4203272232900476e-05,
+      "loss": 0.05,
+      "step": 62825
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4202803431625336e-05,
+      "loss": 0.0489,
+      "step": 62830
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4202334630350196e-05,
+      "loss": 0.0676,
+      "step": 62835
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4201865829075057e-05,
+      "loss": 0.0944,
+      "step": 62840
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4201397027799917e-05,
+      "loss": 0.1006,
+      "step": 62845
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4200928226524777e-05,
+      "loss": 0.1669,
+      "step": 62850
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4200459425249637e-05,
+      "loss": 0.1534,
+      "step": 62855
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4199990623974497e-05,
+      "loss": 0.1898,
+      "step": 62860
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4199521822699359e-05,
+      "loss": 0.2592,
+      "step": 62865
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4199053021424219e-05,
+      "loss": 0.0166,
+      "step": 62870
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.419858422014908e-05,
+      "loss": 0.0357,
+      "step": 62875
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4198115418873942e-05,
+      "loss": 0.0386,
+      "step": 62880
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4197646617598802e-05,
+      "loss": 0.0523,
+      "step": 62885
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4197177816323662e-05,
+      "loss": 0.1094,
+      "step": 62890
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4196709015048522e-05,
+      "loss": 0.1759,
+      "step": 62895
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4196240213773382e-05,
+      "loss": 0.0962,
+      "step": 62900
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4195771412498243e-05,
+      "loss": 0.1184,
+      "step": 62905
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4195302611223103e-05,
+      "loss": 0.2146,
+      "step": 62910
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4194833809947963e-05,
+      "loss": 0.2689,
+      "step": 62915
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4194365008672826e-05,
+      "loss": 0.0281,
+      "step": 62920
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4193896207397686e-05,
+      "loss": 0.0596,
+      "step": 62925
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4193427406122546e-05,
+      "loss": 0.0905,
+      "step": 62930
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4192958604847406e-05,
+      "loss": 0.0953,
+      "step": 62935
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4192489803572266e-05,
+      "loss": 0.0945,
+      "step": 62940
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4192021002297128e-05,
+      "loss": 0.1268,
+      "step": 62945
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4191552201021988e-05,
+      "loss": 0.1404,
+      "step": 62950
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4191083399746848e-05,
+      "loss": 0.0784,
+      "step": 62955
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4190614598471708e-05,
+      "loss": 0.2586,
+      "step": 62960
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.419014579719657e-05,
+      "loss": 0.216,
+      "step": 62965
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.418967699592143e-05,
+      "loss": 0.0615,
+      "step": 62970
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.418920819464629e-05,
+      "loss": 0.0367,
+      "step": 62975
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.418873939337115e-05,
+      "loss": 0.07,
+      "step": 62980
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4188270592096012e-05,
+      "loss": 0.0625,
+      "step": 62985
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4187801790820872e-05,
+      "loss": 0.0718,
+      "step": 62990
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4187332989545732e-05,
+      "loss": 0.1033,
+      "step": 62995
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4186864188270592e-05,
+      "loss": 0.1077,
+      "step": 63000
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4186395386995454e-05,
+      "loss": 0.1611,
+      "step": 63005
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4185926585720314e-05,
+      "loss": 0.2172,
+      "step": 63010
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4185457784445175e-05,
+      "loss": 0.2392,
+      "step": 63015
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4184988983170037e-05,
+      "loss": 0.0908,
+      "step": 63020
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4184520181894897e-05,
+      "loss": 0.0453,
+      "step": 63025
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4184051380619757e-05,
+      "loss": 0.0317,
+      "step": 63030
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4183582579344617e-05,
+      "loss": 0.0595,
+      "step": 63035
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4183113778069477e-05,
+      "loss": 0.0727,
+      "step": 63040
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4182644976794338e-05,
+      "loss": 0.0852,
+      "step": 63045
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4182176175519198e-05,
+      "loss": 0.1202,
+      "step": 63050
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4181707374244058e-05,
+      "loss": 0.1415,
+      "step": 63055
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4181238572968921e-05,
+      "loss": 0.2081,
+      "step": 63060
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4180769771693781e-05,
+      "loss": 0.3448,
+      "step": 63065
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4180300970418641e-05,
+      "loss": 0.0215,
+      "step": 63070
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4179832169143501e-05,
+      "loss": 0.0511,
+      "step": 63075
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4179363367868361e-05,
+      "loss": 0.0203,
+      "step": 63080
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4178894566593223e-05,
+      "loss": 0.0664,
+      "step": 63085
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4178425765318083e-05,
+      "loss": 0.0897,
+      "step": 63090
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4177956964042943e-05,
+      "loss": 0.1132,
+      "step": 63095
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4177488162767803e-05,
+      "loss": 0.1802,
+      "step": 63100
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4177019361492666e-05,
+      "loss": 0.1214,
+      "step": 63105
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4176550560217526e-05,
+      "loss": 0.2478,
+      "step": 63110
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4176081758942386e-05,
+      "loss": 0.299,
+      "step": 63115
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4175612957667246e-05,
+      "loss": 0.0547,
+      "step": 63120
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4175144156392107e-05,
+      "loss": 0.0359,
+      "step": 63125
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4174675355116967e-05,
+      "loss": 0.0356,
+      "step": 63130
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4174206553841827e-05,
+      "loss": 0.0364,
+      "step": 63135
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4173737752566687e-05,
+      "loss": 0.0752,
+      "step": 63140
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4173268951291547e-05,
+      "loss": 0.0793,
+      "step": 63145
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.417280015001641e-05,
+      "loss": 0.1518,
+      "step": 63150
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.417233134874127e-05,
+      "loss": 0.2003,
+      "step": 63155
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.417186254746613e-05,
+      "loss": 0.2679,
+      "step": 63160
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4171393746190992e-05,
+      "loss": 0.3248,
+      "step": 63165
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4170924944915852e-05,
+      "loss": 0.0329,
+      "step": 63170
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4170456143640712e-05,
+      "loss": 0.0632,
+      "step": 63175
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4169987342365572e-05,
+      "loss": 0.1266,
+      "step": 63180
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4169518541090432e-05,
+      "loss": 0.0548,
+      "step": 63185
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4169049739815293e-05,
+      "loss": 0.0715,
+      "step": 63190
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4168580938540153e-05,
+      "loss": 0.0533,
+      "step": 63195
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4168112137265015e-05,
+      "loss": 0.1372,
+      "step": 63200
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4167643335989876e-05,
+      "loss": 0.119,
+      "step": 63205
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4167174534714736e-05,
+      "loss": 0.2117,
+      "step": 63210
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4166705733439596e-05,
+      "loss": 0.3411,
+      "step": 63215
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4166236932164456e-05,
+      "loss": 0.095,
+      "step": 63220
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4165768130889316e-05,
+      "loss": 0.0377,
+      "step": 63225
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4165299329614178e-05,
+      "loss": 0.042,
+      "step": 63230
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4164830528339038e-05,
+      "loss": 0.0263,
+      "step": 63235
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4164361727063898e-05,
+      "loss": 0.1153,
+      "step": 63240
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4163892925788761e-05,
+      "loss": 0.0538,
+      "step": 63245
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4163424124513621e-05,
+      "loss": 0.184,
+      "step": 63250
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4162955323238481e-05,
+      "loss": 0.145,
+      "step": 63255
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.416248652196334e-05,
+      "loss": 0.1811,
+      "step": 63260
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.41620177206882e-05,
+      "loss": 0.322,
+      "step": 63265
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4161548919413062e-05,
+      "loss": 0.0432,
+      "step": 63270
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4161080118137922e-05,
+      "loss": 0.0354,
+      "step": 63275
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4160611316862782e-05,
+      "loss": 0.0424,
+      "step": 63280
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4160142515587642e-05,
+      "loss": 0.0593,
+      "step": 63285
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4159673714312505e-05,
+      "loss": 0.0574,
+      "step": 63290
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4159204913037365e-05,
+      "loss": 0.0868,
+      "step": 63295
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4158736111762225e-05,
+      "loss": 0.1314,
+      "step": 63300
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4158267310487085e-05,
+      "loss": 0.085,
+      "step": 63305
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4157798509211947e-05,
+      "loss": 0.1749,
+      "step": 63310
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4157329707936807e-05,
+      "loss": 0.263,
+      "step": 63315
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4156860906661667e-05,
+      "loss": 0.0535,
+      "step": 63320
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4156392105386527e-05,
+      "loss": 0.0341,
+      "step": 63325
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4155923304111388e-05,
+      "loss": 0.0705,
+      "step": 63330
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4155454502836248e-05,
+      "loss": 0.0689,
+      "step": 63335
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.415498570156111e-05,
+      "loss": 0.0664,
+      "step": 63340
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.415451690028597e-05,
+      "loss": 0.0963,
+      "step": 63345
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4154048099010831e-05,
+      "loss": 0.1253,
+      "step": 63350
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4153579297735691e-05,
+      "loss": 0.1531,
+      "step": 63355
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4153110496460551e-05,
+      "loss": 0.3743,
+      "step": 63360
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4152641695185411e-05,
+      "loss": 0.3862,
+      "step": 63365
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4152172893910273e-05,
+      "loss": 0.0714,
+      "step": 63370
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4151704092635133e-05,
+      "loss": 0.0333,
+      "step": 63375
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4151235291359993e-05,
+      "loss": 0.0459,
+      "step": 63380
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4150766490084854e-05,
+      "loss": 0.024,
+      "step": 63385
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4150297688809716e-05,
+      "loss": 0.0285,
+      "step": 63390
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4149828887534576e-05,
+      "loss": 0.0494,
+      "step": 63395
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4149360086259436e-05,
+      "loss": 0.1061,
+      "step": 63400
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4148891284984296e-05,
+      "loss": 0.1774,
+      "step": 63405
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4148422483709157e-05,
+      "loss": 0.2098,
+      "step": 63410
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4147953682434017e-05,
+      "loss": 0.2257,
+      "step": 63415
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4147484881158877e-05,
+      "loss": 0.0405,
+      "step": 63420
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4147016079883737e-05,
+      "loss": 0.0221,
+      "step": 63425
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.41465472786086e-05,
+      "loss": 0.0382,
+      "step": 63430
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.414607847733346e-05,
+      "loss": 0.0391,
+      "step": 63435
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.414560967605832e-05,
+      "loss": 0.0743,
+      "step": 63440
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.414514087478318e-05,
+      "loss": 0.106,
+      "step": 63445
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4144672073508042e-05,
+      "loss": 0.1007,
+      "step": 63450
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4144203272232902e-05,
+      "loss": 0.1935,
+      "step": 63455
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4143734470957762e-05,
+      "loss": 0.1602,
+      "step": 63460
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4143265669682622e-05,
+      "loss": 0.3175,
+      "step": 63465
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4142796868407482e-05,
+      "loss": 0.0462,
+      "step": 63470
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4142328067132345e-05,
+      "loss": 0.0184,
+      "step": 63475
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4141859265857205e-05,
+      "loss": 0.027,
+      "step": 63480
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4141390464582065e-05,
+      "loss": 0.0426,
+      "step": 63485
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4140921663306927e-05,
+      "loss": 0.0563,
+      "step": 63490
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4140452862031786e-05,
+      "loss": 0.1309,
+      "step": 63495
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4139984060756646e-05,
+      "loss": 0.1223,
+      "step": 63500
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4139515259481506e-05,
+      "loss": 0.1788,
+      "step": 63505
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4139046458206366e-05,
+      "loss": 0.3226,
+      "step": 63510
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4138577656931228e-05,
+      "loss": 0.278,
+      "step": 63515
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4138108855656088e-05,
+      "loss": 0.0424,
+      "step": 63520
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.413764005438095e-05,
+      "loss": 0.0399,
+      "step": 63525
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4137171253105811e-05,
+      "loss": 0.0638,
+      "step": 63530
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4136702451830671e-05,
+      "loss": 0.0928,
+      "step": 63535
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4136233650555531e-05,
+      "loss": 0.0727,
+      "step": 63540
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4135764849280391e-05,
+      "loss": 0.1228,
+      "step": 63545
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.413529604800525e-05,
+      "loss": 0.1091,
+      "step": 63550
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4134827246730112e-05,
+      "loss": 0.0921,
+      "step": 63555
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4134358445454972e-05,
+      "loss": 0.3803,
+      "step": 63560
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4133889644179832e-05,
+      "loss": 0.1736,
+      "step": 63565
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4133420842904696e-05,
+      "loss": 0.0774,
+      "step": 63570
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4132952041629556e-05,
+      "loss": 0.1033,
+      "step": 63575
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4132483240354416e-05,
+      "loss": 0.0655,
+      "step": 63580
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4132014439079275e-05,
+      "loss": 0.1123,
+      "step": 63585
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4131545637804135e-05,
+      "loss": 0.075,
+      "step": 63590
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4131076836528997e-05,
+      "loss": 0.1171,
+      "step": 63595
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4130608035253857e-05,
+      "loss": 0.1863,
+      "step": 63600
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4130139233978717e-05,
+      "loss": 0.0909,
+      "step": 63605
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4129670432703577e-05,
+      "loss": 0.3155,
+      "step": 63610
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.412920163142844e-05,
+      "loss": 0.1992,
+      "step": 63615
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.41287328301533e-05,
+      "loss": 0.0821,
+      "step": 63620
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.412826402887816e-05,
+      "loss": 0.1078,
+      "step": 63625
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.412779522760302e-05,
+      "loss": 0.0328,
+      "step": 63630
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4127326426327882e-05,
+      "loss": 0.0879,
+      "step": 63635
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4126857625052741e-05,
+      "loss": 0.03,
+      "step": 63640
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4126388823777601e-05,
+      "loss": 0.1541,
+      "step": 63645
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4125920022502461e-05,
+      "loss": 0.0869,
+      "step": 63650
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4125451221227321e-05,
+      "loss": 0.1071,
+      "step": 63655
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4124982419952183e-05,
+      "loss": 0.2699,
+      "step": 63660
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4124513618677045e-05,
+      "loss": 0.1667,
+      "step": 63665
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4124044817401904e-05,
+      "loss": 0.0572,
+      "step": 63670
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4123576016126766e-05,
+      "loss": 0.0256,
+      "step": 63675
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4123107214851626e-05,
+      "loss": 0.0349,
+      "step": 63680
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4122638413576486e-05,
+      "loss": 0.0604,
+      "step": 63685
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4122169612301346e-05,
+      "loss": 0.0297,
+      "step": 63690
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4121700811026206e-05,
+      "loss": 0.1569,
+      "step": 63695
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4121232009751067e-05,
+      "loss": 0.0974,
+      "step": 63700
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4120763208475927e-05,
+      "loss": 0.1207,
+      "step": 63705
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4120294407200789e-05,
+      "loss": 0.2527,
+      "step": 63710
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.411982560592565e-05,
+      "loss": 0.2932,
+      "step": 63715
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.411935680465051e-05,
+      "loss": 0.0372,
+      "step": 63720
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.411888800337537e-05,
+      "loss": 0.0309,
+      "step": 63725
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.411841920210023e-05,
+      "loss": 0.019,
+      "step": 63730
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.411795040082509e-05,
+      "loss": 0.0686,
+      "step": 63735
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4117481599549952e-05,
+      "loss": 0.1975,
+      "step": 63740
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4117012798274812e-05,
+      "loss": 0.1837,
+      "step": 63745
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4116543996999672e-05,
+      "loss": 0.1234,
+      "step": 63750
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4116075195724535e-05,
+      "loss": 0.1988,
+      "step": 63755
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4115606394449395e-05,
+      "loss": 0.2116,
+      "step": 63760
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4115137593174255e-05,
+      "loss": 0.2059,
+      "step": 63765
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4114668791899115e-05,
+      "loss": 0.0628,
+      "step": 63770
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4114199990623975e-05,
+      "loss": 0.0345,
+      "step": 63775
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4113731189348837e-05,
+      "loss": 0.0374,
+      "step": 63780
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4113262388073697e-05,
+      "loss": 0.1135,
+      "step": 63785
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4112793586798556e-05,
+      "loss": 0.101,
+      "step": 63790
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4112324785523416e-05,
+      "loss": 0.0271,
+      "step": 63795
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.411185598424828e-05,
+      "loss": 0.0826,
+      "step": 63800
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.411138718297314e-05,
+      "loss": 0.1655,
+      "step": 63805
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4110918381698e-05,
+      "loss": 0.2046,
+      "step": 63810
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.411044958042286e-05,
+      "loss": 0.2569,
+      "step": 63815
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4109980779147721e-05,
+      "loss": 0.0889,
+      "step": 63820
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4109511977872581e-05,
+      "loss": 0.024,
+      "step": 63825
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4109043176597441e-05,
+      "loss": 0.0664,
+      "step": 63830
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4108574375322301e-05,
+      "loss": 0.0197,
+      "step": 63835
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4108105574047163e-05,
+      "loss": 0.0726,
+      "step": 63840
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4107636772772022e-05,
+      "loss": 0.087,
+      "step": 63845
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4107167971496884e-05,
+      "loss": 0.0947,
+      "step": 63850
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4106699170221744e-05,
+      "loss": 0.1588,
+      "step": 63855
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4106230368946606e-05,
+      "loss": 0.2632,
+      "step": 63860
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4105761567671466e-05,
+      "loss": 0.2116,
+      "step": 63865
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4105292766396326e-05,
+      "loss": 0.0705,
+      "step": 63870
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4104823965121185e-05,
+      "loss": 0.0129,
+      "step": 63875
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4104355163846047e-05,
+      "loss": 0.0476,
+      "step": 63880
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4103886362570907e-05,
+      "loss": 0.0482,
+      "step": 63885
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4103417561295767e-05,
+      "loss": 0.0601,
+      "step": 63890
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4102948760020629e-05,
+      "loss": 0.1474,
+      "step": 63895
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.410247995874549e-05,
+      "loss": 0.16,
+      "step": 63900
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.410201115747035e-05,
+      "loss": 0.2903,
+      "step": 63905
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.410154235619521e-05,
+      "loss": 0.2707,
+      "step": 63910
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.410107355492007e-05,
+      "loss": 0.315,
+      "step": 63915
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4100604753644932e-05,
+      "loss": 0.0473,
+      "step": 63920
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4100135952369792e-05,
+      "loss": 0.045,
+      "step": 63925
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4099667151094652e-05,
+      "loss": 0.0782,
+      "step": 63930
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4099198349819511e-05,
+      "loss": 0.0812,
+      "step": 63935
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4098729548544375e-05,
+      "loss": 0.0824,
+      "step": 63940
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4098260747269235e-05,
+      "loss": 0.1176,
+      "step": 63945
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4097791945994095e-05,
+      "loss": 0.151,
+      "step": 63950
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4097323144718955e-05,
+      "loss": 0.1791,
+      "step": 63955
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4096854343443816e-05,
+      "loss": 0.3215,
+      "step": 63960
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4096385542168676e-05,
+      "loss": 0.2454,
+      "step": 63965
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4095916740893536e-05,
+      "loss": 0.047,
+      "step": 63970
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4095447939618396e-05,
+      "loss": 0.0286,
+      "step": 63975
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4094979138343256e-05,
+      "loss": 0.0482,
+      "step": 63980
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4094510337068118e-05,
+      "loss": 0.0548,
+      "step": 63985
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.409404153579298e-05,
+      "loss": 0.0564,
+      "step": 63990
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4093572734517839e-05,
+      "loss": 0.0845,
+      "step": 63995
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.40931039332427e-05,
+      "loss": 0.0653,
+      "step": 64000
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.409263513196756e-05,
+      "loss": 0.1316,
+      "step": 64005
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.409216633069242e-05,
+      "loss": 0.2763,
+      "step": 64010
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.409169752941728e-05,
+      "loss": 0.2955,
+      "step": 64015
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.409122872814214e-05,
+      "loss": 0.0321,
+      "step": 64020
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4090759926867002e-05,
+      "loss": 0.0718,
+      "step": 64025
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4090291125591862e-05,
+      "loss": 0.0463,
+      "step": 64030
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4089822324316724e-05,
+      "loss": 0.0631,
+      "step": 64035
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4089353523041585e-05,
+      "loss": 0.0756,
+      "step": 64040
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4088884721766445e-05,
+      "loss": 0.0511,
+      "step": 64045
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4088415920491305e-05,
+      "loss": 0.1322,
+      "step": 64050
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4087947119216165e-05,
+      "loss": 0.1484,
+      "step": 64055
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4087478317941025e-05,
+      "loss": 0.2167,
+      "step": 64060
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4087009516665887e-05,
+      "loss": 0.4035,
+      "step": 64065
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4086540715390747e-05,
+      "loss": 0.0353,
+      "step": 64070
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4086071914115607e-05,
+      "loss": 0.0812,
+      "step": 64075
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.408560311284047e-05,
+      "loss": 0.0497,
+      "step": 64080
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.408513431156533e-05,
+      "loss": 0.0927,
+      "step": 64085
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.408466551029019e-05,
+      "loss": 0.1061,
+      "step": 64090
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.408419670901505e-05,
+      "loss": 0.0803,
+      "step": 64095
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.408372790773991e-05,
+      "loss": 0.0983,
+      "step": 64100
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4083259106464771e-05,
+      "loss": 0.1288,
+      "step": 64105
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4082790305189631e-05,
+      "loss": 0.1537,
+      "step": 64110
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4082321503914491e-05,
+      "loss": 0.2068,
+      "step": 64115
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4081852702639351e-05,
+      "loss": 0.0357,
+      "step": 64120
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4081383901364214e-05,
+      "loss": 0.0367,
+      "step": 64125
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4080915100089074e-05,
+      "loss": 0.0455,
+      "step": 64130
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4080446298813934e-05,
+      "loss": 0.1085,
+      "step": 64135
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4079977497538794e-05,
+      "loss": 0.1128,
+      "step": 64140
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4079508696263656e-05,
+      "loss": 0.0766,
+      "step": 64145
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4079039894988516e-05,
+      "loss": 0.0769,
+      "step": 64150
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4078571093713376e-05,
+      "loss": 0.1373,
+      "step": 64155
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4078102292438236e-05,
+      "loss": 0.2616,
+      "step": 64160
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4077633491163096e-05,
+      "loss": 0.4626,
+      "step": 64165
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4077164689887957e-05,
+      "loss": 0.0197,
+      "step": 64170
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4076695888612819e-05,
+      "loss": 0.0197,
+      "step": 64175
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4076227087337679e-05,
+      "loss": 0.0383,
+      "step": 64180
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.407575828606254e-05,
+      "loss": 0.1042,
+      "step": 64185
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.40752894847874e-05,
+      "loss": 0.0343,
+      "step": 64190
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.407482068351226e-05,
+      "loss": 0.0483,
+      "step": 64195
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.407435188223712e-05,
+      "loss": 0.0665,
+      "step": 64200
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.407388308096198e-05,
+      "loss": 0.2019,
+      "step": 64205
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4073414279686842e-05,
+      "loss": 0.231,
+      "step": 64210
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4072945478411702e-05,
+      "loss": 0.338,
+      "step": 64215
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4072476677136563e-05,
+      "loss": 0.0589,
+      "step": 64220
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4072007875861425e-05,
+      "loss": 0.0093,
+      "step": 64225
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4071539074586285e-05,
+      "loss": 0.0857,
+      "step": 64230
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4071070273311145e-05,
+      "loss": 0.0601,
+      "step": 64235
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4070601472036005e-05,
+      "loss": 0.1184,
+      "step": 64240
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4070132670760865e-05,
+      "loss": 0.0687,
+      "step": 64245
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4069663869485726e-05,
+      "loss": 0.0938,
+      "step": 64250
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4069195068210586e-05,
+      "loss": 0.0997,
+      "step": 64255
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4068726266935446e-05,
+      "loss": 0.2678,
+      "step": 64260
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.406825746566031e-05,
+      "loss": 0.3459,
+      "step": 64265
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.406778866438517e-05,
+      "loss": 0.0474,
+      "step": 64270
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.406731986311003e-05,
+      "loss": 0.0988,
+      "step": 64275
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.406685106183489e-05,
+      "loss": 0.0423,
+      "step": 64280
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.406638226055975e-05,
+      "loss": 0.0893,
+      "step": 64285
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.406591345928461e-05,
+      "loss": 0.1688,
+      "step": 64290
+    },
+    {
+      "epoch": 3.0,
+      "eval_cer": 0.011432097636787493,
+      "eval_loss": 0.03627169877290726,
+      "eval_runtime": 381.0741,
+      "eval_samples_per_second": 49.99,
+      "eval_steps_per_second": 12.499,
+      "eval_wer": 0.09668894174500699,
+      "step": 64293
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.406544465800947e-05,
+      "loss": 0.2315,
+      "step": 64295
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.406497585673433e-05,
+      "loss": 0.1021,
+      "step": 64300
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.406450705545919e-05,
+      "loss": 0.0167,
+      "step": 64305
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4064038254184052e-05,
+      "loss": 0.0554,
+      "step": 64310
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4063569452908914e-05,
+      "loss": 0.0683,
+      "step": 64315
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4063100651633774e-05,
+      "loss": 0.0589,
+      "step": 64320
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4062631850358634e-05,
+      "loss": 0.1134,
+      "step": 64325
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4062163049083495e-05,
+      "loss": 0.1114,
+      "step": 64330
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4061694247808355e-05,
+      "loss": 0.1605,
+      "step": 64335
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4061225446533215e-05,
+      "loss": 0.2171,
+      "step": 64340
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4060756645258075e-05,
+      "loss": 0.2604,
+      "step": 64345
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4060287843982937e-05,
+      "loss": 0.0621,
+      "step": 64350
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4059819042707797e-05,
+      "loss": 0.0456,
+      "step": 64355
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4059350241432658e-05,
+      "loss": 0.0587,
+      "step": 64360
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4058881440157518e-05,
+      "loss": 0.0389,
+      "step": 64365
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.405841263888238e-05,
+      "loss": 0.0616,
+      "step": 64370
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.405794383760724e-05,
+      "loss": 0.0693,
+      "step": 64375
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.40574750363321e-05,
+      "loss": 0.0516,
+      "step": 64380
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.405700623505696e-05,
+      "loss": 0.1313,
+      "step": 64385
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4056537433781821e-05,
+      "loss": 0.3616,
+      "step": 64390
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4056068632506681e-05,
+      "loss": 0.2899,
+      "step": 64395
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4055599831231541e-05,
+      "loss": 0.0599,
+      "step": 64400
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4055131029956403e-05,
+      "loss": 0.0295,
+      "step": 64405
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4054662228681264e-05,
+      "loss": 0.0761,
+      "step": 64410
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4054193427406124e-05,
+      "loss": 0.0633,
+      "step": 64415
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4053724626130984e-05,
+      "loss": 0.0678,
+      "step": 64420
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4053255824855844e-05,
+      "loss": 0.081,
+      "step": 64425
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4052787023580706e-05,
+      "loss": 0.1015,
+      "step": 64430
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4052318222305566e-05,
+      "loss": 0.1764,
+      "step": 64435
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4051849421030426e-05,
+      "loss": 0.1692,
+      "step": 64440
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4051380619755286e-05,
+      "loss": 0.3618,
+      "step": 64445
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4050911818480149e-05,
+      "loss": 0.0466,
+      "step": 64450
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4050443017205009e-05,
+      "loss": 0.0402,
+      "step": 64455
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4049974215929869e-05,
+      "loss": 0.0041,
+      "step": 64460
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4049505414654729e-05,
+      "loss": 0.0794,
+      "step": 64465
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.404903661337959e-05,
+      "loss": 0.1197,
+      "step": 64470
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.404856781210445e-05,
+      "loss": 0.0774,
+      "step": 64475
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.404809901082931e-05,
+      "loss": 0.1861,
+      "step": 64480
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.404763020955417e-05,
+      "loss": 0.1428,
+      "step": 64485
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.404716140827903e-05,
+      "loss": 0.2123,
+      "step": 64490
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4046692607003892e-05,
+      "loss": 0.254,
+      "step": 64495
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4046223805728753e-05,
+      "loss": 0.0496,
+      "step": 64500
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4045755004453613e-05,
+      "loss": 0.0437,
+      "step": 64505
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4045286203178475e-05,
+      "loss": 0.0811,
+      "step": 64510
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4044817401903335e-05,
+      "loss": 0.0684,
+      "step": 64515
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4044348600628195e-05,
+      "loss": 0.0883,
+      "step": 64520
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4043879799353055e-05,
+      "loss": 0.0396,
+      "step": 64525
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4043410998077915e-05,
+      "loss": 0.0864,
+      "step": 64530
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4042942196802776e-05,
+      "loss": 0.1839,
+      "step": 64535
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4042473395527636e-05,
+      "loss": 0.2589,
+      "step": 64540
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4042004594252498e-05,
+      "loss": 0.3049,
+      "step": 64545
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.404153579297736e-05,
+      "loss": 0.0369,
+      "step": 64550
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.404106699170222e-05,
+      "loss": 0.0151,
+      "step": 64555
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.404059819042708e-05,
+      "loss": 0.0752,
+      "step": 64560
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.404012938915194e-05,
+      "loss": 0.0677,
+      "step": 64565
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.40396605878768e-05,
+      "loss": 0.0562,
+      "step": 64570
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4039191786601661e-05,
+      "loss": 0.0675,
+      "step": 64575
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.403872298532652e-05,
+      "loss": 0.2245,
+      "step": 64580
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.403825418405138e-05,
+      "loss": 0.0523,
+      "step": 64585
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4037785382776244e-05,
+      "loss": 0.1616,
+      "step": 64590
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4037316581501104e-05,
+      "loss": 0.3289,
+      "step": 64595
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4036847780225964e-05,
+      "loss": 0.067,
+      "step": 64600
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4036378978950824e-05,
+      "loss": 0.0411,
+      "step": 64605
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4035910177675684e-05,
+      "loss": 0.0761,
+      "step": 64610
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4035441376400545e-05,
+      "loss": 0.0312,
+      "step": 64615
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4034972575125405e-05,
+      "loss": 0.0734,
+      "step": 64620
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4034503773850265e-05,
+      "loss": 0.1185,
+      "step": 64625
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4034034972575125e-05,
+      "loss": 0.1229,
+      "step": 64630
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4033566171299985e-05,
+      "loss": 0.0916,
+      "step": 64635
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4033097370024848e-05,
+      "loss": 0.174,
+      "step": 64640
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4032628568749708e-05,
+      "loss": 0.3041,
+      "step": 64645
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4032159767474568e-05,
+      "loss": 0.0549,
+      "step": 64650
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.403169096619943e-05,
+      "loss": 0.053,
+      "step": 64655
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.403122216492429e-05,
+      "loss": 0.0258,
+      "step": 64660
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.403075336364915e-05,
+      "loss": 0.0875,
+      "step": 64665
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.403028456237401e-05,
+      "loss": 0.0554,
+      "step": 64670
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.402981576109887e-05,
+      "loss": 0.0884,
+      "step": 64675
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4029346959823731e-05,
+      "loss": 0.1799,
+      "step": 64680
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4028878158548593e-05,
+      "loss": 0.1663,
+      "step": 64685
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4028409357273453e-05,
+      "loss": 0.208,
+      "step": 64690
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4027940555998315e-05,
+      "loss": 0.462,
+      "step": 64695
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4027471754723174e-05,
+      "loss": 0.0676,
+      "step": 64700
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4027002953448034e-05,
+      "loss": 0.0337,
+      "step": 64705
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4026534152172894e-05,
+      "loss": 0.0228,
+      "step": 64710
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4026065350897754e-05,
+      "loss": 0.0328,
+      "step": 64715
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4025596549622616e-05,
+      "loss": 0.0326,
+      "step": 64720
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4025127748347476e-05,
+      "loss": 0.0464,
+      "step": 64725
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4024658947072337e-05,
+      "loss": 0.1157,
+      "step": 64730
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4024190145797199e-05,
+      "loss": 0.1096,
+      "step": 64735
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4023721344522059e-05,
+      "loss": 0.2395,
+      "step": 64740
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4023252543246919e-05,
+      "loss": 0.2545,
+      "step": 64745
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4022783741971779e-05,
+      "loss": 0.0549,
+      "step": 64750
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.402231494069664e-05,
+      "loss": 0.0309,
+      "step": 64755
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.40218461394215e-05,
+      "loss": 0.0336,
+      "step": 64760
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.402137733814636e-05,
+      "loss": 0.0846,
+      "step": 64765
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.402090853687122e-05,
+      "loss": 0.0801,
+      "step": 64770
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4020439735596084e-05,
+      "loss": 0.0981,
+      "step": 64775
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4019970934320944e-05,
+      "loss": 0.0964,
+      "step": 64780
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4019502133045804e-05,
+      "loss": 0.1676,
+      "step": 64785
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4019033331770663e-05,
+      "loss": 0.2262,
+      "step": 64790
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4018564530495525e-05,
+      "loss": 0.3382,
+      "step": 64795
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4018095729220385e-05,
+      "loss": 0.1401,
+      "step": 64800
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4017626927945245e-05,
+      "loss": 0.0242,
+      "step": 64805
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4017158126670105e-05,
+      "loss": 0.0722,
+      "step": 64810
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4016689325394965e-05,
+      "loss": 0.0581,
+      "step": 64815
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4016220524119826e-05,
+      "loss": 0.0565,
+      "step": 64820
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4015751722844688e-05,
+      "loss": 0.0378,
+      "step": 64825
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4015282921569548e-05,
+      "loss": 0.1204,
+      "step": 64830
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.401481412029441e-05,
+      "loss": 0.0845,
+      "step": 64835
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.401434531901927e-05,
+      "loss": 0.1998,
+      "step": 64840
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.401387651774413e-05,
+      "loss": 0.2978,
+      "step": 64845
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.401340771646899e-05,
+      "loss": 0.0672,
+      "step": 64850
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.401293891519385e-05,
+      "loss": 0.0436,
+      "step": 64855
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4012470113918711e-05,
+      "loss": 0.0702,
+      "step": 64860
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4012001312643571e-05,
+      "loss": 0.0439,
+      "step": 64865
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4011532511368433e-05,
+      "loss": 0.0986,
+      "step": 64870
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4011063710093294e-05,
+      "loss": 0.05,
+      "step": 64875
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4010594908818154e-05,
+      "loss": 0.1149,
+      "step": 64880
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4010126107543014e-05,
+      "loss": 0.1209,
+      "step": 64885
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4009657306267874e-05,
+      "loss": 0.1562,
+      "step": 64890
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4009188504992734e-05,
+      "loss": 0.2224,
+      "step": 64895
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4008719703717596e-05,
+      "loss": 0.1391,
+      "step": 64900
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4008250902442455e-05,
+      "loss": 0.0347,
+      "step": 64905
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4007782101167315e-05,
+      "loss": 0.0537,
+      "step": 64910
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4007313299892179e-05,
+      "loss": 0.0274,
+      "step": 64915
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4006844498617039e-05,
+      "loss": 0.049,
+      "step": 64920
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4006375697341899e-05,
+      "loss": 0.0963,
+      "step": 64925
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4005906896066759e-05,
+      "loss": 0.0633,
+      "step": 64930
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4005438094791618e-05,
+      "loss": 0.0924,
+      "step": 64935
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.400496929351648e-05,
+      "loss": 0.2455,
+      "step": 64940
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.400450049224134e-05,
+      "loss": 0.229,
+      "step": 64945
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.40040316909662e-05,
+      "loss": 0.0981,
+      "step": 64950
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.400356288969106e-05,
+      "loss": 0.0606,
+      "step": 64955
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.400309408841592e-05,
+      "loss": 0.0343,
+      "step": 64960
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4002625287140783e-05,
+      "loss": 0.0352,
+      "step": 64965
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4002156485865643e-05,
+      "loss": 0.0588,
+      "step": 64970
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4001687684590503e-05,
+      "loss": 0.0514,
+      "step": 64975
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4001218883315365e-05,
+      "loss": 0.1118,
+      "step": 64980
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4000750082040225e-05,
+      "loss": 0.1195,
+      "step": 64985
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4000281280765085e-05,
+      "loss": 0.1169,
+      "step": 64990
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3999812479489944e-05,
+      "loss": 0.2991,
+      "step": 64995
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3999343678214804e-05,
+      "loss": 0.0672,
+      "step": 65000
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3998874876939666e-05,
+      "loss": 0.0248,
+      "step": 65005
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3998406075664528e-05,
+      "loss": 0.0432,
+      "step": 65010
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3997937274389388e-05,
+      "loss": 0.0714,
+      "step": 65015
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.399746847311425e-05,
+      "loss": 0.1364,
+      "step": 65020
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3996999671839109e-05,
+      "loss": 0.0531,
+      "step": 65025
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3996530870563969e-05,
+      "loss": 0.1025,
+      "step": 65030
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3996062069288829e-05,
+      "loss": 0.1427,
+      "step": 65035
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3995593268013689e-05,
+      "loss": 0.1708,
+      "step": 65040
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.399512446673855e-05,
+      "loss": 0.3074,
+      "step": 65045
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.399465566546341e-05,
+      "loss": 0.1141,
+      "step": 65050
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3994186864188272e-05,
+      "loss": 0.0437,
+      "step": 65055
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3993718062913134e-05,
+      "loss": 0.047,
+      "step": 65060
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3993249261637994e-05,
+      "loss": 0.0521,
+      "step": 65065
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3992780460362854e-05,
+      "loss": 0.04,
+      "step": 65070
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3992311659087714e-05,
+      "loss": 0.132,
+      "step": 65075
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3991842857812573e-05,
+      "loss": 0.0976,
+      "step": 65080
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3991374056537435e-05,
+      "loss": 0.1697,
+      "step": 65085
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3990905255262295e-05,
+      "loss": 0.164,
+      "step": 65090
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3990436453987155e-05,
+      "loss": 0.175,
+      "step": 65095
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3989967652712018e-05,
+      "loss": 0.0484,
+      "step": 65100
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3989498851436878e-05,
+      "loss": 0.0495,
+      "step": 65105
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3989030050161738e-05,
+      "loss": 0.0101,
+      "step": 65110
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3988561248886598e-05,
+      "loss": 0.0512,
+      "step": 65115
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3988092447611458e-05,
+      "loss": 0.0571,
+      "step": 65120
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.398762364633632e-05,
+      "loss": 0.1731,
+      "step": 65125
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.398715484506118e-05,
+      "loss": 0.1818,
+      "step": 65130
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.398668604378604e-05,
+      "loss": 0.1292,
+      "step": 65135
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.39862172425109e-05,
+      "loss": 0.314,
+      "step": 65140
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3985748441235761e-05,
+      "loss": 0.2696,
+      "step": 65145
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3985279639960623e-05,
+      "loss": 0.0891,
+      "step": 65150
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3984810838685483e-05,
+      "loss": 0.0257,
+      "step": 65155
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3984342037410343e-05,
+      "loss": 0.0532,
+      "step": 65160
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3983873236135204e-05,
+      "loss": 0.0476,
+      "step": 65165
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3983404434860064e-05,
+      "loss": 0.0654,
+      "step": 65170
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3982935633584924e-05,
+      "loss": 0.0515,
+      "step": 65175
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3982466832309784e-05,
+      "loss": 0.1277,
+      "step": 65180
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3981998031034646e-05,
+      "loss": 0.0926,
+      "step": 65185
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3981529229759506e-05,
+      "loss": 0.1293,
+      "step": 65190
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3981060428484367e-05,
+      "loss": 0.343,
+      "step": 65195
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3980591627209227e-05,
+      "loss": 0.0822,
+      "step": 65200
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3980122825934089e-05,
+      "loss": 0.0387,
+      "step": 65205
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3979654024658949e-05,
+      "loss": 0.02,
+      "step": 65210
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3979185223383809e-05,
+      "loss": 0.0687,
+      "step": 65215
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3978716422108669e-05,
+      "loss": 0.1311,
+      "step": 65220
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.397824762083353e-05,
+      "loss": 0.0949,
+      "step": 65225
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.397777881955839e-05,
+      "loss": 0.0685,
+      "step": 65230
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.397731001828325e-05,
+      "loss": 0.0929,
+      "step": 65235
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3976841217008112e-05,
+      "loss": 0.2325,
+      "step": 65240
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3976372415732973e-05,
+      "loss": 0.2006,
+      "step": 65245
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3975903614457833e-05,
+      "loss": 0.0695,
+      "step": 65250
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3975434813182693e-05,
+      "loss": 0.0692,
+      "step": 65255
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3974966011907553e-05,
+      "loss": 0.0339,
+      "step": 65260
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3974497210632415e-05,
+      "loss": 0.023,
+      "step": 65265
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3974028409357275e-05,
+      "loss": 0.1405,
+      "step": 65270
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3973559608082135e-05,
+      "loss": 0.1026,
+      "step": 65275
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3973090806806995e-05,
+      "loss": 0.1097,
+      "step": 65280
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3972622005531854e-05,
+      "loss": 0.2419,
+      "step": 65285
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3972153204256718e-05,
+      "loss": 0.1716,
+      "step": 65290
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3971684402981578e-05,
+      "loss": 0.2981,
+      "step": 65295
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3971215601706438e-05,
+      "loss": 0.0676,
+      "step": 65300
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.39707468004313e-05,
+      "loss": 0.0832,
+      "step": 65305
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.397027799915616e-05,
+      "loss": 0.0183,
+      "step": 65310
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.396980919788102e-05,
+      "loss": 0.0464,
+      "step": 65315
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3969340396605879e-05,
+      "loss": 0.0661,
+      "step": 65320
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3968871595330739e-05,
+      "loss": 0.0894,
+      "step": 65325
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.39684027940556e-05,
+      "loss": 0.1564,
+      "step": 65330
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3967933992780462e-05,
+      "loss": 0.0455,
+      "step": 65335
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3967465191505322e-05,
+      "loss": 0.1823,
+      "step": 65340
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3966996390230184e-05,
+      "loss": 0.2856,
+      "step": 65345
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3966527588955044e-05,
+      "loss": 0.0937,
+      "step": 65350
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3966058787679904e-05,
+      "loss": 0.0362,
+      "step": 65355
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3965589986404764e-05,
+      "loss": 0.0686,
+      "step": 65360
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3965121185129624e-05,
+      "loss": 0.0787,
+      "step": 65365
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3964652383854485e-05,
+      "loss": 0.109,
+      "step": 65370
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3964183582579345e-05,
+      "loss": 0.0994,
+      "step": 65375
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3963714781304207e-05,
+      "loss": 0.1041,
+      "step": 65380
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3963245980029068e-05,
+      "loss": 0.0703,
+      "step": 65385
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3962777178753928e-05,
+      "loss": 0.2074,
+      "step": 65390
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3962308377478788e-05,
+      "loss": 0.2059,
+      "step": 65395
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3961839576203648e-05,
+      "loss": 0.0528,
+      "step": 65400
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3961370774928508e-05,
+      "loss": 0.1225,
+      "step": 65405
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.396090197365337e-05,
+      "loss": 0.0581,
+      "step": 65410
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.396043317237823e-05,
+      "loss": 0.0427,
+      "step": 65415
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.395996437110309e-05,
+      "loss": 0.0252,
+      "step": 65420
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3959495569827953e-05,
+      "loss": 0.0547,
+      "step": 65425
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3959026768552813e-05,
+      "loss": 0.0974,
+      "step": 65430
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3958557967277673e-05,
+      "loss": 0.1627,
+      "step": 65435
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3958089166002533e-05,
+      "loss": 0.2218,
+      "step": 65440
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3957620364727393e-05,
+      "loss": 0.3212,
+      "step": 65445
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3957151563452254e-05,
+      "loss": 0.075,
+      "step": 65450
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3956682762177114e-05,
+      "loss": 0.0222,
+      "step": 65455
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3956213960901974e-05,
+      "loss": 0.0682,
+      "step": 65460
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3955745159626834e-05,
+      "loss": 0.0816,
+      "step": 65465
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3955276358351694e-05,
+      "loss": 0.1097,
+      "step": 65470
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3954807557076557e-05,
+      "loss": 0.0523,
+      "step": 65475
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3954338755801417e-05,
+      "loss": 0.08,
+      "step": 65480
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3953869954526277e-05,
+      "loss": 0.163,
+      "step": 65485
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3953401153251139e-05,
+      "loss": 0.0968,
+      "step": 65490
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3952932351975999e-05,
+      "loss": 0.2553,
+      "step": 65495
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3952463550700859e-05,
+      "loss": 0.0787,
+      "step": 65500
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3951994749425719e-05,
+      "loss": 0.0242,
+      "step": 65505
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3951525948150579e-05,
+      "loss": 0.0399,
+      "step": 65510
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.395105714687544e-05,
+      "loss": 0.0426,
+      "step": 65515
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3950588345600302e-05,
+      "loss": 0.1501,
+      "step": 65520
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3950119544325162e-05,
+      "loss": 0.0746,
+      "step": 65525
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3949650743050023e-05,
+      "loss": 0.1603,
+      "step": 65530
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3949181941774883e-05,
+      "loss": 0.1407,
+      "step": 65535
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3948713140499743e-05,
+      "loss": 0.2105,
+      "step": 65540
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3948244339224603e-05,
+      "loss": 0.2679,
+      "step": 65545
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3947775537949463e-05,
+      "loss": 0.0636,
+      "step": 65550
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3947306736674325e-05,
+      "loss": 0.0332,
+      "step": 65555
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3946837935399185e-05,
+      "loss": 0.0484,
+      "step": 65560
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3946369134124046e-05,
+      "loss": 0.0443,
+      "step": 65565
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3945900332848908e-05,
+      "loss": 0.0765,
+      "step": 65570
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3945431531573768e-05,
+      "loss": 0.087,
+      "step": 65575
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3944962730298628e-05,
+      "loss": 0.1172,
+      "step": 65580
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3944493929023488e-05,
+      "loss": 0.1665,
+      "step": 65585
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3944025127748348e-05,
+      "loss": 0.258,
+      "step": 65590
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.394355632647321e-05,
+      "loss": 0.3192,
+      "step": 65595
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.394308752519807e-05,
+      "loss": 0.0779,
+      "step": 65600
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.394261872392293e-05,
+      "loss": 0.0351,
+      "step": 65605
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3942149922647789e-05,
+      "loss": 0.0668,
+      "step": 65610
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3941681121372652e-05,
+      "loss": 0.0433,
+      "step": 65615
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3941212320097512e-05,
+      "loss": 0.0631,
+      "step": 65620
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3940743518822372e-05,
+      "loss": 0.1735,
+      "step": 65625
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3940274717547232e-05,
+      "loss": 0.154,
+      "step": 65630
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3939805916272094e-05,
+      "loss": 0.1654,
+      "step": 65635
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3939337114996954e-05,
+      "loss": 0.1955,
+      "step": 65640
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3938868313721814e-05,
+      "loss": 0.2966,
+      "step": 65645
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3938399512446674e-05,
+      "loss": 0.0721,
+      "step": 65650
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3937930711171535e-05,
+      "loss": 0.106,
+      "step": 65655
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3937461909896397e-05,
+      "loss": 0.0779,
+      "step": 65660
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3936993108621257e-05,
+      "loss": 0.0368,
+      "step": 65665
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3936524307346117e-05,
+      "loss": 0.0609,
+      "step": 65670
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3936055506070978e-05,
+      "loss": 0.073,
+      "step": 65675
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3935586704795838e-05,
+      "loss": 0.1176,
+      "step": 65680
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3935117903520698e-05,
+      "loss": 0.2023,
+      "step": 65685
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.3934649102245558e-05,
+      "loss": 0.1856,
+      "step": 65690
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.393418030097042e-05,
+      "loss": 0.2884,
+      "step": 65695
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.393371149969528e-05,
+      "loss": 0.0588,
+      "step": 65700
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.3933242698420141e-05,
+      "loss": 0.0409,
+      "step": 65705
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.3932773897145001e-05,
+      "loss": 0.0542,
+      "step": 65710
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.3932305095869863e-05,
+      "loss": 0.0748,
+      "step": 65715
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.3931836294594723e-05,
+      "loss": 0.0393,
+      "step": 65720
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.3931367493319583e-05,
+      "loss": 0.0428,
+      "step": 65725
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.3930898692044443e-05,
+      "loss": 0.1729,
+      "step": 65730
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.3930429890769304e-05,
+      "loss": 0.1347,
+      "step": 65735
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.3929961089494164e-05,
+      "loss": 0.1727,
+      "step": 65740
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.3929492288219024e-05,
+      "loss": 0.3092,
+      "step": 65745
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.3929023486943886e-05,
+      "loss": 0.0617,
+      "step": 65750
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.3928554685668748e-05,
+      "loss": 0.0284,
+      "step": 65755
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.3928085884393607e-05,
+      "loss": 0.0499,
+      "step": 65760
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.3927617083118467e-05,
+      "loss": 0.0908,
+      "step": 65765
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.3927148281843327e-05,
+      "loss": 0.0661,
+      "step": 65770
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.3926679480568189e-05,
+      "loss": 0.0754,
+      "step": 65775
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.3926210679293049e-05,
+      "loss": 0.2077,
+      "step": 65780
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.3925741878017909e-05,
+      "loss": 0.2175,
+      "step": 65785
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.3925273076742769e-05,
+      "loss": 0.1797,
+      "step": 65790
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.3924804275467629e-05,
+      "loss": 0.2083,
+      "step": 65795
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.3924335474192492e-05,
+      "loss": 0.0658,
+      "step": 65800
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.3923866672917352e-05,
+      "loss": 0.0514,
+      "step": 65805
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.3923397871642212e-05,
+      "loss": 0.06,
+      "step": 65810
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.3922929070367074e-05,
+      "loss": 0.0353,
+      "step": 65815
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.3922460269091933e-05,
+      "loss": 0.0438,
+      "step": 65820
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.3921991467816793e-05,
+      "loss": 0.1012,
+      "step": 65825
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.3921522666541653e-05,
+      "loss": 0.0892,
+      "step": 65830
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.3921053865266513e-05,
+      "loss": 0.151,
+      "step": 65835
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.3920585063991375e-05,
+      "loss": 0.2743,
+      "step": 65840
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.3920116262716236e-05,
+      "loss": 0.3603,
+      "step": 65845
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.3919647461441096e-05,
+      "loss": 0.0907,
+      "step": 65850
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.3919178660165958e-05,
+      "loss": 0.0014,
+      "step": 65855
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.3918709858890818e-05,
+      "loss": 0.0464,
+      "step": 65860
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.3918241057615678e-05,
+      "loss": 0.0833,
+      "step": 65865
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.3917772256340538e-05,
+      "loss": 0.0198,
+      "step": 65870
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.3917303455065398e-05,
+      "loss": 0.0867,
+      "step": 65875
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.391683465379026e-05,
+      "loss": 0.1497,
+      "step": 65880
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.391636585251512e-05,
+      "loss": 0.1157,
+      "step": 65885
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.3915897051239981e-05,
+      "loss": 0.1333,
+      "step": 65890
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.3915428249964843e-05,
+      "loss": 0.3905,
+      "step": 65895
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.3914959448689703e-05,
+      "loss": 0.0356,
+      "step": 65900
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.3914490647414562e-05,
+      "loss": 0.0283,
+      "step": 65905
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.3914021846139422e-05,
+      "loss": 0.0624,
+      "step": 65910
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.3913553044864282e-05,
+      "loss": 0.0387,
+      "step": 65915
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.3913084243589144e-05,
+      "loss": 0.1258,
+      "step": 65920
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.3912615442314004e-05,
+      "loss": 0.0784,
+      "step": 65925
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.3912146641038864e-05,
+      "loss": 0.0804,
+      "step": 65930
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.3911677839763724e-05,
+      "loss": 0.1573,
+      "step": 65935
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.3911209038488587e-05,
+      "loss": 0.2832,
+      "step": 65940
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.3910740237213447e-05,
+      "loss": 0.328,
+      "step": 65945
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.3910271435938307e-05,
+      "loss": 0.0674,
+      "step": 65950
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.3909802634663167e-05,
+      "loss": 0.0282,
+      "step": 65955
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.3909333833388029e-05,
+      "loss": 0.018,
+      "step": 65960
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.3908865032112888e-05,
+      "loss": 0.0294,
+      "step": 65965
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.3908396230837748e-05,
+      "loss": 0.1143,
+      "step": 65970
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.3907927429562608e-05,
+      "loss": 0.0887,
+      "step": 65975
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.3907458628287468e-05,
+      "loss": 0.1257,
+      "step": 65980
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.3906989827012332e-05,
+      "loss": 0.1163,
+      "step": 65985
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.3906521025737192e-05,
+      "loss": 0.1974,
+      "step": 65990
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.3906052224462051e-05,
+      "loss": 0.2213,
+      "step": 65995
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.3905583423186913e-05,
+      "loss": 0.1337,
+      "step": 66000
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.3905114621911773e-05,
+      "loss": 0.0456,
+      "step": 66005
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.3904645820636633e-05,
+      "loss": 0.0686,
+      "step": 66010
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.3904177019361493e-05,
+      "loss": 0.0671,
+      "step": 66015
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.3903708218086353e-05,
+      "loss": 0.0511,
+      "step": 66020
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.3903239416811214e-05,
+      "loss": 0.1142,
+      "step": 66025
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.3902770615536076e-05,
+      "loss": 0.1178,
+      "step": 66030
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.3902301814260936e-05,
+      "loss": 0.1264,
+      "step": 66035
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.3901833012985798e-05,
+      "loss": 0.1253,
+      "step": 66040
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.3901364211710658e-05,
+      "loss": 0.272,
+      "step": 66045
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.3900895410435517e-05,
+      "loss": 0.072,
+      "step": 66050
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.3900426609160377e-05,
+      "loss": 0.1135,
+      "step": 66055
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.3899957807885237e-05,
+      "loss": 0.1074,
+      "step": 66060
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.3899489006610099e-05,
+      "loss": 0.0989,
+      "step": 66065
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.3899020205334959e-05,
+      "loss": 0.068,
+      "step": 66070
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.389855140405982e-05,
+      "loss": 0.053,
+      "step": 66075
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.3898082602784682e-05,
+      "loss": 0.1008,
+      "step": 66080
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.3897613801509542e-05,
+      "loss": 0.0979,
+      "step": 66085
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.3897145000234402e-05,
+      "loss": 0.1121,
+      "step": 66090
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.3896676198959262e-05,
+      "loss": 0.2136,
+      "step": 66095
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.3896207397684122e-05,
+      "loss": 0.0855,
+      "step": 66100
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.3895738596408984e-05,
+      "loss": 0.0216,
+      "step": 66105
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.3895269795133843e-05,
+      "loss": 0.0825,
+      "step": 66110
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.3894800993858703e-05,
+      "loss": 0.019,
+      "step": 66115
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.3894332192583563e-05,
+      "loss": 0.0268,
+      "step": 66120
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.3893863391308427e-05,
+      "loss": 0.0669,
+      "step": 66125
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.3893394590033287e-05,
+      "loss": 0.0652,
+      "step": 66130
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.3892925788758147e-05,
+      "loss": 0.1235,
+      "step": 66135
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.3892456987483006e-05,
+      "loss": 0.2344,
+      "step": 66140
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.3891988186207868e-05,
+      "loss": 0.2408,
+      "step": 66145
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.3891519384932728e-05,
+      "loss": 0.0598,
+      "step": 66150
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.3891050583657588e-05,
+      "loss": 0.0141,
+      "step": 66155
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.3890581782382448e-05,
+      "loss": 0.0588,
+      "step": 66160
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.389011298110731e-05,
+      "loss": 0.0216,
+      "step": 66165
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.3889644179832171e-05,
+      "loss": 0.102,
+      "step": 66170
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.3889175378557031e-05,
+      "loss": 0.058,
+      "step": 66175
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.3888706577281893e-05,
+      "loss": 0.0824,
+      "step": 66180
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.3888237776006753e-05,
+      "loss": 0.177,
+      "step": 66185
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.3887768974731613e-05,
+      "loss": 0.2118,
+      "step": 66190
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.3887300173456473e-05,
+      "loss": 0.2732,
+      "step": 66195
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.3886831372181332e-05,
+      "loss": 0.0782,
+      "step": 66200
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.3886362570906194e-05,
+      "loss": 0.0291,
+      "step": 66205
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.3885893769631054e-05,
+      "loss": 0.0364,
+      "step": 66210
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.3885424968355916e-05,
+      "loss": 0.041,
+      "step": 66215
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.3884956167080777e-05,
+      "loss": 0.0625,
+      "step": 66220
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.3884487365805637e-05,
+      "loss": 0.0589,
+      "step": 66225
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.3884018564530497e-05,
+      "loss": 0.1163,
+      "step": 66230
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.3883549763255357e-05,
+      "loss": 0.0797,
+      "step": 66235
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.3883080961980217e-05,
+      "loss": 0.1096,
+      "step": 66240
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.3882612160705079e-05,
+      "loss": 0.1487,
+      "step": 66245
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.3882143359429939e-05,
+      "loss": 0.0677,
+      "step": 66250
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.3881674558154798e-05,
+      "loss": 0.0646,
+      "step": 66255
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.3881205756879658e-05,
+      "loss": 0.0805,
+      "step": 66260
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.3880736955604522e-05,
+      "loss": 0.1717,
+      "step": 66265
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.3880268154329382e-05,
+      "loss": 0.0432,
+      "step": 66270
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.3879799353054242e-05,
+      "loss": 0.1104,
+      "step": 66275
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.3879330551779102e-05,
+      "loss": 0.1144,
+      "step": 66280
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.3878861750503963e-05,
+      "loss": 0.1326,
+      "step": 66285
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.3878392949228823e-05,
+      "loss": 0.4025,
+      "step": 66290
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.3877924147953683e-05,
+      "loss": 0.3015,
+      "step": 66295
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.3877455346678543e-05,
+      "loss": 0.0762,
+      "step": 66300
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.3876986545403403e-05,
+      "loss": 0.0335,
+      "step": 66305
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.3876517744128266e-05,
+      "loss": 0.0364,
+      "step": 66310
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.3876048942853126e-05,
+      "loss": 0.0831,
+      "step": 66315
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.3875580141577986e-05,
+      "loss": 0.0909,
+      "step": 66320
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.3875111340302848e-05,
+      "loss": 0.0958,
+      "step": 66325
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.3874642539027708e-05,
+      "loss": 0.109,
+      "step": 66330
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.3874173737752568e-05,
+      "loss": 0.1368,
+      "step": 66335
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.3873704936477428e-05,
+      "loss": 0.2237,
+      "step": 66340
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.3873236135202287e-05,
+      "loss": 0.3289,
+      "step": 66345
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.3872767333927149e-05,
+      "loss": 0.0626,
+      "step": 66350
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.387229853265201e-05,
+      "loss": 0.0189,
+      "step": 66355
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.387182973137687e-05,
+      "loss": 0.0907,
+      "step": 66360
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.3871360930101732e-05,
+      "loss": 0.0416,
+      "step": 66365
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.3870892128826592e-05,
+      "loss": 0.0711,
+      "step": 66370
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.3870423327551452e-05,
+      "loss": 0.1234,
+      "step": 66375
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.3869954526276312e-05,
+      "loss": 0.1226,
+      "step": 66380
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.3869485725001172e-05,
+      "loss": 0.1751,
+      "step": 66385
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.3869016923726034e-05,
+      "loss": 0.2324,
+      "step": 66390
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.3868548122450894e-05,
+      "loss": 0.2637,
+      "step": 66395
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.3868079321175755e-05,
+      "loss": 0.0733,
+      "step": 66400
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.3867610519900617e-05,
+      "loss": 0.0436,
+      "step": 66405
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.3867141718625477e-05,
+      "loss": 0.0274,
+      "step": 66410
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.3866672917350337e-05,
+      "loss": 0.049,
+      "step": 66415
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.3866204116075197e-05,
+      "loss": 0.0902,
+      "step": 66420
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.3865735314800057e-05,
+      "loss": 0.1148,
+      "step": 66425
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.3865266513524918e-05,
+      "loss": 0.1224,
+      "step": 66430
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.3864797712249778e-05,
+      "loss": 0.2542,
+      "step": 66435
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.3864328910974638e-05,
+      "loss": 0.1721,
+      "step": 66440
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.3863860109699498e-05,
+      "loss": 0.314,
+      "step": 66445
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.3863391308424361e-05,
+      "loss": 0.0618,
+      "step": 66450
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.3862922507149221e-05,
+      "loss": 0.0659,
+      "step": 66455
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.3862453705874081e-05,
+      "loss": 0.0642,
+      "step": 66460
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.3861984904598941e-05,
+      "loss": 0.0547,
+      "step": 66465
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.3861516103323803e-05,
+      "loss": 0.0489,
+      "step": 66470
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.3861047302048663e-05,
+      "loss": 0.1195,
+      "step": 66475
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.3860578500773523e-05,
+      "loss": 0.2332,
+      "step": 66480
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.3860109699498383e-05,
+      "loss": 0.1851,
+      "step": 66485
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.3859640898223242e-05,
+      "loss": 0.142,
+      "step": 66490
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.3859172096948106e-05,
+      "loss": 0.2409,
+      "step": 66495
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.3858703295672966e-05,
+      "loss": 0.0916,
+      "step": 66500
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.3858234494397826e-05,
+      "loss": 0.025,
+      "step": 66505
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.3857765693122687e-05,
+      "loss": 0.0263,
+      "step": 66510
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.3857296891847547e-05,
+      "loss": 0.1125,
+      "step": 66515
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.3856828090572407e-05,
+      "loss": 0.0423,
+      "step": 66520
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.3856359289297267e-05,
+      "loss": 0.1117,
+      "step": 66525
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.3855890488022129e-05,
+      "loss": 0.1736,
+      "step": 66530
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.3855421686746989e-05,
+      "loss": 0.1654,
+      "step": 66535
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.385495288547185e-05,
+      "loss": 0.2988,
+      "step": 66540
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.385448408419671e-05,
+      "loss": 0.4467,
+      "step": 66545
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.3854015282921572e-05,
+      "loss": 0.0604,
+      "step": 66550
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.3853546481646432e-05,
+      "loss": 0.029,
+      "step": 66555
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.3853077680371292e-05,
+      "loss": 0.0642,
+      "step": 66560
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.3852608879096152e-05,
+      "loss": 0.0539,
+      "step": 66565
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.3852140077821013e-05,
+      "loss": 0.054,
+      "step": 66570
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.3851671276545873e-05,
+      "loss": 0.0307,
+      "step": 66575
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.3851202475270733e-05,
+      "loss": 0.0699,
+      "step": 66580
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.3850733673995593e-05,
+      "loss": 0.0521,
+      "step": 66585
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.3850264872720456e-05,
+      "loss": 0.2003,
+      "step": 66590
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.3849796071445316e-05,
+      "loss": 0.2092,
+      "step": 66595
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.3849327270170176e-05,
+      "loss": 0.0655,
+      "step": 66600
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.3848858468895036e-05,
+      "loss": 0.057,
+      "step": 66605
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.3848389667619898e-05,
+      "loss": 0.0298,
+      "step": 66610
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.3847920866344758e-05,
+      "loss": 0.0235,
+      "step": 66615
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.3847452065069618e-05,
+      "loss": 0.1046,
+      "step": 66620
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.3846983263794478e-05,
+      "loss": 0.0526,
+      "step": 66625
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.3846514462519338e-05,
+      "loss": 0.1352,
+      "step": 66630
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.3846045661244201e-05,
+      "loss": 0.1467,
+      "step": 66635
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.384557685996906e-05,
+      "loss": 0.3234,
+      "step": 66640
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.384510805869392e-05,
+      "loss": 0.4519,
+      "step": 66645
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.3844639257418782e-05,
+      "loss": 0.0615,
+      "step": 66650
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.3844170456143642e-05,
+      "loss": 0.0226,
+      "step": 66655
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.3843701654868502e-05,
+      "loss": 0.0299,
+      "step": 66660
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.3843232853593362e-05,
+      "loss": 0.0906,
+      "step": 66665
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.3842764052318222e-05,
+      "loss": 0.1091,
+      "step": 66670
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.3842295251043084e-05,
+      "loss": 0.0796,
+      "step": 66675
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.3841826449767945e-05,
+      "loss": 0.0716,
+      "step": 66680
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.3841357648492805e-05,
+      "loss": 0.2111,
+      "step": 66685
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.3840888847217667e-05,
+      "loss": 0.2017,
+      "step": 66690
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.3840420045942527e-05,
+      "loss": 0.2499,
+      "step": 66695
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.3839951244667387e-05,
+      "loss": 0.0438,
+      "step": 66700
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.3839482443392247e-05,
+      "loss": 0.0432,
+      "step": 66705
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.3839013642117107e-05,
+      "loss": 0.0451,
+      "step": 66710
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.3838544840841968e-05,
+      "loss": 0.0343,
+      "step": 66715
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.3838076039566828e-05,
+      "loss": 0.0664,
+      "step": 66720
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.383760723829169e-05,
+      "loss": 0.0706,
+      "step": 66725
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.3837138437016551e-05,
+      "loss": 0.1731,
+      "step": 66730
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.3836669635741411e-05,
+      "loss": 0.1815,
+      "step": 66735
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.3836200834466271e-05,
+      "loss": 0.1428,
+      "step": 66740
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.3835732033191131e-05,
+      "loss": 0.3716,
+      "step": 66745
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.3835263231915991e-05,
+      "loss": 0.0278,
+      "step": 66750
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.3834794430640853e-05,
+      "loss": 0.046,
+      "step": 66755
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.3834325629365713e-05,
+      "loss": 0.1035,
+      "step": 66760
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.3833856828090573e-05,
+      "loss": 0.1144,
+      "step": 66765
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.3833388026815433e-05,
+      "loss": 0.0783,
+      "step": 66770
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.3832919225540296e-05,
+      "loss": 0.0529,
+      "step": 66775
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.3832450424265156e-05,
+      "loss": 0.089,
+      "step": 66780
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.3831981622990016e-05,
+      "loss": 0.1363,
+      "step": 66785
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.3831512821714876e-05,
+      "loss": 0.212,
+      "step": 66790
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.3831044020439737e-05,
+      "loss": 0.1733,
+      "step": 66795
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.3830575219164597e-05,
+      "loss": 0.067,
+      "step": 66800
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.3830106417889457e-05,
+      "loss": 0.0803,
+      "step": 66805
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.3829637616614317e-05,
+      "loss": 0.0371,
+      "step": 66810
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.3829168815339177e-05,
+      "loss": 0.0515,
+      "step": 66815
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.382870001406404e-05,
+      "loss": 0.0837,
+      "step": 66820
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.38282312127889e-05,
+      "loss": 0.0697,
+      "step": 66825
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.382776241151376e-05,
+      "loss": 0.1489,
+      "step": 66830
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.3827293610238622e-05,
+      "loss": 0.1778,
+      "step": 66835
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.3826824808963482e-05,
+      "loss": 0.1401,
+      "step": 66840
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.3826356007688342e-05,
+      "loss": 0.3149,
+      "step": 66845
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.3825887206413202e-05,
+      "loss": 0.0503,
+      "step": 66850
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.3825418405138062e-05,
+      "loss": 0.027,
+      "step": 66855
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.3824949603862923e-05,
+      "loss": 0.0344,
+      "step": 66860
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.3824480802587785e-05,
+      "loss": 0.0539,
+      "step": 66865
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.3824012001312645e-05,
+      "loss": 0.0438,
+      "step": 66870
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.3823543200037506e-05,
+      "loss": 0.0387,
+      "step": 66875
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.3823074398762366e-05,
+      "loss": 0.1085,
+      "step": 66880
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.3822605597487226e-05,
+      "loss": 0.1261,
+      "step": 66885
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.3822136796212086e-05,
+      "loss": 0.2468,
+      "step": 66890
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.3821667994936946e-05,
+      "loss": 0.2092,
+      "step": 66895
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.3821199193661808e-05,
+      "loss": 0.0453,
+      "step": 66900
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.3820730392386668e-05,
+      "loss": 0.0149,
+      "step": 66905
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.3820261591111528e-05,
+      "loss": 0.0878,
+      "step": 66910
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.3819792789836391e-05,
+      "loss": 0.1053,
+      "step": 66915
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.3819323988561251e-05,
+      "loss": 0.1154,
+      "step": 66920
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.3818855187286111e-05,
+      "loss": 0.0957,
+      "step": 66925
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.3818386386010971e-05,
+      "loss": 0.1194,
+      "step": 66930
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.381791758473583e-05,
+      "loss": 0.1477,
+      "step": 66935
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.3817448783460692e-05,
+      "loss": 0.1528,
+      "step": 66940
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.3816979982185552e-05,
+      "loss": 0.3295,
+      "step": 66945
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.3816511180910412e-05,
+      "loss": 0.0486,
+      "step": 66950
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.3816042379635272e-05,
+      "loss": 0.0517,
+      "step": 66955
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.3815573578360136e-05,
+      "loss": 0.0239,
+      "step": 66960
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.3815104777084995e-05,
+      "loss": 0.0443,
+      "step": 66965
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.3814635975809855e-05,
+      "loss": 0.083,
+      "step": 66970
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.3814167174534715e-05,
+      "loss": 0.0842,
+      "step": 66975
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.3813698373259577e-05,
+      "loss": 0.071,
+      "step": 66980
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.3813229571984437e-05,
+      "loss": 0.1215,
+      "step": 66985
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.3812760770709297e-05,
+      "loss": 0.1975,
+      "step": 66990
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.3812291969434157e-05,
+      "loss": 0.312,
+      "step": 66995
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.3811823168159018e-05,
+      "loss": 0.0263,
+      "step": 67000
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.381135436688388e-05,
+      "loss": 0.022,
+      "step": 67005
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.381088556560874e-05,
+      "loss": 0.0698,
+      "step": 67010
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.38104167643336e-05,
+      "loss": 0.0665,
+      "step": 67015
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.3809947963058462e-05,
+      "loss": 0.0565,
+      "step": 67020
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.3809479161783321e-05,
+      "loss": 0.0716,
+      "step": 67025
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.3809010360508181e-05,
+      "loss": 0.0787,
+      "step": 67030
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.3808541559233041e-05,
+      "loss": 0.1136,
+      "step": 67035
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.3808072757957903e-05,
+      "loss": 0.1218,
+      "step": 67040
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.3807603956682763e-05,
+      "loss": 0.2986,
+      "step": 67045
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.3807135155407625e-05,
+      "loss": 0.0709,
+      "step": 67050
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.3806666354132484e-05,
+      "loss": 0.0823,
+      "step": 67055
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.3806197552857346e-05,
+      "loss": 0.0219,
+      "step": 67060
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.3805728751582206e-05,
+      "loss": 0.0453,
+      "step": 67065
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.3805259950307066e-05,
+      "loss": 0.0388,
+      "step": 67070
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.3804791149031926e-05,
+      "loss": 0.1246,
+      "step": 67075
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.3804322347756787e-05,
+      "loss": 0.1001,
+      "step": 67080
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.3803853546481647e-05,
+      "loss": 0.1526,
+      "step": 67085
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.3803384745206507e-05,
+      "loss": 0.2058,
+      "step": 67090
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.3802915943931367e-05,
+      "loss": 0.3274,
+      "step": 67095
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.380244714265623e-05,
+      "loss": 0.0589,
+      "step": 67100
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.380197834138109e-05,
+      "loss": 0.0554,
+      "step": 67105
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.380150954010595e-05,
+      "loss": 0.1107,
+      "step": 67110
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.380104073883081e-05,
+      "loss": 0.0373,
+      "step": 67115
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.3800571937555672e-05,
+      "loss": 0.1125,
+      "step": 67120
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.3800103136280532e-05,
+      "loss": 0.0933,
+      "step": 67125
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.3799634335005392e-05,
+      "loss": 0.0467,
+      "step": 67130
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.3799165533730252e-05,
+      "loss": 0.1961,
+      "step": 67135
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.3798696732455112e-05,
+      "loss": 0.1006,
+      "step": 67140
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.3798227931179975e-05,
+      "loss": 0.2407,
+      "step": 67145
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.3797759129904835e-05,
+      "loss": 0.0181,
+      "step": 67150
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.3797290328629695e-05,
+      "loss": 0.0193,
+      "step": 67155
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.3796821527354557e-05,
+      "loss": 0.0602,
+      "step": 67160
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.3796352726079417e-05,
+      "loss": 0.0818,
+      "step": 67165
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.3795883924804276e-05,
+      "loss": 0.0914,
+      "step": 67170
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.3795415123529136e-05,
+      "loss": 0.0872,
+      "step": 67175
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.3794946322253996e-05,
+      "loss": 0.1196,
+      "step": 67180
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.3794477520978858e-05,
+      "loss": 0.1093,
+      "step": 67185
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.379400871970372e-05,
+      "loss": 0.1819,
+      "step": 67190
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.379353991842858e-05,
+      "loss": 0.3717,
+      "step": 67195
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.3793071117153441e-05,
+      "loss": 0.0776,
+      "step": 67200
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.3792602315878301e-05,
+      "loss": 0.026,
+      "step": 67205
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.3792133514603161e-05,
+      "loss": 0.0606,
+      "step": 67210
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.3791664713328021e-05,
+      "loss": 0.0693,
+      "step": 67215
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.3791195912052881e-05,
+      "loss": 0.0327,
+      "step": 67220
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.3790727110777743e-05,
+      "loss": 0.0602,
+      "step": 67225
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.3790258309502602e-05,
+      "loss": 0.0975,
+      "step": 67230
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.3789789508227462e-05,
+      "loss": 0.1581,
+      "step": 67235
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.3789320706952326e-05,
+      "loss": 0.1922,
+      "step": 67240
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.3788851905677186e-05,
+      "loss": 0.1326,
+      "step": 67245
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.3788383104402046e-05,
+      "loss": 0.0908,
+      "step": 67250
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.3787914303126905e-05,
+      "loss": 0.0064,
+      "step": 67255
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.3787445501851765e-05,
+      "loss": 0.0279,
+      "step": 67260
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.3786976700576627e-05,
+      "loss": 0.0307,
+      "step": 67265
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.3786507899301487e-05,
+      "loss": 0.0717,
+      "step": 67270
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.3786039098026347e-05,
+      "loss": 0.0845,
+      "step": 67275
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.3785570296751207e-05,
+      "loss": 0.1032,
+      "step": 67280
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.378510149547607e-05,
+      "loss": 0.1643,
+      "step": 67285
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.378463269420093e-05,
+      "loss": 0.243,
+      "step": 67290
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.378416389292579e-05,
+      "loss": 0.3675,
+      "step": 67295
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.378369509165065e-05,
+      "loss": 0.0633,
+      "step": 67300
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.3783226290375512e-05,
+      "loss": 0.0273,
+      "step": 67305
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.3782757489100372e-05,
+      "loss": 0.047,
+      "step": 67310
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.3782288687825231e-05,
+      "loss": 0.0805,
+      "step": 67315
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.3781819886550091e-05,
+      "loss": 0.1093,
+      "step": 67320
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.3781351085274951e-05,
+      "loss": 0.0728,
+      "step": 67325
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.3780882283999815e-05,
+      "loss": 0.0528,
+      "step": 67330
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.3780413482724675e-05,
+      "loss": 0.1021,
+      "step": 67335
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.3779944681449535e-05,
+      "loss": 0.2603,
+      "step": 67340
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.3779475880174396e-05,
+      "loss": 0.2996,
+      "step": 67345
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.3779007078899256e-05,
+      "loss": 0.0848,
+      "step": 67350
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.3778538277624116e-05,
+      "loss": 0.0372,
+      "step": 67355
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.3778069476348976e-05,
+      "loss": 0.0614,
+      "step": 67360
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.3777600675073836e-05,
+      "loss": 0.0727,
+      "step": 67365
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.3777131873798698e-05,
+      "loss": 0.0873,
+      "step": 67370
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.3776663072523559e-05,
+      "loss": 0.1126,
+      "step": 67375
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.3776194271248419e-05,
+      "loss": 0.1982,
+      "step": 67380
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.377572546997328e-05,
+      "loss": 0.2663,
+      "step": 67385
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.377525666869814e-05,
+      "loss": 0.2309,
+      "step": 67390
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.3774787867423e-05,
+      "loss": 0.3462,
+      "step": 67395
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.377431906614786e-05,
+      "loss": 0.0791,
+      "step": 67400
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.377385026487272e-05,
+      "loss": 0.0226,
+      "step": 67405
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.3773381463597582e-05,
+      "loss": 0.0631,
+      "step": 67410
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.3772912662322442e-05,
+      "loss": 0.0799,
+      "step": 67415
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.3772443861047302e-05,
+      "loss": 0.0407,
+      "step": 67420
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.3771975059772165e-05,
+      "loss": 0.0385,
+      "step": 67425
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.3771506258497025e-05,
+      "loss": 0.0817,
+      "step": 67430
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.3771037457221885e-05,
+      "loss": 0.1169,
+      "step": 67435
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.3770568655946745e-05,
+      "loss": 0.1531,
+      "step": 67440
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.3770099854671605e-05,
+      "loss": 0.1626,
+      "step": 67445
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.3769631053396467e-05,
+      "loss": 0.0748,
+      "step": 67450
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.3769162252121327e-05,
+      "loss": 0.0773,
+      "step": 67455
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.3768693450846186e-05,
+      "loss": 0.0212,
+      "step": 67460
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.3768224649571046e-05,
+      "loss": 0.0787,
+      "step": 67465
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.376775584829591e-05,
+      "loss": 0.0926,
+      "step": 67470
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.376728704702077e-05,
+      "loss": 0.1528,
+      "step": 67475
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.376681824574563e-05,
+      "loss": 0.2009,
+      "step": 67480
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.376634944447049e-05,
+      "loss": 0.1525,
+      "step": 67485
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.3765880643195351e-05,
+      "loss": 0.2424,
+      "step": 67490
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.3765411841920211e-05,
+      "loss": 0.4591,
+      "step": 67495
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.3764943040645071e-05,
+      "loss": 0.0752,
+      "step": 67500
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.3764474239369931e-05,
+      "loss": 0.009,
+      "step": 67505
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.3764005438094793e-05,
+      "loss": 0.087,
+      "step": 67510
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.3763536636819654e-05,
+      "loss": 0.0634,
+      "step": 67515
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.3763067835544514e-05,
+      "loss": 0.0393,
+      "step": 67520
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.3762599034269374e-05,
+      "loss": 0.0283,
+      "step": 67525
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.3762130232994236e-05,
+      "loss": 0.1217,
+      "step": 67530
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.3761661431719096e-05,
+      "loss": 0.1377,
+      "step": 67535
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.3761192630443956e-05,
+      "loss": 0.2084,
+      "step": 67540
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.3760723829168816e-05,
+      "loss": 0.2998,
+      "step": 67545
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.3760255027893677e-05,
+      "loss": 0.0727,
+      "step": 67550
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.3759786226618537e-05,
+      "loss": 0.0368,
+      "step": 67555
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.3759317425343397e-05,
+      "loss": 0.0344,
+      "step": 67560
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.3758848624068259e-05,
+      "loss": 0.0686,
+      "step": 67565
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.375837982279312e-05,
+      "loss": 0.1097,
+      "step": 67570
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.375791102151798e-05,
+      "loss": 0.2716,
+      "step": 67575
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.375744222024284e-05,
+      "loss": 0.1313,
+      "step": 67580
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.37569734189677e-05,
+      "loss": 0.117,
+      "step": 67585
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.3756504617692562e-05,
+      "loss": 0.1824,
+      "step": 67590
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.3756035816417422e-05,
+      "loss": 0.1543,
+      "step": 67595
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.3755567015142282e-05,
+      "loss": 0.122,
+      "step": 67600
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.3755098213867142e-05,
+      "loss": 0.0074,
+      "step": 67605
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.3754629412592005e-05,
+      "loss": 0.0307,
+      "step": 67610
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.3754160611316865e-05,
+      "loss": 0.0396,
+      "step": 67615
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.3753691810041725e-05,
+      "loss": 0.0495,
+      "step": 67620
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.3753223008766585e-05,
+      "loss": 0.0975,
+      "step": 67625
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.3752754207491446e-05,
+      "loss": 0.1073,
+      "step": 67630
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.3752285406216306e-05,
+      "loss": 0.1438,
+      "step": 67635
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.3751816604941166e-05,
+      "loss": 0.2011,
+      "step": 67640
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.3751347803666026e-05,
+      "loss": 0.3405,
+      "step": 67645
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.3750879002390886e-05,
+      "loss": 0.0957,
+      "step": 67650
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.375041020111575e-05,
+      "loss": 0.0259,
+      "step": 67655
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.374994139984061e-05,
+      "loss": 0.0578,
+      "step": 67660
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.374947259856547e-05,
+      "loss": 0.1437,
+      "step": 67665
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.374900379729033e-05,
+      "loss": 0.0596,
+      "step": 67670
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.374853499601519e-05,
+      "loss": 0.0759,
+      "step": 67675
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.374806619474005e-05,
+      "loss": 0.0801,
+      "step": 67680
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.374759739346491e-05,
+      "loss": 0.1489,
+      "step": 67685
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.374712859218977e-05,
+      "loss": 0.0848,
+      "step": 67690
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.3746659790914632e-05,
+      "loss": 0.237,
+      "step": 67695
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.3746190989639492e-05,
+      "loss": 0.0866,
+      "step": 67700
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.3745722188364354e-05,
+      "loss": 0.0325,
+      "step": 67705
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.3745253387089215e-05,
+      "loss": 0.0331,
+      "step": 67710
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.3744784585814075e-05,
+      "loss": 0.0228,
+      "step": 67715
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.3744315784538935e-05,
+      "loss": 0.1033,
+      "step": 67720
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.3743846983263795e-05,
+      "loss": 0.1214,
+      "step": 67725
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.3743378181988655e-05,
+      "loss": 0.0933,
+      "step": 67730
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.3742909380713517e-05,
+      "loss": 0.1463,
+      "step": 67735
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.3742440579438377e-05,
+      "loss": 0.2893,
+      "step": 67740
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.3741971778163237e-05,
+      "loss": 0.3865,
+      "step": 67745
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.37415029768881e-05,
+      "loss": 0.0846,
+      "step": 67750
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.374103417561296e-05,
+      "loss": 0.0302,
+      "step": 67755
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.374056537433782e-05,
+      "loss": 0.0473,
+      "step": 67760
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.374009657306268e-05,
+      "loss": 0.018,
+      "step": 67765
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.373962777178754e-05,
+      "loss": 0.1362,
+      "step": 67770
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.3739158970512401e-05,
+      "loss": 0.057,
+      "step": 67775
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.3738690169237261e-05,
+      "loss": 0.1749,
+      "step": 67780
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.3738221367962121e-05,
+      "loss": 0.1016,
+      "step": 67785
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.3737752566686981e-05,
+      "loss": 0.1584,
+      "step": 67790
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.3737283765411844e-05,
+      "loss": 0.357,
+      "step": 67795
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.3736814964136704e-05,
+      "loss": 0.0303,
+      "step": 67800
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.3736346162861564e-05,
+      "loss": 0.0423,
+      "step": 67805
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.3735877361586424e-05,
+      "loss": 0.0347,
+      "step": 67810
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.3735408560311286e-05,
+      "loss": 0.0268,
+      "step": 67815
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.3734939759036146e-05,
+      "loss": 0.0234,
+      "step": 67820
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.3734470957761006e-05,
+      "loss": 0.0924,
+      "step": 67825
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.3734002156485866e-05,
+      "loss": 0.0546,
+      "step": 67830
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.3733533355210726e-05,
+      "loss": 0.0996,
+      "step": 67835
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.3733064553935589e-05,
+      "loss": 0.127,
+      "step": 67840
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.3732595752660449e-05,
+      "loss": 0.2586,
+      "step": 67845
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.3732126951385309e-05,
+      "loss": 0.0788,
+      "step": 67850
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.373165815011017e-05,
+      "loss": 0.0596,
+      "step": 67855
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.373118934883503e-05,
+      "loss": 0.0302,
+      "step": 67860
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.373072054755989e-05,
+      "loss": 0.0224,
+      "step": 67865
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.373025174628475e-05,
+      "loss": 0.0651,
+      "step": 67870
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.372978294500961e-05,
+      "loss": 0.1478,
+      "step": 67875
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.3729314143734472e-05,
+      "loss": 0.0709,
+      "step": 67880
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.3728845342459332e-05,
+      "loss": 0.1736,
+      "step": 67885
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.3728376541184193e-05,
+      "loss": 0.2604,
+      "step": 67890
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.3727907739909055e-05,
+      "loss": 0.3258,
+      "step": 67895
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.3727438938633915e-05,
+      "loss": 0.1177,
+      "step": 67900
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.3726970137358775e-05,
+      "loss": 0.0203,
+      "step": 67905
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.3726501336083635e-05,
+      "loss": 0.0373,
+      "step": 67910
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.3726032534808495e-05,
+      "loss": 0.0882,
+      "step": 67915
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.3725563733533356e-05,
+      "loss": 0.1121,
+      "step": 67920
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.3725094932258216e-05,
+      "loss": 0.1319,
+      "step": 67925
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.3724626130983076e-05,
+      "loss": 0.1484,
+      "step": 67930
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.372415732970794e-05,
+      "loss": 0.1881,
+      "step": 67935
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.37236885284328e-05,
+      "loss": 0.1164,
+      "step": 67940
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.372321972715766e-05,
+      "loss": 0.3235,
+      "step": 67945
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.372275092588252e-05,
+      "loss": 0.0919,
+      "step": 67950
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.3722282124607381e-05,
+      "loss": 0.0481,
+      "step": 67955
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.372181332333224e-05,
+      "loss": 0.0297,
+      "step": 67960
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.37213445220571e-05,
+      "loss": 0.0337,
+      "step": 67965
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.372087572078196e-05,
+      "loss": 0.0748,
+      "step": 67970
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.372040691950682e-05,
+      "loss": 0.0674,
+      "step": 67975
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.3719938118231684e-05,
+      "loss": 0.08,
+      "step": 67980
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.3719469316956544e-05,
+      "loss": 0.1805,
+      "step": 67985
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.3719000515681404e-05,
+      "loss": 0.1943,
+      "step": 67990
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.3718531714406265e-05,
+      "loss": 0.2545,
+      "step": 67995
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.3718062913131125e-05,
+      "loss": 0.077,
+      "step": 68000
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.3717594111855985e-05,
+      "loss": 0.0349,
+      "step": 68005
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.3717125310580845e-05,
+      "loss": 0.0293,
+      "step": 68010
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.3716656509305705e-05,
+      "loss": 0.1356,
+      "step": 68015
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.3716187708030567e-05,
+      "loss": 0.1144,
+      "step": 68020
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.3715718906755427e-05,
+      "loss": 0.1098,
+      "step": 68025
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.3715250105480288e-05,
+      "loss": 0.1073,
+      "step": 68030
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.371478130420515e-05,
+      "loss": 0.1827,
+      "step": 68035
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.371431250293001e-05,
+      "loss": 0.2102,
+      "step": 68040
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.371384370165487e-05,
+      "loss": 0.2809,
+      "step": 68045
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.371337490037973e-05,
+      "loss": 0.0934,
+      "step": 68050
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.371290609910459e-05,
+      "loss": 0.0633,
+      "step": 68055
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.3712437297829451e-05,
+      "loss": 0.0299,
+      "step": 68060
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.3711968496554311e-05,
+      "loss": 0.0281,
+      "step": 68065
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.3711499695279171e-05,
+      "loss": 0.0962,
+      "step": 68070
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.3711030894004035e-05,
+      "loss": 0.0608,
+      "step": 68075
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.3710562092728894e-05,
+      "loss": 0.1047,
+      "step": 68080
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.3710093291453754e-05,
+      "loss": 0.1032,
+      "step": 68085
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.3709624490178614e-05,
+      "loss": 0.2665,
+      "step": 68090
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.3709155688903474e-05,
+      "loss": 0.4392,
+      "step": 68095
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.3708686887628336e-05,
+      "loss": 0.06,
+      "step": 68100
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.3708218086353196e-05,
+      "loss": 0.0181,
+      "step": 68105
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.3707749285078056e-05,
+      "loss": 0.0404,
+      "step": 68110
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.3707280483802916e-05,
+      "loss": 0.0858,
+      "step": 68115
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.3706811682527779e-05,
+      "loss": 0.091,
+      "step": 68120
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.3706342881252639e-05,
+      "loss": 0.1108,
+      "step": 68125
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.3705874079977499e-05,
+      "loss": 0.1246,
+      "step": 68130
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.3705405278702359e-05,
+      "loss": 0.0982,
+      "step": 68135
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.370493647742722e-05,
+      "loss": 0.2201,
+      "step": 68140
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.370446767615208e-05,
+      "loss": 0.1963,
+      "step": 68145
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.370399887487694e-05,
+      "loss": 0.0729,
+      "step": 68150
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.37035300736018e-05,
+      "loss": 0.0392,
+      "step": 68155
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.370306127232666e-05,
+      "loss": 0.0932,
+      "step": 68160
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.3702592471051524e-05,
+      "loss": 0.0605,
+      "step": 68165
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.3702123669776383e-05,
+      "loss": 0.0771,
+      "step": 68170
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.3701654868501243e-05,
+      "loss": 0.1162,
+      "step": 68175
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.3701186067226105e-05,
+      "loss": 0.0961,
+      "step": 68180
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.3700717265950965e-05,
+      "loss": 0.1156,
+      "step": 68185
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.3700248464675825e-05,
+      "loss": 0.148,
+      "step": 68190
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.3699779663400685e-05,
+      "loss": 0.2526,
+      "step": 68195
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.3699310862125545e-05,
+      "loss": 0.0837,
+      "step": 68200
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.3698842060850406e-05,
+      "loss": 0.0659,
+      "step": 68205
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.3698373259575266e-05,
+      "loss": 0.0308,
+      "step": 68210
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.3697904458300128e-05,
+      "loss": 0.0924,
+      "step": 68215
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.369743565702499e-05,
+      "loss": 0.0603,
+      "step": 68220
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.369696685574985e-05,
+      "loss": 0.0701,
+      "step": 68225
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.369649805447471e-05,
+      "loss": 0.0573,
+      "step": 68230
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.369602925319957e-05,
+      "loss": 0.1506,
+      "step": 68235
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.369556045192443e-05,
+      "loss": 0.6045,
+      "step": 68240
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.3695091650649291e-05,
+      "loss": 0.303,
+      "step": 68245
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.3694622849374151e-05,
+      "loss": 0.0841,
+      "step": 68250
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.369415404809901e-05,
+      "loss": 0.0456,
+      "step": 68255
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.3693685246823874e-05,
+      "loss": 0.0308,
+      "step": 68260
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.3693216445548734e-05,
+      "loss": 0.0283,
+      "step": 68265
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.3692747644273594e-05,
+      "loss": 0.069,
+      "step": 68270
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.3692278842998454e-05,
+      "loss": 0.1158,
+      "step": 68275
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.3691810041723314e-05,
+      "loss": 0.1096,
+      "step": 68280
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.3691341240448175e-05,
+      "loss": 0.1556,
+      "step": 68285
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.3690872439173035e-05,
+      "loss": 0.1588,
+      "step": 68290
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.3690403637897895e-05,
+      "loss": 0.3156,
+      "step": 68295
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.3689934836622755e-05,
+      "loss": 0.1504,
+      "step": 68300
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.3689466035347619e-05,
+      "loss": 0.0297,
+      "step": 68305
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.3688997234072479e-05,
+      "loss": 0.0265,
+      "step": 68310
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.3688528432797338e-05,
+      "loss": 0.043,
+      "step": 68315
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.3688059631522198e-05,
+      "loss": 0.2241,
+      "step": 68320
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.368759083024706e-05,
+      "loss": 0.1192,
+      "step": 68325
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.368712202897192e-05,
+      "loss": 0.0722,
+      "step": 68330
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.368665322769678e-05,
+      "loss": 0.17,
+      "step": 68335
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.368618442642164e-05,
+      "loss": 0.171,
+      "step": 68340
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.3685715625146501e-05,
+      "loss": 0.2061,
+      "step": 68345
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.3685246823871361e-05,
+      "loss": 0.0677,
+      "step": 68350
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.3684778022596223e-05,
+      "loss": 0.022,
+      "step": 68355
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.3684309221321083e-05,
+      "loss": 0.0457,
+      "step": 68360
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.3683840420045945e-05,
+      "loss": 0.0643,
+      "step": 68365
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.3683371618770805e-05,
+      "loss": 0.0821,
+      "step": 68370
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.3682902817495664e-05,
+      "loss": 0.0636,
+      "step": 68375
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.3682434016220524e-05,
+      "loss": 0.1447,
+      "step": 68380
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.3681965214945386e-05,
+      "loss": 0.1489,
+      "step": 68385
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.3681496413670246e-05,
+      "loss": 0.1166,
+      "step": 68390
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.3681027612395106e-05,
+      "loss": 0.4197,
+      "step": 68395
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.3680558811119968e-05,
+      "loss": 0.0833,
+      "step": 68400
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.3680090009844829e-05,
+      "loss": 0.0463,
+      "step": 68405
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.3679621208569689e-05,
+      "loss": 0.051,
+      "step": 68410
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.3679152407294549e-05,
+      "loss": 0.0393,
+      "step": 68415
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.3678683606019409e-05,
+      "loss": 0.0909,
+      "step": 68420
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.367821480474427e-05,
+      "loss": 0.2237,
+      "step": 68425
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.367774600346913e-05,
+      "loss": 0.0817,
+      "step": 68430
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.367727720219399e-05,
+      "loss": 0.0923,
+      "step": 68435
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.367680840091885e-05,
+      "loss": 0.188,
+      "step": 68440
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.3676339599643714e-05,
+      "loss": 0.3209,
+      "step": 68445
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.3675870798368574e-05,
+      "loss": 0.0419,
+      "step": 68450
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.3675401997093434e-05,
+      "loss": 0.0279,
+      "step": 68455
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.3674933195818294e-05,
+      "loss": 0.0472,
+      "step": 68460
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.3674464394543155e-05,
+      "loss": 0.0767,
+      "step": 68465
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.3673995593268015e-05,
+      "loss": 0.039,
+      "step": 68470
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.3673526791992875e-05,
+      "loss": 0.1062,
+      "step": 68475
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.3673057990717735e-05,
+      "loss": 0.0748,
+      "step": 68480
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.3672589189442595e-05,
+      "loss": 0.2356,
+      "step": 68485
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.3672120388167458e-05,
+      "loss": 0.2333,
+      "step": 68490
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.3671651586892318e-05,
+      "loss": 0.2404,
+      "step": 68495
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.3671182785617178e-05,
+      "loss": 0.0869,
+      "step": 68500
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.367071398434204e-05,
+      "loss": 0.0531,
+      "step": 68505
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.36702451830669e-05,
+      "loss": 0.075,
+      "step": 68510
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.366977638179176e-05,
+      "loss": 0.0588,
+      "step": 68515
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.366930758051662e-05,
+      "loss": 0.0472,
+      "step": 68520
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.366883877924148e-05,
+      "loss": 0.0844,
+      "step": 68525
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.3668369977966341e-05,
+      "loss": 0.1068,
+      "step": 68530
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.3667901176691201e-05,
+      "loss": 0.2066,
+      "step": 68535
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.3667432375416063e-05,
+      "loss": 0.2088,
+      "step": 68540
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.3666963574140924e-05,
+      "loss": 0.3059,
+      "step": 68545
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.3666494772865784e-05,
+      "loss": 0.0681,
+      "step": 68550
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.3666025971590644e-05,
+      "loss": 0.0341,
+      "step": 68555
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.3665557170315504e-05,
+      "loss": 0.0286,
+      "step": 68560
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.3665088369040364e-05,
+      "loss": 0.0394,
+      "step": 68565
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.3664619567765226e-05,
+      "loss": 0.0279,
+      "step": 68570
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.3664150766490086e-05,
+      "loss": 0.0567,
+      "step": 68575
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.3663681965214945e-05,
+      "loss": 0.0947,
+      "step": 68580
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.3663213163939809e-05,
+      "loss": 0.126,
+      "step": 68585
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.3662744362664669e-05,
+      "loss": 0.1452,
+      "step": 68590
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.3662275561389529e-05,
+      "loss": 0.2892,
+      "step": 68595
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.3661806760114389e-05,
+      "loss": 0.0578,
+      "step": 68600
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.3661337958839249e-05,
+      "loss": 0.0673,
+      "step": 68605
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.366086915756411e-05,
+      "loss": 0.1013,
+      "step": 68610
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.366040035628897e-05,
+      "loss": 0.0439,
+      "step": 68615
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.365993155501383e-05,
+      "loss": 0.0907,
+      "step": 68620
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.365946275373869e-05,
+      "loss": 0.1349,
+      "step": 68625
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.3658993952463553e-05,
+      "loss": 0.0859,
+      "step": 68630
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.3658525151188413e-05,
+      "loss": 0.1666,
+      "step": 68635
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.3658056349913273e-05,
+      "loss": 0.1575,
+      "step": 68640
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.3657587548638133e-05,
+      "loss": 0.321,
+      "step": 68645
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.3657118747362995e-05,
+      "loss": 0.0574,
+      "step": 68650
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.3656649946087855e-05,
+      "loss": 0.054,
+      "step": 68655
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.3656181144812715e-05,
+      "loss": 0.0162,
+      "step": 68660
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.3655712343537574e-05,
+      "loss": 0.0822,
+      "step": 68665
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.3655243542262434e-05,
+      "loss": 0.0454,
+      "step": 68670
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.3654774740987296e-05,
+      "loss": 0.0711,
+      "step": 68675
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.3654305939712158e-05,
+      "loss": 0.1444,
+      "step": 68680
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.3653837138437018e-05,
+      "loss": 0.0962,
+      "step": 68685
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.365336833716188e-05,
+      "loss": 0.1342,
+      "step": 68690
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.365289953588674e-05,
+      "loss": 0.2585,
+      "step": 68695
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.3652430734611599e-05,
+      "loss": 0.0507,
+      "step": 68700
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.3651961933336459e-05,
+      "loss": 0.0703,
+      "step": 68705
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.3651493132061319e-05,
+      "loss": 0.0602,
+      "step": 68710
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.365102433078618e-05,
+      "loss": 0.0916,
+      "step": 68715
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.365055552951104e-05,
+      "loss": 0.0549,
+      "step": 68720
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.3650086728235902e-05,
+      "loss": 0.0701,
+      "step": 68725
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.3649617926960764e-05,
+      "loss": 0.0628,
+      "step": 68730
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.3649149125685624e-05,
+      "loss": 0.0906,
+      "step": 68735
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.3648680324410484e-05,
+      "loss": 0.1594,
+      "step": 68740
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.3648211523135344e-05,
+      "loss": 0.2705,
+      "step": 68745
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.3647742721860204e-05,
+      "loss": 0.0764,
+      "step": 68750
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.3647273920585065e-05,
+      "loss": 0.0282,
+      "step": 68755
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.3646805119309925e-05,
+      "loss": 0.068,
+      "step": 68760
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.3646336318034785e-05,
+      "loss": 0.0627,
+      "step": 68765
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.3645867516759648e-05,
+      "loss": 0.0624,
+      "step": 68770
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.3645398715484508e-05,
+      "loss": 0.0597,
+      "step": 68775
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.3644929914209368e-05,
+      "loss": 0.1637,
+      "step": 68780
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.3644461112934228e-05,
+      "loss": 0.1517,
+      "step": 68785
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.3643992311659088e-05,
+      "loss": 0.2563,
+      "step": 68790
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.364352351038395e-05,
+      "loss": 0.2644,
+      "step": 68795
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.364305470910881e-05,
+      "loss": 0.0429,
+      "step": 68800
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.364258590783367e-05,
+      "loss": 0.008,
+      "step": 68805
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.364211710655853e-05,
+      "loss": 0.0423,
+      "step": 68810
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.3641648305283393e-05,
+      "loss": 0.0706,
+      "step": 68815
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.3641179504008253e-05,
+      "loss": 0.0768,
+      "step": 68820
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.3640710702733113e-05,
+      "loss": 0.0976,
+      "step": 68825
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.3640241901457973e-05,
+      "loss": 0.0463,
+      "step": 68830
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.3639773100182834e-05,
+      "loss": 0.1814,
+      "step": 68835
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.3639304298907694e-05,
+      "loss": 0.0591,
+      "step": 68840
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.3638835497632554e-05,
+      "loss": 0.302,
+      "step": 68845
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.3638366696357414e-05,
+      "loss": 0.0831,
+      "step": 68850
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.3637897895082276e-05,
+      "loss": 0.0241,
+      "step": 68855
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.3637429093807136e-05,
+      "loss": 0.0345,
+      "step": 68860
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.3636960292531997e-05,
+      "loss": 0.0299,
+      "step": 68865
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.3636491491256857e-05,
+      "loss": 0.1228,
+      "step": 68870
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.3636022689981719e-05,
+      "loss": 0.0709,
+      "step": 68875
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.3635553888706579e-05,
+      "loss": 0.2013,
+      "step": 68880
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.3635085087431439e-05,
+      "loss": 0.1109,
+      "step": 68885
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.3634616286156299e-05,
+      "loss": 0.1237,
+      "step": 68890
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.363414748488116e-05,
+      "loss": 0.2264,
+      "step": 68895
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.363367868360602e-05,
+      "loss": 0.0495,
+      "step": 68900
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.363320988233088e-05,
+      "loss": 0.0278,
+      "step": 68905
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.3632741081055742e-05,
+      "loss": 0.0226,
+      "step": 68910
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.3632272279780603e-05,
+      "loss": 0.0756,
+      "step": 68915
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.3631803478505463e-05,
+      "loss": 0.0944,
+      "step": 68920
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.3631334677230323e-05,
+      "loss": 0.1001,
+      "step": 68925
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.3630865875955183e-05,
+      "loss": 0.0958,
+      "step": 68930
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.3630397074680045e-05,
+      "loss": 0.0488,
+      "step": 68935
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.3629928273404905e-05,
+      "loss": 0.2112,
+      "step": 68940
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.3629459472129765e-05,
+      "loss": 0.311,
+      "step": 68945
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.3628990670854625e-05,
+      "loss": 0.0435,
+      "step": 68950
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.3628521869579488e-05,
+      "loss": 0.0489,
+      "step": 68955
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.3628053068304348e-05,
+      "loss": 0.0378,
+      "step": 68960
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.3627584267029208e-05,
+      "loss": 0.0346,
+      "step": 68965
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.3627115465754068e-05,
+      "loss": 0.0223,
+      "step": 68970
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.362664666447893e-05,
+      "loss": 0.0454,
+      "step": 68975
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.362617786320379e-05,
+      "loss": 0.0608,
+      "step": 68980
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.362570906192865e-05,
+      "loss": 0.132,
+      "step": 68985
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.3625240260653509e-05,
+      "loss": 0.1778,
+      "step": 68990
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.3624771459378369e-05,
+      "loss": 0.3151,
+      "step": 68995
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.362430265810323e-05,
+      "loss": 0.0527,
+      "step": 69000
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.3623833856828092e-05,
+      "loss": 0.0827,
+      "step": 69005
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.3623365055552952e-05,
+      "loss": 0.0368,
+      "step": 69010
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.3622896254277814e-05,
+      "loss": 0.1629,
+      "step": 69015
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.3622427453002674e-05,
+      "loss": 0.1078,
+      "step": 69020
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.3621958651727534e-05,
+      "loss": 0.1879,
+      "step": 69025
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.3621489850452394e-05,
+      "loss": 0.1383,
+      "step": 69030
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.3621021049177254e-05,
+      "loss": 0.1159,
+      "step": 69035
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.3620552247902115e-05,
+      "loss": 0.3463,
+      "step": 69040
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.3620083446626975e-05,
+      "loss": 0.2195,
+      "step": 69045
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.3619614645351837e-05,
+      "loss": 0.0743,
+      "step": 69050
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.3619145844076698e-05,
+      "loss": 0.031,
+      "step": 69055
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.3618677042801558e-05,
+      "loss": 0.0395,
+      "step": 69060
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.3618208241526418e-05,
+      "loss": 0.0818,
+      "step": 69065
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.3617739440251278e-05,
+      "loss": 0.0429,
+      "step": 69070
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.3617270638976138e-05,
+      "loss": 0.077,
+      "step": 69075
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.3616801837701e-05,
+      "loss": 0.1415,
+      "step": 69080
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.361633303642586e-05,
+      "loss": 0.1771,
+      "step": 69085
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.361586423515072e-05,
+      "loss": 0.2503,
+      "step": 69090
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.3615395433875583e-05,
+      "loss": 0.2144,
+      "step": 69095
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.3614926632600443e-05,
+      "loss": 0.0527,
+      "step": 69100
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.3614457831325303e-05,
+      "loss": 0.0181,
+      "step": 69105
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.3613989030050163e-05,
+      "loss": 0.0112,
+      "step": 69110
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.3613520228775023e-05,
+      "loss": 0.1278,
+      "step": 69115
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.3613051427499884e-05,
+      "loss": 0.0671,
+      "step": 69120
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.3612582626224744e-05,
+      "loss": 0.0654,
+      "step": 69125
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.3612113824949604e-05,
+      "loss": 0.1508,
+      "step": 69130
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.3611645023674464e-05,
+      "loss": 0.1947,
+      "step": 69135
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.3611176222399327e-05,
+      "loss": 0.2565,
+      "step": 69140
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.3610707421124187e-05,
+      "loss": 0.2243,
+      "step": 69145
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.3610238619849047e-05,
+      "loss": 0.0372,
+      "step": 69150
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.3609769818573907e-05,
+      "loss": 0.025,
+      "step": 69155
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.3609301017298769e-05,
+      "loss": 0.0489,
+      "step": 69160
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.3608832216023629e-05,
+      "loss": 0.0977,
+      "step": 69165
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.3608363414748489e-05,
+      "loss": 0.0823,
+      "step": 69170
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.3607894613473349e-05,
+      "loss": 0.0635,
+      "step": 69175
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.3607425812198209e-05,
+      "loss": 0.0701,
+      "step": 69180
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.360695701092307e-05,
+      "loss": 0.1179,
+      "step": 69185
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.3606488209647932e-05,
+      "loss": 0.118,
+      "step": 69190
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.3606019408372792e-05,
+      "loss": 0.2901,
+      "step": 69195
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.3605550607097653e-05,
+      "loss": 0.082,
+      "step": 69200
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.3605081805822513e-05,
+      "loss": 0.0496,
+      "step": 69205
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.3604613004547373e-05,
+      "loss": 0.0609,
+      "step": 69210
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.3604144203272233e-05,
+      "loss": 0.1,
+      "step": 69215
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.3603675401997093e-05,
+      "loss": 0.0398,
+      "step": 69220
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.3603206600721955e-05,
+      "loss": 0.097,
+      "step": 69225
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.3602737799446815e-05,
+      "loss": 0.135,
+      "step": 69230
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.3602268998171676e-05,
+      "loss": 0.1163,
+      "step": 69235
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.3601800196896538e-05,
+      "loss": 0.2599,
+      "step": 69240
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.3601331395621398e-05,
+      "loss": 0.2447,
+      "step": 69245
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.3600862594346258e-05,
+      "loss": 0.0495,
+      "step": 69250
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.3600393793071118e-05,
+      "loss": 0.0333,
+      "step": 69255
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.3599924991795978e-05,
+      "loss": 0.0179,
+      "step": 69260
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.359945619052084e-05,
+      "loss": 0.061,
+      "step": 69265
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.35989873892457e-05,
+      "loss": 0.1018,
+      "step": 69270
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.359851858797056e-05,
+      "loss": 0.0976,
+      "step": 69275
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.3598049786695423e-05,
+      "loss": 0.1241,
+      "step": 69280
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.3597580985420283e-05,
+      "loss": 0.1119,
+      "step": 69285
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.3597112184145142e-05,
+      "loss": 0.1587,
+      "step": 69290
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.3596643382870002e-05,
+      "loss": 0.1354,
+      "step": 69295
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.3596174581594862e-05,
+      "loss": 0.0734,
+      "step": 69300
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.3595705780319724e-05,
+      "loss": 0.0203,
+      "step": 69305
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.3595236979044584e-05,
+      "loss": 0.0254,
+      "step": 69310
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.3594768177769444e-05,
+      "loss": 0.0609,
+      "step": 69315
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.3594299376494304e-05,
+      "loss": 0.1144,
+      "step": 69320
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.3593830575219165e-05,
+      "loss": 0.1696,
+      "step": 69325
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.3593361773944027e-05,
+      "loss": 0.1502,
+      "step": 69330
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.3592892972668887e-05,
+      "loss": 0.103,
+      "step": 69335
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.3592424171393747e-05,
+      "loss": 0.3154,
+      "step": 69340
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.3591955370118608e-05,
+      "loss": 0.2723,
+      "step": 69345
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.3591486568843468e-05,
+      "loss": 0.0576,
+      "step": 69350
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.3591017767568328e-05,
+      "loss": 0.0225,
+      "step": 69355
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.3590548966293188e-05,
+      "loss": 0.0424,
+      "step": 69360
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.359008016501805e-05,
+      "loss": 0.0495,
+      "step": 69365
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.358961136374291e-05,
+      "loss": 0.0497,
+      "step": 69370
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.3589142562467771e-05,
+      "loss": 0.0975,
+      "step": 69375
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.3588673761192633e-05,
+      "loss": 0.1527,
+      "step": 69380
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.3588204959917493e-05,
+      "loss": 0.1149,
+      "step": 69385
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.3587736158642353e-05,
+      "loss": 0.1807,
+      "step": 69390
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.3587267357367213e-05,
+      "loss": 0.1963,
+      "step": 69395
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.3586798556092073e-05,
+      "loss": 0.0231,
+      "step": 69400
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.3586329754816934e-05,
+      "loss": 0.0145,
+      "step": 69405
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.3585860953541794e-05,
+      "loss": 0.026,
+      "step": 69410
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.3585392152266654e-05,
+      "loss": 0.0397,
+      "step": 69415
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.3584923350991518e-05,
+      "loss": 0.0624,
+      "step": 69420
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.3584454549716378e-05,
+      "loss": 0.1085,
+      "step": 69425
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.3583985748441238e-05,
+      "loss": 0.1823,
+      "step": 69430
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.3583516947166097e-05,
+      "loss": 0.1782,
+      "step": 69435
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.3583048145890957e-05,
+      "loss": 0.2047,
+      "step": 69440
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.3582579344615819e-05,
+      "loss": 0.3008,
+      "step": 69445
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.3582110543340679e-05,
+      "loss": 0.0884,
+      "step": 69450
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.3581641742065539e-05,
+      "loss": 0.0262,
+      "step": 69455
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.3581172940790399e-05,
+      "loss": 0.0235,
+      "step": 69460
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.3580704139515262e-05,
+      "loss": 0.0497,
+      "step": 69465
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.3580235338240122e-05,
+      "loss": 0.037,
+      "step": 69470
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.3579766536964982e-05,
+      "loss": 0.0763,
+      "step": 69475
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.3579297735689842e-05,
+      "loss": 0.1038,
+      "step": 69480
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.3578828934414704e-05,
+      "loss": 0.2086,
+      "step": 69485
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.3578360133139563e-05,
+      "loss": 0.1528,
+      "step": 69490
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.3577891331864423e-05,
+      "loss": 0.2077,
+      "step": 69495
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.3577422530589283e-05,
+      "loss": 0.0728,
+      "step": 69500
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.3576953729314143e-05,
+      "loss": 0.0365,
+      "step": 69505
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.3576484928039005e-05,
+      "loss": 0.031,
+      "step": 69510
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.3576016126763867e-05,
+      "loss": 0.039,
+      "step": 69515
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.3575547325488726e-05,
+      "loss": 0.0433,
+      "step": 69520
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.3575078524213588e-05,
+      "loss": 0.0939,
+      "step": 69525
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.3574609722938448e-05,
+      "loss": 0.1011,
+      "step": 69530
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.3574140921663308e-05,
+      "loss": 0.1292,
+      "step": 69535
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.3573672120388168e-05,
+      "loss": 0.2253,
+      "step": 69540
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.3573203319113028e-05,
+      "loss": 0.3289,
+      "step": 69545
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.357273451783789e-05,
+      "loss": 0.0613,
+      "step": 69550
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.357226571656275e-05,
+      "loss": 0.0207,
+      "step": 69555
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.3571796915287611e-05,
+      "loss": 0.0707,
+      "step": 69560
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.3571328114012473e-05,
+      "loss": 0.1089,
+      "step": 69565
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.3570859312737333e-05,
+      "loss": 0.0599,
+      "step": 69570
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.3570390511462193e-05,
+      "loss": 0.1222,
+      "step": 69575
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.3569921710187052e-05,
+      "loss": 0.1018,
+      "step": 69580
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.3569452908911912e-05,
+      "loss": 0.1716,
+      "step": 69585
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.3568984107636774e-05,
+      "loss": 0.2636,
+      "step": 69590
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.3568515306361634e-05,
+      "loss": 0.2356,
+      "step": 69595
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.3568046505086494e-05,
+      "loss": 0.0779,
+      "step": 69600
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.3567577703811357e-05,
+      "loss": 0.0668,
+      "step": 69605
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.3567108902536217e-05,
+      "loss": 0.0626,
+      "step": 69610
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.3566640101261077e-05,
+      "loss": 0.0641,
+      "step": 69615
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.3566171299985937e-05,
+      "loss": 0.1447,
+      "step": 69620
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.3565702498710797e-05,
+      "loss": 0.0985,
+      "step": 69625
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.3565233697435659e-05,
+      "loss": 0.0882,
+      "step": 69630
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.3564764896160519e-05,
+      "loss": 0.1848,
+      "step": 69635
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.3564296094885378e-05,
+      "loss": 0.1694,
+      "step": 69640
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.3563827293610238e-05,
+      "loss": 0.2375,
+      "step": 69645
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.3563358492335098e-05,
+      "loss": 0.0846,
+      "step": 69650
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.3562889691059962e-05,
+      "loss": 0.0379,
+      "step": 69655
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.3562420889784822e-05,
+      "loss": 0.0343,
+      "step": 69660
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.3561952088509682e-05,
+      "loss": 0.0637,
+      "step": 69665
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.3561483287234543e-05,
+      "loss": 0.1055,
+      "step": 69670
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.3561014485959403e-05,
+      "loss": 0.0717,
+      "step": 69675
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.3560545684684263e-05,
+      "loss": 0.1089,
+      "step": 69680
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.3560076883409123e-05,
+      "loss": 0.1697,
+      "step": 69685
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.3559608082133985e-05,
+      "loss": 0.2271,
+      "step": 69690
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.3559139280858844e-05,
+      "loss": 0.2439,
+      "step": 69695
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.3558670479583706e-05,
+      "loss": 0.0538,
+      "step": 69700
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.3558201678308566e-05,
+      "loss": 0.0417,
+      "step": 69705
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.3557732877033428e-05,
+      "loss": 0.0547,
+      "step": 69710
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.3557264075758288e-05,
+      "loss": 0.1037,
+      "step": 69715
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.3556795274483148e-05,
+      "loss": 0.102,
+      "step": 69720
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.3556326473208007e-05,
+      "loss": 0.049,
+      "step": 69725
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.3555857671932869e-05,
+      "loss": 0.1282,
+      "step": 69730
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.3555388870657729e-05,
+      "loss": 0.1482,
+      "step": 69735
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.3554920069382589e-05,
+      "loss": 0.1134,
+      "step": 69740
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.355445126810745e-05,
+      "loss": 0.167,
+      "step": 69745
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.3553982466832312e-05,
+      "loss": 0.0429,
+      "step": 69750
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.3553513665557172e-05,
+      "loss": 0.0229,
+      "step": 69755
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.3553044864282032e-05,
+      "loss": 0.0685,
+      "step": 69760
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.3552576063006892e-05,
+      "loss": 0.0867,
+      "step": 69765
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.3552107261731754e-05,
+      "loss": 0.1179,
+      "step": 69770
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.3551638460456614e-05,
+      "loss": 0.0445,
+      "step": 69775
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.3551169659181474e-05,
+      "loss": 0.1649,
+      "step": 69780
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.3550700857906333e-05,
+      "loss": 0.1136,
+      "step": 69785
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.3550232056631197e-05,
+      "loss": 0.1493,
+      "step": 69790
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.3549763255356057e-05,
+      "loss": 0.3378,
+      "step": 69795
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.3549294454080917e-05,
+      "loss": 0.0582,
+      "step": 69800
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.3548825652805777e-05,
+      "loss": 0.0214,
+      "step": 69805
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.3548356851530638e-05,
+      "loss": 0.0592,
+      "step": 69810
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.3547888050255498e-05,
+      "loss": 0.0658,
+      "step": 69815
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.3547419248980358e-05,
+      "loss": 0.0869,
+      "step": 69820
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.3546950447705218e-05,
+      "loss": 0.1109,
+      "step": 69825
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.3546481646430078e-05,
+      "loss": 0.1647,
+      "step": 69830
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.354601284515494e-05,
+      "loss": 0.2254,
+      "step": 69835
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.3545544043879801e-05,
+      "loss": 0.249,
+      "step": 69840
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.3545075242604661e-05,
+      "loss": 0.232,
+      "step": 69845
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.3544606441329523e-05,
+      "loss": 0.0793,
+      "step": 69850
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.3544137640054383e-05,
+      "loss": 0.0497,
+      "step": 69855
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.3543668838779243e-05,
+      "loss": 0.0398,
+      "step": 69860
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.3543200037504103e-05,
+      "loss": 0.0742,
+      "step": 69865
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.3542731236228963e-05,
+      "loss": 0.0944,
+      "step": 69870
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.3542262434953824e-05,
+      "loss": 0.0623,
+      "step": 69875
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.3541793633678684e-05,
+      "loss": 0.088,
+      "step": 69880
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.3541324832403546e-05,
+      "loss": 0.1227,
+      "step": 69885
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.3540856031128407e-05,
+      "loss": 0.2153,
+      "step": 69890
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.3540387229853267e-05,
+      "loss": 0.2853,
+      "step": 69895
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.3539918428578127e-05,
+      "loss": 0.0896,
+      "step": 69900
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.3539449627302987e-05,
+      "loss": 0.0529,
+      "step": 69905
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.3538980826027847e-05,
+      "loss": 0.0535,
+      "step": 69910
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.3538512024752709e-05,
+      "loss": 0.071,
+      "step": 69915
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.3538043223477569e-05,
+      "loss": 0.0571,
+      "step": 69920
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.3537574422202429e-05,
+      "loss": 0.0381,
+      "step": 69925
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.3537105620927292e-05,
+      "loss": 0.174,
+      "step": 69930
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.3536636819652152e-05,
+      "loss": 0.1899,
+      "step": 69935
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.3536168018377012e-05,
+      "loss": 0.1559,
+      "step": 69940
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.3535699217101872e-05,
+      "loss": 0.3481,
+      "step": 69945
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.3535230415826732e-05,
+      "loss": 0.0759,
+      "step": 69950
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.3534761614551593e-05,
+      "loss": 0.0254,
+      "step": 69955
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.3534292813276453e-05,
+      "loss": 0.0272,
+      "step": 69960
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.3533824012001313e-05,
+      "loss": 0.0522,
+      "step": 69965
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.3533355210726173e-05,
+      "loss": 0.0842,
+      "step": 69970
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.3532886409451033e-05,
+      "loss": 0.1161,
+      "step": 69975
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.3532417608175896e-05,
+      "loss": 0.0978,
+      "step": 69980
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.3531948806900756e-05,
+      "loss": 0.0851,
+      "step": 69985
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.3531480005625616e-05,
+      "loss": 0.2412,
+      "step": 69990
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.3531011204350478e-05,
+      "loss": 0.291,
+      "step": 69995
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.3530542403075338e-05,
+      "loss": 0.0615,
+      "step": 70000
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.3530073601800198e-05,
+      "loss": 0.049,
+      "step": 70005
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.3529604800525058e-05,
+      "loss": 0.0562,
+      "step": 70010
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.3529135999249918e-05,
+      "loss": 0.0864,
+      "step": 70015
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.3528667197974779e-05,
+      "loss": 0.044,
+      "step": 70020
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.352819839669964e-05,
+      "loss": 0.1173,
+      "step": 70025
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.35277295954245e-05,
+      "loss": 0.1263,
+      "step": 70030
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.3527260794149362e-05,
+      "loss": 0.0906,
+      "step": 70035
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.3526791992874222e-05,
+      "loss": 0.2683,
+      "step": 70040
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.3526323191599082e-05,
+      "loss": 0.2091,
+      "step": 70045
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.3525854390323942e-05,
+      "loss": 0.1217,
+      "step": 70050
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.3525385589048802e-05,
+      "loss": 0.0359,
+      "step": 70055
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.3524916787773664e-05,
+      "loss": 0.0479,
+      "step": 70060
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.3524447986498524e-05,
+      "loss": 0.0724,
+      "step": 70065
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.3523979185223385e-05,
+      "loss": 0.0834,
+      "step": 70070
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.3523510383948247e-05,
+      "loss": 0.0925,
+      "step": 70075
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.3523041582673107e-05,
+      "loss": 0.173,
+      "step": 70080
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.3522572781397967e-05,
+      "loss": 0.1308,
+      "step": 70085
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.3522103980122827e-05,
+      "loss": 0.1221,
+      "step": 70090
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.3521635178847687e-05,
+      "loss": 0.3521,
+      "step": 70095
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.3521166377572548e-05,
+      "loss": 0.0234,
+      "step": 70100
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.3520697576297408e-05,
+      "loss": 0.0109,
+      "step": 70105
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.3520228775022268e-05,
+      "loss": 0.0234,
+      "step": 70110
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.3519759973747131e-05,
+      "loss": 0.0545,
+      "step": 70115
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.3519291172471991e-05,
+      "loss": 0.0509,
+      "step": 70120
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.3518822371196851e-05,
+      "loss": 0.1194,
+      "step": 70125
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.3518353569921711e-05,
+      "loss": 0.1149,
+      "step": 70130
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.3517884768646571e-05,
+      "loss": 0.1618,
+      "step": 70135
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.3517415967371433e-05,
+      "loss": 0.1269,
+      "step": 70140
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.3516947166096293e-05,
+      "loss": 0.3106,
+      "step": 70145
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.3516478364821153e-05,
+      "loss": 0.1003,
+      "step": 70150
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.3516009563546013e-05,
+      "loss": 0.0187,
+      "step": 70155
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.3515540762270874e-05,
+      "loss": 0.0811,
+      "step": 70160
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.3515071960995736e-05,
+      "loss": 0.008,
+      "step": 70165
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.3514603159720596e-05,
+      "loss": 0.0612,
+      "step": 70170
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.3514134358445456e-05,
+      "loss": 0.0913,
+      "step": 70175
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.3513665557170317e-05,
+      "loss": 0.1169,
+      "step": 70180
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.3513196755895177e-05,
+      "loss": 0.2338,
+      "step": 70185
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3512727954620037e-05,
+      "loss": 0.2062,
+      "step": 70190
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3512259153344897e-05,
+      "loss": 0.2692,
+      "step": 70195
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3511790352069759e-05,
+      "loss": 0.0322,
+      "step": 70200
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3511321550794619e-05,
+      "loss": 0.0478,
+      "step": 70205
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.351085274951948e-05,
+      "loss": 0.0719,
+      "step": 70210
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.351038394824434e-05,
+      "loss": 0.0476,
+      "step": 70215
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3509915146969202e-05,
+      "loss": 0.0856,
+      "step": 70220
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3509446345694062e-05,
+      "loss": 0.0642,
+      "step": 70225
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3508977544418922e-05,
+      "loss": 0.1311,
+      "step": 70230
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3508508743143782e-05,
+      "loss": 0.0779,
+      "step": 70235
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3508039941868643e-05,
+      "loss": 0.1567,
+      "step": 70240
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3507571140593503e-05,
+      "loss": 0.3294,
+      "step": 70245
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3507102339318363e-05,
+      "loss": 0.0495,
+      "step": 70250
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3506633538043225e-05,
+      "loss": 0.0297,
+      "step": 70255
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3506164736768086e-05,
+      "loss": 0.0638,
+      "step": 70260
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3505695935492946e-05,
+      "loss": 0.0689,
+      "step": 70265
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3505227134217806e-05,
+      "loss": 0.0922,
+      "step": 70270
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3504758332942666e-05,
+      "loss": 0.055,
+      "step": 70275
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3504289531667528e-05,
+      "loss": 0.0506,
+      "step": 70280
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3503820730392388e-05,
+      "loss": 0.2447,
+      "step": 70285
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3503351929117248e-05,
+      "loss": 0.152,
+      "step": 70290
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3502883127842108e-05,
+      "loss": 0.4533,
+      "step": 70295
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3502414326566968e-05,
+      "loss": 0.0593,
+      "step": 70300
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3501945525291831e-05,
+      "loss": 0.0576,
+      "step": 70305
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3501476724016691e-05,
+      "loss": 0.1412,
+      "step": 70310
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.350100792274155e-05,
+      "loss": 0.067,
+      "step": 70315
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3500539121466412e-05,
+      "loss": 0.0748,
+      "step": 70320
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3500070320191272e-05,
+      "loss": 0.0804,
+      "step": 70325
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3499601518916132e-05,
+      "loss": 0.1018,
+      "step": 70330
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3499132717640992e-05,
+      "loss": 0.2179,
+      "step": 70335
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3498663916365852e-05,
+      "loss": 0.1462,
+      "step": 70340
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3498195115090714e-05,
+      "loss": 0.3059,
+      "step": 70345
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3497726313815575e-05,
+      "loss": 0.0822,
+      "step": 70350
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3497257512540435e-05,
+      "loss": 0.0157,
+      "step": 70355
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3496788711265297e-05,
+      "loss": 0.0288,
+      "step": 70360
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3496319909990157e-05,
+      "loss": 0.0865,
+      "step": 70365
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3495851108715017e-05,
+      "loss": 0.0594,
+      "step": 70370
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3495382307439877e-05,
+      "loss": 0.0708,
+      "step": 70375
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3494913506164737e-05,
+      "loss": 0.0819,
+      "step": 70380
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3494444704889598e-05,
+      "loss": 0.3972,
+      "step": 70385
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3493975903614458e-05,
+      "loss": 0.2576,
+      "step": 70390
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.349350710233932e-05,
+      "loss": 0.3062,
+      "step": 70395
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3493038301064182e-05,
+      "loss": 0.0972,
+      "step": 70400
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3492569499789041e-05,
+      "loss": 0.0176,
+      "step": 70405
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3492100698513901e-05,
+      "loss": 0.0343,
+      "step": 70410
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3491631897238761e-05,
+      "loss": 0.0536,
+      "step": 70415
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3491163095963621e-05,
+      "loss": 0.0684,
+      "step": 70420
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3490694294688483e-05,
+      "loss": 0.1144,
+      "step": 70425
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3490225493413343e-05,
+      "loss": 0.1388,
+      "step": 70430
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3489756692138203e-05,
+      "loss": 0.1507,
+      "step": 70435
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3489287890863066e-05,
+      "loss": 0.2201,
+      "step": 70440
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3488819089587926e-05,
+      "loss": 0.3038,
+      "step": 70445
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3488350288312786e-05,
+      "loss": 0.0431,
+      "step": 70450
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3487881487037646e-05,
+      "loss": 0.0485,
+      "step": 70455
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3487412685762506e-05,
+      "loss": 0.0658,
+      "step": 70460
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3486943884487367e-05,
+      "loss": 0.0789,
+      "step": 70465
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3486475083212227e-05,
+      "loss": 0.0781,
+      "step": 70470
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3486006281937087e-05,
+      "loss": 0.0563,
+      "step": 70475
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3485537480661947e-05,
+      "loss": 0.1637,
+      "step": 70480
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3485068679386807e-05,
+      "loss": 0.1455,
+      "step": 70485
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.348459987811167e-05,
+      "loss": 0.169,
+      "step": 70490
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.348413107683653e-05,
+      "loss": 0.3557,
+      "step": 70495
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.348366227556139e-05,
+      "loss": 0.0485,
+      "step": 70500
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3483193474286252e-05,
+      "loss": 0.0281,
+      "step": 70505
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3482724673011112e-05,
+      "loss": 0.0363,
+      "step": 70510
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3482255871735972e-05,
+      "loss": 0.0769,
+      "step": 70515
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3481787070460832e-05,
+      "loss": 0.031,
+      "step": 70520
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3481318269185692e-05,
+      "loss": 0.081,
+      "step": 70525
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3480849467910553e-05,
+      "loss": 0.0901,
+      "step": 70530
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3480380666635415e-05,
+      "loss": 0.093,
+      "step": 70535
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3479911865360275e-05,
+      "loss": 0.1909,
+      "step": 70540
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3479443064085137e-05,
+      "loss": 0.2227,
+      "step": 70545
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3478974262809996e-05,
+      "loss": 0.0348,
+      "step": 70550
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3478505461534856e-05,
+      "loss": 0.0182,
+      "step": 70555
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3478036660259716e-05,
+      "loss": 0.024,
+      "step": 70560
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3477567858984576e-05,
+      "loss": 0.0567,
+      "step": 70565
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3477099057709438e-05,
+      "loss": 0.0617,
+      "step": 70570
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3476630256434298e-05,
+      "loss": 0.0485,
+      "step": 70575
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.347616145515916e-05,
+      "loss": 0.0487,
+      "step": 70580
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3475692653884021e-05,
+      "loss": 0.3816,
+      "step": 70585
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3475223852608881e-05,
+      "loss": 0.2829,
+      "step": 70590
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3474755051333741e-05,
+      "loss": 0.357,
+      "step": 70595
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3474286250058601e-05,
+      "loss": 0.0774,
+      "step": 70600
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.347381744878346e-05,
+      "loss": 0.0361,
+      "step": 70605
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3473348647508322e-05,
+      "loss": 0.0453,
+      "step": 70610
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3472879846233182e-05,
+      "loss": 0.0449,
+      "step": 70615
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3472411044958042e-05,
+      "loss": 0.0499,
+      "step": 70620
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3471942243682902e-05,
+      "loss": 0.048,
+      "step": 70625
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3471473442407766e-05,
+      "loss": 0.2074,
+      "step": 70630
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3471004641132626e-05,
+      "loss": 0.1713,
+      "step": 70635
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3470535839857485e-05,
+      "loss": 0.281,
+      "step": 70640
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3470067038582345e-05,
+      "loss": 0.2752,
+      "step": 70645
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3469598237307207e-05,
+      "loss": 0.0978,
+      "step": 70650
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3469129436032067e-05,
+      "loss": 0.0477,
+      "step": 70655
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3468660634756927e-05,
+      "loss": 0.0398,
+      "step": 70660
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3468191833481787e-05,
+      "loss": 0.0324,
+      "step": 70665
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3467723032206648e-05,
+      "loss": 0.0819,
+      "step": 70670
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.346725423093151e-05,
+      "loss": 0.076,
+      "step": 70675
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.346678542965637e-05,
+      "loss": 0.057,
+      "step": 70680
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.346631662838123e-05,
+      "loss": 0.1844,
+      "step": 70685
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3465847827106092e-05,
+      "loss": 0.2069,
+      "step": 70690
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3465379025830952e-05,
+      "loss": 0.1684,
+      "step": 70695
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3464910224555811e-05,
+      "loss": 0.0384,
+      "step": 70700
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3464441423280671e-05,
+      "loss": 0.0533,
+      "step": 70705
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3463972622005533e-05,
+      "loss": 0.0424,
+      "step": 70710
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3463503820730393e-05,
+      "loss": 0.0333,
+      "step": 70715
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3463035019455255e-05,
+      "loss": 0.0665,
+      "step": 70720
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3462566218180114e-05,
+      "loss": 0.0811,
+      "step": 70725
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3462097416904976e-05,
+      "loss": 0.0911,
+      "step": 70730
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3461628615629836e-05,
+      "loss": 0.2091,
+      "step": 70735
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3461159814354696e-05,
+      "loss": 0.2925,
+      "step": 70740
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3460691013079556e-05,
+      "loss": 0.2814,
+      "step": 70745
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3460222211804418e-05,
+      "loss": 0.0556,
+      "step": 70750
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3459753410529277e-05,
+      "loss": 0.0091,
+      "step": 70755
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3459284609254137e-05,
+      "loss": 0.0401,
+      "step": 70760
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3458815807978999e-05,
+      "loss": 0.0593,
+      "step": 70765
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.345834700670386e-05,
+      "loss": 0.046,
+      "step": 70770
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.345787820542872e-05,
+      "loss": 0.075,
+      "step": 70775
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.345740940415358e-05,
+      "loss": 0.0843,
+      "step": 70780
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.345694060287844e-05,
+      "loss": 0.2834,
+      "step": 70785
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3456471801603302e-05,
+      "loss": 0.354,
+      "step": 70790
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3456003000328162e-05,
+      "loss": 0.2956,
+      "step": 70795
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3455534199053022e-05,
+      "loss": 0.044,
+      "step": 70800
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3455065397777882e-05,
+      "loss": 0.0318,
+      "step": 70805
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3454596596502742e-05,
+      "loss": 0.0744,
+      "step": 70810
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3454127795227605e-05,
+      "loss": 0.0909,
+      "step": 70815
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3453658993952465e-05,
+      "loss": 0.0814,
+      "step": 70820
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3453190192677325e-05,
+      "loss": 0.1356,
+      "step": 70825
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3452721391402187e-05,
+      "loss": 0.1235,
+      "step": 70830
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3452252590127047e-05,
+      "loss": 0.0741,
+      "step": 70835
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3451783788851907e-05,
+      "loss": 0.1165,
+      "step": 70840
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3451314987576766e-05,
+      "loss": 0.3005,
+      "step": 70845
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3450846186301626e-05,
+      "loss": 0.0731,
+      "step": 70850
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3450377385026488e-05,
+      "loss": 0.034,
+      "step": 70855
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.344990858375135e-05,
+      "loss": 0.0277,
+      "step": 70860
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.344943978247621e-05,
+      "loss": 0.0857,
+      "step": 70865
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3448970981201071e-05,
+      "loss": 0.0274,
+      "step": 70870
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3448502179925931e-05,
+      "loss": 0.1408,
+      "step": 70875
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3448033378650791e-05,
+      "loss": 0.0956,
+      "step": 70880
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3447564577375651e-05,
+      "loss": 0.0891,
+      "step": 70885
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3447095776100511e-05,
+      "loss": 0.1134,
+      "step": 70890
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3446626974825373e-05,
+      "loss": 0.2421,
+      "step": 70895
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3446158173550233e-05,
+      "loss": 0.0405,
+      "step": 70900
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3445689372275094e-05,
+      "loss": 0.0647,
+      "step": 70905
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3445220570999956e-05,
+      "loss": 0.0569,
+      "step": 70910
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3444751769724816e-05,
+      "loss": 0.0657,
+      "step": 70915
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3444282968449676e-05,
+      "loss": 0.0433,
+      "step": 70920
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3443814167174536e-05,
+      "loss": 0.1249,
+      "step": 70925
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3443345365899395e-05,
+      "loss": 0.1427,
+      "step": 70930
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3442876564624257e-05,
+      "loss": 0.1352,
+      "step": 70935
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3442407763349117e-05,
+      "loss": 0.154,
+      "step": 70940
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3441938962073977e-05,
+      "loss": 0.2397,
+      "step": 70945
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3441470160798837e-05,
+      "loss": 0.0717,
+      "step": 70950
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.34410013595237e-05,
+      "loss": 0.0249,
+      "step": 70955
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.344053255824856e-05,
+      "loss": 0.076,
+      "step": 70960
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.344006375697342e-05,
+      "loss": 0.0485,
+      "step": 70965
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.343959495569828e-05,
+      "loss": 0.071,
+      "step": 70970
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3439126154423142e-05,
+      "loss": 0.18,
+      "step": 70975
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3438657353148002e-05,
+      "loss": 0.1468,
+      "step": 70980
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3438188551872862e-05,
+      "loss": 0.0761,
+      "step": 70985
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3437719750597721e-05,
+      "loss": 0.1708,
+      "step": 70990
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3437250949322581e-05,
+      "loss": 0.2662,
+      "step": 70995
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3436782148047445e-05,
+      "loss": 0.0523,
+      "step": 71000
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3436313346772305e-05,
+      "loss": 0.0473,
+      "step": 71005
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3435844545497165e-05,
+      "loss": 0.0661,
+      "step": 71010
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3435375744222026e-05,
+      "loss": 0.0808,
+      "step": 71015
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3434906942946886e-05,
+      "loss": 0.0806,
+      "step": 71020
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3434438141671746e-05,
+      "loss": 0.0982,
+      "step": 71025
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3433969340396606e-05,
+      "loss": 0.1275,
+      "step": 71030
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3433500539121466e-05,
+      "loss": 0.1666,
+      "step": 71035
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3433031737846328e-05,
+      "loss": 0.1886,
+      "step": 71040
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.343256293657119e-05,
+      "loss": 0.1818,
+      "step": 71045
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3432094135296049e-05,
+      "loss": 0.0647,
+      "step": 71050
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.343162533402091e-05,
+      "loss": 0.0322,
+      "step": 71055
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.343115653274577e-05,
+      "loss": 0.0411,
+      "step": 71060
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.343068773147063e-05,
+      "loss": 0.0963,
+      "step": 71065
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.343021893019549e-05,
+      "loss": 0.1114,
+      "step": 71070
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.342975012892035e-05,
+      "loss": 0.0667,
+      "step": 71075
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3429281327645212e-05,
+      "loss": 0.0841,
+      "step": 71080
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3428812526370072e-05,
+      "loss": 0.1584,
+      "step": 71085
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3428343725094934e-05,
+      "loss": 0.1747,
+      "step": 71090
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3427874923819795e-05,
+      "loss": 0.3115,
+      "step": 71095
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3427406122544655e-05,
+      "loss": 0.0685,
+      "step": 71100
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3426937321269515e-05,
+      "loss": 0.0193,
+      "step": 71105
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3426468519994375e-05,
+      "loss": 0.065,
+      "step": 71110
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3425999718719237e-05,
+      "loss": 0.0504,
+      "step": 71115
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3425530917444097e-05,
+      "loss": 0.0395,
+      "step": 71120
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3425062116168957e-05,
+      "loss": 0.0502,
+      "step": 71125
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3424593314893817e-05,
+      "loss": 0.0907,
+      "step": 71130
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3424124513618676e-05,
+      "loss": 0.1055,
+      "step": 71135
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.342365571234354e-05,
+      "loss": 0.0928,
+      "step": 71140
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.34231869110684e-05,
+      "loss": 0.343,
+      "step": 71145
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.342271810979326e-05,
+      "loss": 0.0368,
+      "step": 71150
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3422249308518121e-05,
+      "loss": 0.0282,
+      "step": 71155
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3421780507242981e-05,
+      "loss": 0.0772,
+      "step": 71160
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3421311705967841e-05,
+      "loss": 0.0514,
+      "step": 71165
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3420842904692701e-05,
+      "loss": 0.077,
+      "step": 71170
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3420374103417561e-05,
+      "loss": 0.1319,
+      "step": 71175
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3419905302142423e-05,
+      "loss": 0.1325,
+      "step": 71180
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3419436500867284e-05,
+      "loss": 0.1731,
+      "step": 71185
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3418967699592144e-05,
+      "loss": 0.2676,
+      "step": 71190
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3418498898317006e-05,
+      "loss": 0.4007,
+      "step": 71195
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3418030097041866e-05,
+      "loss": 0.1035,
+      "step": 71200
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3417561295766726e-05,
+      "loss": 0.1052,
+      "step": 71205
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3417092494491586e-05,
+      "loss": 0.0426,
+      "step": 71210
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3416623693216446e-05,
+      "loss": 0.0513,
+      "step": 71215
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3416154891941307e-05,
+      "loss": 0.0408,
+      "step": 71220
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3415686090666167e-05,
+      "loss": 0.1096,
+      "step": 71225
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3415217289391029e-05,
+      "loss": 0.0922,
+      "step": 71230
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.341474848811589e-05,
+      "loss": 0.1576,
+      "step": 71235
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.341427968684075e-05,
+      "loss": 0.118,
+      "step": 71240
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.341381088556561e-05,
+      "loss": 0.2277,
+      "step": 71245
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.341334208429047e-05,
+      "loss": 0.0906,
+      "step": 71250
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.341287328301533e-05,
+      "loss": 0.0256,
+      "step": 71255
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3412404481740192e-05,
+      "loss": 0.0239,
+      "step": 71260
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3411935680465052e-05,
+      "loss": 0.021,
+      "step": 71265
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3411466879189912e-05,
+      "loss": 0.0638,
+      "step": 71270
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3410998077914772e-05,
+      "loss": 0.0908,
+      "step": 71275
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3410529276639635e-05,
+      "loss": 0.1054,
+      "step": 71280
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3410060475364495e-05,
+      "loss": 0.1925,
+      "step": 71285
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3409591674089355e-05,
+      "loss": 0.2934,
+      "step": 71290
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3409122872814215e-05,
+      "loss": 0.3971,
+      "step": 71295
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3408654071539076e-05,
+      "loss": 0.0439,
+      "step": 71300
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3408185270263936e-05,
+      "loss": 0.0476,
+      "step": 71305
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3407716468988796e-05,
+      "loss": 0.0662,
+      "step": 71310
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3407247667713656e-05,
+      "loss": 0.0805,
+      "step": 71315
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3406778866438516e-05,
+      "loss": 0.0783,
+      "step": 71320
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.340631006516338e-05,
+      "loss": 0.047,
+      "step": 71325
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.340584126388824e-05,
+      "loss": 0.1047,
+      "step": 71330
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.34053724626131e-05,
+      "loss": 0.1287,
+      "step": 71335
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3404903661337961e-05,
+      "loss": 0.1862,
+      "step": 71340
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.340443486006282e-05,
+      "loss": 0.2412,
+      "step": 71345
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.340396605878768e-05,
+      "loss": 0.0788,
+      "step": 71350
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.340349725751254e-05,
+      "loss": 0.0357,
+      "step": 71355
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.34030284562374e-05,
+      "loss": 0.0551,
+      "step": 71360
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3402559654962262e-05,
+      "loss": 0.0659,
+      "step": 71365
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3402090853687124e-05,
+      "loss": 0.0647,
+      "step": 71370
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3401622052411984e-05,
+      "loss": 0.079,
+      "step": 71375
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3401153251136845e-05,
+      "loss": 0.112,
+      "step": 71380
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3400684449861705e-05,
+      "loss": 0.1248,
+      "step": 71385
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3400215648586565e-05,
+      "loss": 0.1932,
+      "step": 71390
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3399746847311425e-05,
+      "loss": 0.4284,
+      "step": 71395
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3399278046036285e-05,
+      "loss": 0.0631,
+      "step": 71400
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3398809244761147e-05,
+      "loss": 0.0085,
+      "step": 71405
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3398340443486007e-05,
+      "loss": 0.0515,
+      "step": 71410
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3397871642210868e-05,
+      "loss": 0.0528,
+      "step": 71415
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.339740284093573e-05,
+      "loss": 0.1182,
+      "step": 71420
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.339693403966059e-05,
+      "loss": 0.142,
+      "step": 71425
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.339646523838545e-05,
+      "loss": 0.0624,
+      "step": 71430
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.339599643711031e-05,
+      "loss": 0.1191,
+      "step": 71435
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.339552763583517e-05,
+      "loss": 0.2345,
+      "step": 71440
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3395058834560031e-05,
+      "loss": 0.2952,
+      "step": 71445
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3394590033284891e-05,
+      "loss": 0.0598,
+      "step": 71450
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3394121232009751e-05,
+      "loss": 0.0184,
+      "step": 71455
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3393652430734611e-05,
+      "loss": 0.0507,
+      "step": 71460
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3393183629459474e-05,
+      "loss": 0.0529,
+      "step": 71465
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3392714828184334e-05,
+      "loss": 0.0637,
+      "step": 71470
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3392246026909194e-05,
+      "loss": 0.1361,
+      "step": 71475
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3391777225634054e-05,
+      "loss": 0.0945,
+      "step": 71480
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3391308424358916e-05,
+      "loss": 0.0916,
+      "step": 71485
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3390839623083776e-05,
+      "loss": 0.0792,
+      "step": 71490
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3390370821808636e-05,
+      "loss": 0.3219,
+      "step": 71495
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3389902020533496e-05,
+      "loss": 0.0405,
+      "step": 71500
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3389433219258357e-05,
+      "loss": 0.0338,
+      "step": 71505
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3388964417983219e-05,
+      "loss": 0.053,
+      "step": 71510
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3388495616708079e-05,
+      "loss": 0.0713,
+      "step": 71515
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3388026815432939e-05,
+      "loss": 0.0541,
+      "step": 71520
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.33875580141578e-05,
+      "loss": 0.0825,
+      "step": 71525
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.338708921288266e-05,
+      "loss": 0.1124,
+      "step": 71530
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.338662041160752e-05,
+      "loss": 0.1191,
+      "step": 71535
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.338615161033238e-05,
+      "loss": 0.2045,
+      "step": 71540
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3385682809057242e-05,
+      "loss": 0.4005,
+      "step": 71545
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3385214007782102e-05,
+      "loss": 0.1029,
+      "step": 71550
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3384745206506963e-05,
+      "loss": 0.0426,
+      "step": 71555
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3384276405231823e-05,
+      "loss": 0.0687,
+      "step": 71560
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3383807603956685e-05,
+      "loss": 0.0593,
+      "step": 71565
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3383338802681545e-05,
+      "loss": 0.0722,
+      "step": 71570
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3382870001406405e-05,
+      "loss": 0.0619,
+      "step": 71575
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3382401200131265e-05,
+      "loss": 0.144,
+      "step": 71580
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3381932398856126e-05,
+      "loss": 0.1801,
+      "step": 71585
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3381463597580986e-05,
+      "loss": 0.2007,
+      "step": 71590
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3380994796305846e-05,
+      "loss": 0.2975,
+      "step": 71595
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3380525995030706e-05,
+      "loss": 0.066,
+      "step": 71600
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.338005719375557e-05,
+      "loss": 0.0438,
+      "step": 71605
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.337958839248043e-05,
+      "loss": 0.061,
+      "step": 71610
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.337911959120529e-05,
+      "loss": 0.0309,
+      "step": 71615
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.337865078993015e-05,
+      "loss": 0.0702,
+      "step": 71620
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3378181988655011e-05,
+      "loss": 0.0805,
+      "step": 71625
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3377713187379871e-05,
+      "loss": 0.0509,
+      "step": 71630
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.337724438610473e-05,
+      "loss": 0.1516,
+      "step": 71635
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.337677558482959e-05,
+      "loss": 0.1109,
+      "step": 71640
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.337630678355445e-05,
+      "loss": 0.2836,
+      "step": 71645
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3375837982279314e-05,
+      "loss": 0.0521,
+      "step": 71650
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3375369181004174e-05,
+      "loss": 0.0361,
+      "step": 71655
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3374900379729034e-05,
+      "loss": 0.0419,
+      "step": 71660
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3374431578453896e-05,
+      "loss": 0.0647,
+      "step": 71665
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3373962777178755e-05,
+      "loss": 0.0941,
+      "step": 71670
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3373493975903615e-05,
+      "loss": 0.1481,
+      "step": 71675
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3373025174628475e-05,
+      "loss": 0.1705,
+      "step": 71680
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3372556373353335e-05,
+      "loss": 0.1946,
+      "step": 71685
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3372087572078197e-05,
+      "loss": 0.2952,
+      "step": 71690
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3371618770803059e-05,
+      "loss": 0.3711,
+      "step": 71695
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3371149969527918e-05,
+      "loss": 0.0434,
+      "step": 71700
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.337068116825278e-05,
+      "loss": 0.0257,
+      "step": 71705
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.337021236697764e-05,
+      "loss": 0.0518,
+      "step": 71710
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.33697435657025e-05,
+      "loss": 0.041,
+      "step": 71715
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.336927476442736e-05,
+      "loss": 0.0884,
+      "step": 71720
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.336880596315222e-05,
+      "loss": 0.0258,
+      "step": 71725
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3368337161877081e-05,
+      "loss": 0.1479,
+      "step": 71730
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3367868360601941e-05,
+      "loss": 0.1033,
+      "step": 71735
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3367399559326803e-05,
+      "loss": 0.2549,
+      "step": 71740
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3366930758051665e-05,
+      "loss": 0.1745,
+      "step": 71745
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3366461956776525e-05,
+      "loss": 0.0645,
+      "step": 71750
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3365993155501384e-05,
+      "loss": 0.0305,
+      "step": 71755
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3365524354226244e-05,
+      "loss": 0.0346,
+      "step": 71760
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3365055552951104e-05,
+      "loss": 0.0896,
+      "step": 71765
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3364586751675966e-05,
+      "loss": 0.1029,
+      "step": 71770
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3364117950400826e-05,
+      "loss": 0.0547,
+      "step": 71775
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3363649149125686e-05,
+      "loss": 0.0774,
+      "step": 71780
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3363180347850546e-05,
+      "loss": 0.1384,
+      "step": 71785
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3362711546575409e-05,
+      "loss": 0.1239,
+      "step": 71790
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3362242745300269e-05,
+      "loss": 0.236,
+      "step": 71795
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3361773944025129e-05,
+      "loss": 0.0586,
+      "step": 71800
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3361305142749989e-05,
+      "loss": 0.0177,
+      "step": 71805
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.336083634147485e-05,
+      "loss": 0.0119,
+      "step": 71810
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.336036754019971e-05,
+      "loss": 0.059,
+      "step": 71815
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.335989873892457e-05,
+      "loss": 0.0429,
+      "step": 71820
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.335942993764943e-05,
+      "loss": 0.0979,
+      "step": 71825
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.335896113637429e-05,
+      "loss": 0.1286,
+      "step": 71830
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3358492335099154e-05,
+      "loss": 0.1293,
+      "step": 71835
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3358023533824014e-05,
+      "loss": 0.223,
+      "step": 71840
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3357554732548873e-05,
+      "loss": 0.1792,
+      "step": 71845
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3357085931273735e-05,
+      "loss": 0.0705,
+      "step": 71850
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3356617129998595e-05,
+      "loss": 0.0182,
+      "step": 71855
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3356148328723455e-05,
+      "loss": 0.0345,
+      "step": 71860
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3355679527448315e-05,
+      "loss": 0.0467,
+      "step": 71865
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3355210726173175e-05,
+      "loss": 0.0458,
+      "step": 71870
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3354741924898036e-05,
+      "loss": 0.081,
+      "step": 71875
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3354273123622898e-05,
+      "loss": 0.1493,
+      "step": 71880
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3353804322347758e-05,
+      "loss": 0.1628,
+      "step": 71885
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.335333552107262e-05,
+      "loss": 0.1667,
+      "step": 71890
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.335286671979748e-05,
+      "loss": 0.2669,
+      "step": 71895
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.335239791852234e-05,
+      "loss": 0.0896,
+      "step": 71900
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.33519291172472e-05,
+      "loss": 0.0787,
+      "step": 71905
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.335146031597206e-05,
+      "loss": 0.0165,
+      "step": 71910
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.3350991514696921e-05,
+      "loss": 0.0706,
+      "step": 71915
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.3350522713421781e-05,
+      "loss": 0.0208,
+      "step": 71920
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.3350053912146641e-05,
+      "loss": 0.0542,
+      "step": 71925
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.3349585110871504e-05,
+      "loss": 0.1398,
+      "step": 71930
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.3349116309596364e-05,
+      "loss": 0.1703,
+      "step": 71935
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.3348647508321224e-05,
+      "loss": 0.1941,
+      "step": 71940
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.3348178707046084e-05,
+      "loss": 0.2125,
+      "step": 71945
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.3347709905770944e-05,
+      "loss": 0.069,
+      "step": 71950
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.3347241104495806e-05,
+      "loss": 0.0295,
+      "step": 71955
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.3346772303220665e-05,
+      "loss": 0.0448,
+      "step": 71960
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.3346303501945525e-05,
+      "loss": 0.033,
+      "step": 71965
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.3345834700670385e-05,
+      "loss": 0.0627,
+      "step": 71970
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.3345365899395249e-05,
+      "loss": 0.1029,
+      "step": 71975
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.3344897098120109e-05,
+      "loss": 0.1823,
+      "step": 71980
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.3344428296844969e-05,
+      "loss": 0.1794,
+      "step": 71985
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.3343959495569828e-05,
+      "loss": 0.2201,
+      "step": 71990
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.334349069429469e-05,
+      "loss": 0.5181,
+      "step": 71995
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.334302189301955e-05,
+      "loss": 0.083,
+      "step": 72000
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.334255309174441e-05,
+      "loss": 0.0247,
+      "step": 72005
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.334208429046927e-05,
+      "loss": 0.0589,
+      "step": 72010
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.3341615489194132e-05,
+      "loss": 0.0756,
+      "step": 72015
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.3341146687918993e-05,
+      "loss": 0.0276,
+      "step": 72020
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.3340677886643853e-05,
+      "loss": 0.0197,
+      "step": 72025
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.3340209085368713e-05,
+      "loss": 0.0849,
+      "step": 72030
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.3339740284093575e-05,
+      "loss": 0.1106,
+      "step": 72035
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.3339271482818435e-05,
+      "loss": 0.1956,
+      "step": 72040
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.3338802681543295e-05,
+      "loss": 0.2315,
+      "step": 72045
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.3338333880268154e-05,
+      "loss": 0.0434,
+      "step": 72050
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.3337865078993016e-05,
+      "loss": 0.0338,
+      "step": 72055
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.3337396277717876e-05,
+      "loss": 0.0314,
+      "step": 72060
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.3336927476442738e-05,
+      "loss": 0.0524,
+      "step": 72065
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.3336458675167598e-05,
+      "loss": 0.061,
+      "step": 72070
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.333598987389246e-05,
+      "loss": 0.0573,
+      "step": 72075
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.3335521072617319e-05,
+      "loss": 0.0893,
+      "step": 72080
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.3335052271342179e-05,
+      "loss": 0.1066,
+      "step": 72085
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.3334583470067039e-05,
+      "loss": 0.2155,
+      "step": 72090
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.33341146687919e-05,
+      "loss": 0.3624,
+      "step": 72095
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.333364586751676e-05,
+      "loss": 0.0759,
+      "step": 72100
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.333317706624162e-05,
+      "loss": 0.0728,
+      "step": 72105
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.333270826496648e-05,
+      "loss": 0.0529,
+      "step": 72110
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.3332239463691344e-05,
+      "loss": 0.059,
+      "step": 72115
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.3331770662416204e-05,
+      "loss": 0.0703,
+      "step": 72120
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.3331301861141064e-05,
+      "loss": 0.0768,
+      "step": 72125
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.3330833059865924e-05,
+      "loss": 0.127,
+      "step": 72130
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.3330364258590785e-05,
+      "loss": 0.1048,
+      "step": 72135
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.3329895457315645e-05,
+      "loss": 0.3312,
+      "step": 72140
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.3329426656040505e-05,
+      "loss": 0.3051,
+      "step": 72145
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.3328957854765365e-05,
+      "loss": 0.0489,
+      "step": 72150
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.3328489053490225e-05,
+      "loss": 0.0036,
+      "step": 72155
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.3328020252215088e-05,
+      "loss": 0.0514,
+      "step": 72160
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.3327551450939948e-05,
+      "loss": 0.0611,
+      "step": 72165
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.3327082649664808e-05,
+      "loss": 0.0991,
+      "step": 72170
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.332661384838967e-05,
+      "loss": 0.0367,
+      "step": 72175
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.332614504711453e-05,
+      "loss": 0.1138,
+      "step": 72180
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.332567624583939e-05,
+      "loss": 0.221,
+      "step": 72185
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.332520744456425e-05,
+      "loss": 0.1647,
+      "step": 72190
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.332473864328911e-05,
+      "loss": 0.3998,
+      "step": 72195
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.3324269842013971e-05,
+      "loss": 0.0491,
+      "step": 72200
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.3323801040738833e-05,
+      "loss": 0.0598,
+      "step": 72205
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.3323332239463693e-05,
+      "loss": 0.0408,
+      "step": 72210
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.3322863438188554e-05,
+      "loss": 0.0376,
+      "step": 72215
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.3322394636913414e-05,
+      "loss": 0.0411,
+      "step": 72220
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.3321925835638274e-05,
+      "loss": 0.1011,
+      "step": 72225
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.3321457034363134e-05,
+      "loss": 0.0658,
+      "step": 72230
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.3320988233087994e-05,
+      "loss": 0.0787,
+      "step": 72235
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.3320519431812856e-05,
+      "loss": 0.1882,
+      "step": 72240
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.3320050630537716e-05,
+      "loss": 0.3239,
+      "step": 72245
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.3319581829262576e-05,
+      "loss": 0.097,
+      "step": 72250
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.3319113027987439e-05,
+      "loss": 0.0214,
+      "step": 72255
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.3318644226712299e-05,
+      "loss": 0.0313,
+      "step": 72260
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.3318175425437159e-05,
+      "loss": 0.1312,
+      "step": 72265
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.3317706624162019e-05,
+      "loss": 0.0164,
+      "step": 72270
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.3317237822886879e-05,
+      "loss": 0.0828,
+      "step": 72275
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.331676902161174e-05,
+      "loss": 0.1816,
+      "step": 72280
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.33163002203366e-05,
+      "loss": 0.1546,
+      "step": 72285
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.331583141906146e-05,
+      "loss": 0.2239,
+      "step": 72290
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.331536261778632e-05,
+      "loss": 0.3649,
+      "step": 72295
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.3314893816511183e-05,
+      "loss": 0.0784,
+      "step": 72300
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.3314425015236043e-05,
+      "loss": 0.0467,
+      "step": 72305
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.3313956213960903e-05,
+      "loss": 0.0673,
+      "step": 72310
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.3313487412685763e-05,
+      "loss": 0.0611,
+      "step": 72315
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.3313018611410625e-05,
+      "loss": 0.054,
+      "step": 72320
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.3312549810135485e-05,
+      "loss": 0.229,
+      "step": 72325
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.3312081008860345e-05,
+      "loss": 0.0997,
+      "step": 72330
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.3311612207585205e-05,
+      "loss": 0.1775,
+      "step": 72335
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.3311143406310064e-05,
+      "loss": 0.2334,
+      "step": 72340
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.3310674605034928e-05,
+      "loss": 0.1633,
+      "step": 72345
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.3310205803759788e-05,
+      "loss": 0.0776,
+      "step": 72350
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.3309737002484648e-05,
+      "loss": 0.0365,
+      "step": 72355
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.330926820120951e-05,
+      "loss": 0.0346,
+      "step": 72360
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.330879939993437e-05,
+      "loss": 0.0875,
+      "step": 72365
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.330833059865923e-05,
+      "loss": 0.0641,
+      "step": 72370
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.3307861797384089e-05,
+      "loss": 0.0793,
+      "step": 72375
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.3307392996108949e-05,
+      "loss": 0.0476,
+      "step": 72380
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.330692419483381e-05,
+      "loss": 0.1841,
+      "step": 72385
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.3306455393558672e-05,
+      "loss": 0.2389,
+      "step": 72390
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.3305986592283532e-05,
+      "loss": 0.399,
+      "step": 72395
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.3305517791008394e-05,
+      "loss": 0.0409,
+      "step": 72400
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.3305048989733254e-05,
+      "loss": 0.018,
+      "step": 72405
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.3304580188458114e-05,
+      "loss": 0.0465,
+      "step": 72410
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.3304111387182974e-05,
+      "loss": 0.0337,
+      "step": 72415
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.3303642585907834e-05,
+      "loss": 0.0483,
+      "step": 72420
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.3303173784632695e-05,
+      "loss": 0.0754,
+      "step": 72425
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.3302704983357555e-05,
+      "loss": 0.0944,
+      "step": 72430
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.3302236182082415e-05,
+      "loss": 0.1161,
+      "step": 72435
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.3301767380807278e-05,
+      "loss": 0.1641,
+      "step": 72440
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.3301298579532138e-05,
+      "loss": 0.2003,
+      "step": 72445
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.3300829778256998e-05,
+      "loss": 0.0955,
+      "step": 72450
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.3300360976981858e-05,
+      "loss": 0.0162,
+      "step": 72455
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.3299892175706718e-05,
+      "loss": 0.0436,
+      "step": 72460
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.329942337443158e-05,
+      "loss": 0.0443,
+      "step": 72465
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.329895457315644e-05,
+      "loss": 0.0388,
+      "step": 72470
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.32984857718813e-05,
+      "loss": 0.0516,
+      "step": 72475
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.329801697060616e-05,
+      "loss": 0.1215,
+      "step": 72480
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.3297548169331023e-05,
+      "loss": 0.1472,
+      "step": 72485
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.3297079368055883e-05,
+      "loss": 0.1324,
+      "step": 72490
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.3296610566780743e-05,
+      "loss": 0.3926,
+      "step": 72495
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.3296141765505603e-05,
+      "loss": 0.0776,
+      "step": 72500
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.3295672964230464e-05,
+      "loss": 0.0354,
+      "step": 72505
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.3295204162955324e-05,
+      "loss": 0.0843,
+      "step": 72510
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.3294735361680184e-05,
+      "loss": 0.0604,
+      "step": 72515
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.3294266560405044e-05,
+      "loss": 0.0771,
+      "step": 72520
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.3293797759129906e-05,
+      "loss": 0.0755,
+      "step": 72525
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.3293328957854767e-05,
+      "loss": 0.1144,
+      "step": 72530
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.3292860156579627e-05,
+      "loss": 0.1494,
+      "step": 72535
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.3292391355304489e-05,
+      "loss": 0.1948,
+      "step": 72540
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.3291922554029349e-05,
+      "loss": 0.3107,
+      "step": 72545
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.3291453752754209e-05,
+      "loss": 0.0527,
+      "step": 72550
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.3290984951479069e-05,
+      "loss": 0.0309,
+      "step": 72555
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.3290516150203929e-05,
+      "loss": 0.0505,
+      "step": 72560
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.329004734892879e-05,
+      "loss": 0.0256,
+      "step": 72565
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.328957854765365e-05,
+      "loss": 0.0659,
+      "step": 72570
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.328910974637851e-05,
+      "loss": 0.0804,
+      "step": 72575
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.3288640945103373e-05,
+      "loss": 0.0771,
+      "step": 72580
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.3288172143828233e-05,
+      "loss": 0.21,
+      "step": 72585
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.3287703342553093e-05,
+      "loss": 0.233,
+      "step": 72590
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.3287234541277953e-05,
+      "loss": 0.4205,
+      "step": 72595
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.3286765740002813e-05,
+      "loss": 0.0724,
+      "step": 72600
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.3286296938727675e-05,
+      "loss": 0.0352,
+      "step": 72605
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.3285828137452535e-05,
+      "loss": 0.0254,
+      "step": 72610
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.3285359336177395e-05,
+      "loss": 0.0759,
+      "step": 72615
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.3284890534902255e-05,
+      "loss": 0.0816,
+      "step": 72620
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.3284421733627118e-05,
+      "loss": 0.0902,
+      "step": 72625
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.3283952932351978e-05,
+      "loss": 0.088,
+      "step": 72630
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.3283484131076838e-05,
+      "loss": 0.0907,
+      "step": 72635
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.3283015329801698e-05,
+      "loss": 0.1708,
+      "step": 72640
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.328254652852656e-05,
+      "loss": 0.2892,
+      "step": 72645
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.328207772725142e-05,
+      "loss": 0.0518,
+      "step": 72650
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.328160892597628e-05,
+      "loss": 0.0324,
+      "step": 72655
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.328114012470114e-05,
+      "loss": 0.0135,
+      "step": 72660
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.3280671323425999e-05,
+      "loss": 0.0409,
+      "step": 72665
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.3280202522150862e-05,
+      "loss": 0.0561,
+      "step": 72670
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.3279733720875722e-05,
+      "loss": 0.0636,
+      "step": 72675
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.3279264919600582e-05,
+      "loss": 0.0697,
+      "step": 72680
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.3278796118325444e-05,
+      "loss": 0.1301,
+      "step": 72685
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.3278327317050304e-05,
+      "loss": 0.1251,
+      "step": 72690
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.3277858515775164e-05,
+      "loss": 0.1802,
+      "step": 72695
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.3277389714500024e-05,
+      "loss": 0.0625,
+      "step": 72700
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.3276920913224884e-05,
+      "loss": 0.0129,
+      "step": 72705
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.3276452111949745e-05,
+      "loss": 0.0352,
+      "step": 72710
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.3275983310674607e-05,
+      "loss": 0.0247,
+      "step": 72715
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.3275514509399467e-05,
+      "loss": 0.0596,
+      "step": 72720
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.3275045708124329e-05,
+      "loss": 0.1163,
+      "step": 72725
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.3274576906849188e-05,
+      "loss": 0.1109,
+      "step": 72730
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.3274108105574048e-05,
+      "loss": 0.1857,
+      "step": 72735
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.3273639304298908e-05,
+      "loss": 0.2121,
+      "step": 72740
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.3273170503023768e-05,
+      "loss": 0.5159,
+      "step": 72745
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.327270170174863e-05,
+      "loss": 0.0355,
+      "step": 72750
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.327223290047349e-05,
+      "loss": 0.0755,
+      "step": 72755
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.327176409919835e-05,
+      "loss": 0.0358,
+      "step": 72760
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.3271295297923213e-05,
+      "loss": 0.0475,
+      "step": 72765
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.3270826496648073e-05,
+      "loss": 0.0639,
+      "step": 72770
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.3270357695372933e-05,
+      "loss": 0.0674,
+      "step": 72775
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.3269888894097793e-05,
+      "loss": 0.0946,
+      "step": 72780
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.3269420092822653e-05,
+      "loss": 0.1409,
+      "step": 72785
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.3268951291547514e-05,
+      "loss": 0.1262,
+      "step": 72790
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.3268482490272374e-05,
+      "loss": 0.2623,
+      "step": 72795
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.3268013688997234e-05,
+      "loss": 0.0845,
+      "step": 72800
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.3267544887722094e-05,
+      "loss": 0.0133,
+      "step": 72805
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.3267076086446958e-05,
+      "loss": 0.0781,
+      "step": 72810
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.3266607285171817e-05,
+      "loss": 0.021,
+      "step": 72815
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.3266138483896677e-05,
+      "loss": 0.1001,
+      "step": 72820
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.3265669682621537e-05,
+      "loss": 0.1239,
+      "step": 72825
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.3265200881346399e-05,
+      "loss": 0.1063,
+      "step": 72830
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.3264732080071259e-05,
+      "loss": 0.0912,
+      "step": 72835
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.3264263278796119e-05,
+      "loss": 0.2155,
+      "step": 72840
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.3263794477520979e-05,
+      "loss": 0.2851,
+      "step": 72845
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.3263325676245839e-05,
+      "loss": 0.0518,
+      "step": 72850
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.3262856874970702e-05,
+      "loss": 0.0152,
+      "step": 72855
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.3262388073695562e-05,
+      "loss": 0.0193,
+      "step": 72860
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.3261919272420422e-05,
+      "loss": 0.093,
+      "step": 72865
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.3261450471145284e-05,
+      "loss": 0.0918,
+      "step": 72870
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.3260981669870143e-05,
+      "loss": 0.0729,
+      "step": 72875
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.3260512868595003e-05,
+      "loss": 0.1065,
+      "step": 72880
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.3260044067319863e-05,
+      "loss": 0.1023,
+      "step": 72885
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.3259575266044725e-05,
+      "loss": 0.2665,
+      "step": 72890
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.3259106464769585e-05,
+      "loss": 0.1425,
+      "step": 72895
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.3258637663494445e-05,
+      "loss": 0.1003,
+      "step": 72900
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.3258168862219306e-05,
+      "loss": 0.0328,
+      "step": 72905
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.3257700060944168e-05,
+      "loss": 0.025,
+      "step": 72910
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.3257231259669028e-05,
+      "loss": 0.0856,
+      "step": 72915
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.3256762458393888e-05,
+      "loss": 0.0648,
+      "step": 72920
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.3256293657118748e-05,
+      "loss": 0.1638,
+      "step": 72925
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.325582485584361e-05,
+      "loss": 0.1195,
+      "step": 72930
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.325535605456847e-05,
+      "loss": 0.0723,
+      "step": 72935
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.325488725329333e-05,
+      "loss": 0.193,
+      "step": 72940
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.325441845201819e-05,
+      "loss": 0.3241,
+      "step": 72945
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.3253949650743053e-05,
+      "loss": 0.0862,
+      "step": 72950
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.3253480849467913e-05,
+      "loss": 0.0328,
+      "step": 72955
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.3253012048192772e-05,
+      "loss": 0.0547,
+      "step": 72960
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.3252543246917632e-05,
+      "loss": 0.0331,
+      "step": 72965
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.3252074445642494e-05,
+      "loss": 0.0603,
+      "step": 72970
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.3251605644367354e-05,
+      "loss": 0.1321,
+      "step": 72975
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.3251136843092214e-05,
+      "loss": 0.0986,
+      "step": 72980
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.3250668041817074e-05,
+      "loss": 0.1706,
+      "step": 72985
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.3250199240541934e-05,
+      "loss": 0.1643,
+      "step": 72990
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.3249730439266797e-05,
+      "loss": 0.2303,
+      "step": 72995
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.3249261637991657e-05,
+      "loss": 0.0926,
+      "step": 73000
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.3248792836716517e-05,
+      "loss": 0.0249,
+      "step": 73005
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.3248324035441379e-05,
+      "loss": 0.0443,
+      "step": 73010
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.3247855234166239e-05,
+      "loss": 0.0351,
+      "step": 73015
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.3247386432891098e-05,
+      "loss": 0.1067,
+      "step": 73020
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.3246917631615958e-05,
+      "loss": 0.0794,
+      "step": 73025
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.3246448830340818e-05,
+      "loss": 0.0929,
+      "step": 73030
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.324598002906568e-05,
+      "loss": 0.1354,
+      "step": 73035
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.3245511227790542e-05,
+      "loss": 0.2728,
+      "step": 73040
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.3245042426515402e-05,
+      "loss": 0.2449,
+      "step": 73045
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.3244573625240263e-05,
+      "loss": 0.0582,
+      "step": 73050
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.3244104823965123e-05,
+      "loss": 0.0717,
+      "step": 73055
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.3243636022689983e-05,
+      "loss": 0.037,
+      "step": 73060
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.3243167221414843e-05,
+      "loss": 0.066,
+      "step": 73065
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.3242698420139703e-05,
+      "loss": 0.1115,
+      "step": 73070
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.3242229618864565e-05,
+      "loss": 0.057,
+      "step": 73075
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.3241760817589424e-05,
+      "loss": 0.0675,
+      "step": 73080
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.3241292016314284e-05,
+      "loss": 0.1726,
+      "step": 73085
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.3240823215039148e-05,
+      "loss": 0.1837,
+      "step": 73090
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.3240354413764008e-05,
+      "loss": 0.2014,
+      "step": 73095
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.3239885612488868e-05,
+      "loss": 0.0381,
+      "step": 73100
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.3239416811213728e-05,
+      "loss": 0.0316,
+      "step": 73105
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.3238948009938587e-05,
+      "loss": 0.046,
+      "step": 73110
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.3238479208663449e-05,
+      "loss": 0.0846,
+      "step": 73115
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.3238010407388309e-05,
+      "loss": 0.1258,
+      "step": 73120
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.3237541606113169e-05,
+      "loss": 0.0894,
+      "step": 73125
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.3237072804838029e-05,
+      "loss": 0.0433,
+      "step": 73130
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.3236604003562892e-05,
+      "loss": 0.137,
+      "step": 73135
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.3236135202287752e-05,
+      "loss": 0.104,
+      "step": 73140
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.3235666401012612e-05,
+      "loss": 0.2982,
+      "step": 73145
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.3235197599737472e-05,
+      "loss": 0.0598,
+      "step": 73150
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.3234728798462334e-05,
+      "loss": 0.0419,
+      "step": 73155
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.3234259997187194e-05,
+      "loss": 0.0112,
+      "step": 73160
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.3233791195912053e-05,
+      "loss": 0.0433,
+      "step": 73165
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.3233322394636913e-05,
+      "loss": 0.1115,
+      "step": 73170
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.3232853593361773e-05,
+      "loss": 0.0278,
+      "step": 73175
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.3232384792086637e-05,
+      "loss": 0.1111,
+      "step": 73180
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.3231915990811497e-05,
+      "loss": 0.1012,
+      "step": 73185
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.3231447189536357e-05,
+      "loss": 0.1412,
+      "step": 73190
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.3230978388261218e-05,
+      "loss": 0.1721,
+      "step": 73195
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.3230509586986078e-05,
+      "loss": 0.093,
+      "step": 73200
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.3230040785710938e-05,
+      "loss": 0.0337,
+      "step": 73205
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.3229571984435798e-05,
+      "loss": 0.0732,
+      "step": 73210
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.3229103183160658e-05,
+      "loss": 0.0859,
+      "step": 73215
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.322863438188552e-05,
+      "loss": 0.0842,
+      "step": 73220
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.322816558061038e-05,
+      "loss": 0.043,
+      "step": 73225
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.3227696779335241e-05,
+      "loss": 0.1238,
+      "step": 73230
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.3227227978060103e-05,
+      "loss": 0.1395,
+      "step": 73235
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.3226759176784963e-05,
+      "loss": 0.161,
+      "step": 73240
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.3226290375509823e-05,
+      "loss": 0.1628,
+      "step": 73245
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.3225821574234683e-05,
+      "loss": 0.0807,
+      "step": 73250
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.3225352772959542e-05,
+      "loss": 0.0274,
+      "step": 73255
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.3224883971684404e-05,
+      "loss": 0.0187,
+      "step": 73260
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.3224415170409264e-05,
+      "loss": 0.0621,
+      "step": 73265
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.3223946369134124e-05,
+      "loss": 0.0558,
+      "step": 73270
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.3223477567858987e-05,
+      "loss": 0.077,
+      "step": 73275
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.3223008766583847e-05,
+      "loss": 0.0494,
+      "step": 73280
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.3222539965308707e-05,
+      "loss": 0.1365,
+      "step": 73285
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.3222071164033567e-05,
+      "loss": 0.1824,
+      "step": 73290
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.3221602362758427e-05,
+      "loss": 0.4051,
+      "step": 73295
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.3221133561483289e-05,
+      "loss": 0.065,
+      "step": 73300
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.3220664760208149e-05,
+      "loss": 0.0081,
+      "step": 73305
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.3220195958933009e-05,
+      "loss": 0.0359,
+      "step": 73310
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.3219727157657868e-05,
+      "loss": 0.0216,
+      "step": 73315
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.3219258356382732e-05,
+      "loss": 0.1492,
+      "step": 73320
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.3218789555107592e-05,
+      "loss": 0.1478,
+      "step": 73325
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.3218320753832452e-05,
+      "loss": 0.1755,
+      "step": 73330
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.3217851952557312e-05,
+      "loss": 0.1802,
+      "step": 73335
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.3217383151282173e-05,
+      "loss": 0.2669,
+      "step": 73340
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.3216914350007033e-05,
+      "loss": 0.2471,
+      "step": 73345
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.3216445548731893e-05,
+      "loss": 0.053,
+      "step": 73350
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.3215976747456753e-05,
+      "loss": 0.0466,
+      "step": 73355
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.3215507946181615e-05,
+      "loss": 0.013,
+      "step": 73360
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.3215039144906476e-05,
+      "loss": 0.0486,
+      "step": 73365
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.3214570343631336e-05,
+      "loss": 0.0598,
+      "step": 73370
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.3214101542356196e-05,
+      "loss": 0.0522,
+      "step": 73375
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.3213632741081058e-05,
+      "loss": 0.1005,
+      "step": 73380
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.3213163939805918e-05,
+      "loss": 0.075,
+      "step": 73385
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.3212695138530778e-05,
+      "loss": 0.2631,
+      "step": 73390
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.3212226337255638e-05,
+      "loss": 0.2347,
+      "step": 73395
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.32117575359805e-05,
+      "loss": 0.0567,
+      "step": 73400
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.3211288734705359e-05,
+      "loss": 0.0364,
+      "step": 73405
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.3210819933430219e-05,
+      "loss": 0.023,
+      "step": 73410
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.321035113215508e-05,
+      "loss": 0.0908,
+      "step": 73415
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.3209882330879942e-05,
+      "loss": 0.0697,
+      "step": 73420
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.3209413529604802e-05,
+      "loss": 0.0808,
+      "step": 73425
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.3208944728329662e-05,
+      "loss": 0.0747,
+      "step": 73430
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.3208475927054522e-05,
+      "loss": 0.1391,
+      "step": 73435
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.3208007125779384e-05,
+      "loss": 0.3241,
+      "step": 73440
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.3207538324504244e-05,
+      "loss": 0.2686,
+      "step": 73445
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.3207069523229104e-05,
+      "loss": 0.0759,
+      "step": 73450
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.3206600721953964e-05,
+      "loss": 0.026,
+      "step": 73455
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.3206131920678827e-05,
+      "loss": 0.0234,
+      "step": 73460
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.3205663119403687e-05,
+      "loss": 0.0817,
+      "step": 73465
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.3205194318128547e-05,
+      "loss": 0.0864,
+      "step": 73470
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.3204725516853407e-05,
+      "loss": 0.0828,
+      "step": 73475
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.3204256715578268e-05,
+      "loss": 0.0999,
+      "step": 73480
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.3203787914303128e-05,
+      "loss": 0.1457,
+      "step": 73485
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.3203319113027988e-05,
+      "loss": 0.1746,
+      "step": 73490
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.3202850311752848e-05,
+      "loss": 0.201,
+      "step": 73495
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.3202381510477708e-05,
+      "loss": 0.0825,
+      "step": 73500
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.3201912709202571e-05,
+      "loss": 0.017,
+      "step": 73505
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.3201443907927431e-05,
+      "loss": 0.0643,
+      "step": 73510
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.3200975106652291e-05,
+      "loss": 0.0708,
+      "step": 73515
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.3200506305377153e-05,
+      "loss": 0.0419,
+      "step": 73520
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.3200037504102013e-05,
+      "loss": 0.1007,
+      "step": 73525
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.3199568702826873e-05,
+      "loss": 0.1826,
+      "step": 73530
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.3199099901551733e-05,
+      "loss": 0.136,
+      "step": 73535
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.3198631100276593e-05,
+      "loss": 0.1944,
+      "step": 73540
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.3198162299001454e-05,
+      "loss": 0.4173,
+      "step": 73545
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.3197693497726314e-05,
+      "loss": 0.105,
+      "step": 73550
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.3197224696451176e-05,
+      "loss": 0.0587,
+      "step": 73555
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.3196755895176037e-05,
+      "loss": 0.0138,
+      "step": 73560
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.3196287093900897e-05,
+      "loss": 0.0695,
+      "step": 73565
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.3195818292625757e-05,
+      "loss": 0.0544,
+      "step": 73570
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.3195349491350617e-05,
+      "loss": 0.0769,
+      "step": 73575
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.3194880690075477e-05,
+      "loss": 0.0871,
+      "step": 73580
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.3194411888800339e-05,
+      "loss": 0.116,
+      "step": 73585
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.3193943087525199e-05,
+      "loss": 0.2275,
+      "step": 73590
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.3193474286250059e-05,
+      "loss": 0.207,
+      "step": 73595
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.3193005484974922e-05,
+      "loss": 0.052,
+      "step": 73600
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.3192536683699782e-05,
+      "loss": 0.077,
+      "step": 73605
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.3192067882424642e-05,
+      "loss": 0.0335,
+      "step": 73610
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.3191599081149502e-05,
+      "loss": 0.0224,
+      "step": 73615
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.3191130279874362e-05,
+      "loss": 0.0545,
+      "step": 73620
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.3190661478599223e-05,
+      "loss": 0.1288,
+      "step": 73625
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.3190192677324083e-05,
+      "loss": 0.117,
+      "step": 73630
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.3189723876048943e-05,
+      "loss": 0.1976,
+      "step": 73635
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.3189255074773803e-05,
+      "loss": 0.1778,
+      "step": 73640
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.3188786273498666e-05,
+      "loss": 0.2328,
+      "step": 73645
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.3188317472223526e-05,
+      "loss": 0.0614,
+      "step": 73650
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.3187848670948386e-05,
+      "loss": 0.0183,
+      "step": 73655
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.3187379869673246e-05,
+      "loss": 0.0346,
+      "step": 73660
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.3186911068398108e-05,
+      "loss": 0.0467,
+      "step": 73665
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.3186442267122968e-05,
+      "loss": 0.0368,
+      "step": 73670
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.3185973465847828e-05,
+      "loss": 0.0949,
+      "step": 73675
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.3185504664572688e-05,
+      "loss": 0.092,
+      "step": 73680
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.3185035863297548e-05,
+      "loss": 0.1353,
+      "step": 73685
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.3184567062022411e-05,
+      "loss": 0.2217,
+      "step": 73690
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.318409826074727e-05,
+      "loss": 0.2987,
+      "step": 73695
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.318362945947213e-05,
+      "loss": 0.0678,
+      "step": 73700
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.3183160658196992e-05,
+      "loss": 0.0372,
+      "step": 73705
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.3182691856921852e-05,
+      "loss": 0.0267,
+      "step": 73710
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.3182223055646712e-05,
+      "loss": 0.0613,
+      "step": 73715
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.3181754254371572e-05,
+      "loss": 0.0556,
+      "step": 73720
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.3181285453096432e-05,
+      "loss": 0.111,
+      "step": 73725
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.3180816651821294e-05,
+      "loss": 0.0856,
+      "step": 73730
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.3180347850546154e-05,
+      "loss": 0.1449,
+      "step": 73735
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.3179879049271015e-05,
+      "loss": 0.1621,
+      "step": 73740
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.3179410247995877e-05,
+      "loss": 0.2623,
+      "step": 73745
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.3178941446720737e-05,
+      "loss": 0.0542,
+      "step": 73750
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.3178472645445597e-05,
+      "loss": 0.0172,
+      "step": 73755
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.3178003844170457e-05,
+      "loss": 0.0505,
+      "step": 73760
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.3177535042895317e-05,
+      "loss": 0.088,
+      "step": 73765
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.3177066241620178e-05,
+      "loss": 0.0588,
+      "step": 73770
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.3176597440345038e-05,
+      "loss": 0.063,
+      "step": 73775
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.3176128639069898e-05,
+      "loss": 0.1176,
+      "step": 73780
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.3175659837794761e-05,
+      "loss": 0.1186,
+      "step": 73785
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.3175191036519621e-05,
+      "loss": 0.1963,
+      "step": 73790
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.3174722235244481e-05,
+      "loss": 0.2739,
+      "step": 73795
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.3174253433969341e-05,
+      "loss": 0.0369,
+      "step": 73800
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.3173784632694201e-05,
+      "loss": 0.015,
+      "step": 73805
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.3173315831419063e-05,
+      "loss": 0.0756,
+      "step": 73810
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.3172847030143923e-05,
+      "loss": 0.0836,
+      "step": 73815
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.3172378228868783e-05,
+      "loss": 0.0515,
+      "step": 73820
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.3171909427593643e-05,
+      "loss": 0.0864,
+      "step": 73825
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.3171440626318506e-05,
+      "loss": 0.107,
+      "step": 73830
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.3170971825043366e-05,
+      "loss": 0.223,
+      "step": 73835
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.3170503023768226e-05,
+      "loss": 0.2617,
+      "step": 73840
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.3170034222493086e-05,
+      "loss": 0.372,
+      "step": 73845
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.3169565421217947e-05,
+      "loss": 0.0329,
+      "step": 73850
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.3169096619942807e-05,
+      "loss": 0.0405,
+      "step": 73855
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.3168627818667667e-05,
+      "loss": 0.0614,
+      "step": 73860
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.3168159017392527e-05,
+      "loss": 0.0546,
+      "step": 73865
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.3167690216117389e-05,
+      "loss": 0.0568,
+      "step": 73870
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.3167221414842249e-05,
+      "loss": 0.053,
+      "step": 73875
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.316675261356711e-05,
+      "loss": 0.0676,
+      "step": 73880
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.316628381229197e-05,
+      "loss": 0.0723,
+      "step": 73885
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.3165815011016832e-05,
+      "loss": 0.1877,
+      "step": 73890
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.3165346209741692e-05,
+      "loss": 0.3457,
+      "step": 73895
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.3164877408466552e-05,
+      "loss": 0.046,
+      "step": 73900
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.3164408607191412e-05,
+      "loss": 0.009,
+      "step": 73905
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.3163939805916273e-05,
+      "loss": 0.0821,
+      "step": 73910
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.3163471004641133e-05,
+      "loss": 0.0406,
+      "step": 73915
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.3163002203365993e-05,
+      "loss": 0.0372,
+      "step": 73920
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.3162533402090855e-05,
+      "loss": 0.1434,
+      "step": 73925
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.3162064600815717e-05,
+      "loss": 0.1279,
+      "step": 73930
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.3161595799540576e-05,
+      "loss": 0.1362,
+      "step": 73935
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.3161126998265436e-05,
+      "loss": 0.177,
+      "step": 73940
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.3160658196990296e-05,
+      "loss": 0.2619,
+      "step": 73945
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.3160189395715158e-05,
+      "loss": 0.0307,
+      "step": 73950
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.3159720594440018e-05,
+      "loss": 0.0362,
+      "step": 73955
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.3159251793164878e-05,
+      "loss": 0.0463,
+      "step": 73960
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.3158782991889738e-05,
+      "loss": 0.0827,
+      "step": 73965
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.3158314190614601e-05,
+      "loss": 0.0723,
+      "step": 73970
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.3157845389339461e-05,
+      "loss": 0.1391,
+      "step": 73975
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.3157376588064321e-05,
+      "loss": 0.0436,
+      "step": 73980
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.3156907786789181e-05,
+      "loss": 0.1241,
+      "step": 73985
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.3156438985514042e-05,
+      "loss": 0.1112,
+      "step": 73990
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.3155970184238902e-05,
+      "loss": 0.3744,
+      "step": 73995
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.3155501382963762e-05,
+      "loss": 0.0241,
+      "step": 74000
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.3155032581688622e-05,
+      "loss": 0.0648,
+      "step": 74005
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.3154563780413482e-05,
+      "loss": 0.0572,
+      "step": 74010
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.3154094979138346e-05,
+      "loss": 0.0271,
+      "step": 74015
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.3153626177863205e-05,
+      "loss": 0.1576,
+      "step": 74020
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.3153157376588065e-05,
+      "loss": 0.0899,
+      "step": 74025
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.3152688575312927e-05,
+      "loss": 0.135,
+      "step": 74030
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.3152219774037787e-05,
+      "loss": 0.0804,
+      "step": 74035
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.3151750972762647e-05,
+      "loss": 0.3208,
+      "step": 74040
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.3151282171487507e-05,
+      "loss": 0.2127,
+      "step": 74045
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.3150813370212367e-05,
+      "loss": 0.0741,
+      "step": 74050
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.3150344568937228e-05,
+      "loss": 0.0224,
+      "step": 74055
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.3149875767662088e-05,
+      "loss": 0.042,
+      "step": 74060
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.314940696638695e-05,
+      "loss": 0.0661,
+      "step": 74065
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.3148938165111812e-05,
+      "loss": 0.0695,
+      "step": 74070
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.3148469363836672e-05,
+      "loss": 0.1331,
+      "step": 74075
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.3148000562561531e-05,
+      "loss": 0.1019,
+      "step": 74080
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.3147531761286391e-05,
+      "loss": 0.1089,
+      "step": 74085
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.3147062960011251e-05,
+      "loss": 0.1379,
+      "step": 74090
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.3146594158736113e-05,
+      "loss": 0.2481,
+      "step": 74095
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.3146125357460973e-05,
+      "loss": 0.1092,
+      "step": 74100
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.3145656556185833e-05,
+      "loss": 0.0486,
+      "step": 74105
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.3145187754910696e-05,
+      "loss": 0.0154,
+      "step": 74110
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.3144718953635556e-05,
+      "loss": 0.0222,
+      "step": 74115
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.3144250152360416e-05,
+      "loss": 0.0986,
+      "step": 74120
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.3143781351085276e-05,
+      "loss": 0.073,
+      "step": 74125
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.3143312549810136e-05,
+      "loss": 0.087,
+      "step": 74130
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.3142843748534998e-05,
+      "loss": 0.0813,
+      "step": 74135
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.3142374947259857e-05,
+      "loss": 0.1903,
+      "step": 74140
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.3141906145984717e-05,
+      "loss": 0.259,
+      "step": 74145
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.3141437344709577e-05,
+      "loss": 0.0568,
+      "step": 74150
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.314096854343444e-05,
+      "loss": 0.0046,
+      "step": 74155
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.31404997421593e-05,
+      "loss": 0.0352,
+      "step": 74160
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.314003094088416e-05,
+      "loss": 0.0488,
+      "step": 74165
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.313956213960902e-05,
+      "loss": 0.0836,
+      "step": 74170
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.3139093338333882e-05,
+      "loss": 0.081,
+      "step": 74175
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.3138624537058742e-05,
+      "loss": 0.1126,
+      "step": 74180
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.3138155735783602e-05,
+      "loss": 0.1061,
+      "step": 74185
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.3137686934508462e-05,
+      "loss": 0.2805,
+      "step": 74190
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.3137218133233322e-05,
+      "loss": 0.3446,
+      "step": 74195
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.3136749331958183e-05,
+      "loss": 0.0647,
+      "step": 74200
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.3136280530683045e-05,
+      "loss": 0.0504,
+      "step": 74205
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.3135811729407905e-05,
+      "loss": 0.0499,
+      "step": 74210
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.3135342928132767e-05,
+      "loss": 0.0752,
+      "step": 74215
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.3134874126857627e-05,
+      "loss": 0.0907,
+      "step": 74220
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.3134405325582486e-05,
+      "loss": 0.0531,
+      "step": 74225
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.3133936524307346e-05,
+      "loss": 0.2022,
+      "step": 74230
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.3133467723032206e-05,
+      "loss": 0.1445,
+      "step": 74235
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.3132998921757068e-05,
+      "loss": 0.2684,
+      "step": 74240
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.3132530120481928e-05,
+      "loss": 0.2548,
+      "step": 74245
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.313206131920679e-05,
+      "loss": 0.0683,
+      "step": 74250
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.3131592517931651e-05,
+      "loss": 0.0156,
+      "step": 74255
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.3131123716656511e-05,
+      "loss": 0.0489,
+      "step": 74260
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.3130654915381371e-05,
+      "loss": 0.0534,
+      "step": 74265
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.3130186114106231e-05,
+      "loss": 0.0794,
+      "step": 74270
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.3129717312831091e-05,
+      "loss": 0.0462,
+      "step": 74275
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.3129248511555953e-05,
+      "loss": 0.126,
+      "step": 74280
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.3128779710280812e-05,
+      "loss": 0.1405,
+      "step": 74285
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.3128310909005672e-05,
+      "loss": 0.211,
+      "step": 74290
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.3127842107730536e-05,
+      "loss": 0.3788,
+      "step": 74295
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.3127373306455396e-05,
+      "loss": 0.0817,
+      "step": 74300
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.3126904505180256e-05,
+      "loss": 0.0454,
+      "step": 74305
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.3126435703905116e-05,
+      "loss": 0.0408,
+      "step": 74310
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.3125966902629977e-05,
+      "loss": 0.0503,
+      "step": 74315
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.3125498101354837e-05,
+      "loss": 0.0483,
+      "step": 74320
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.3125029300079697e-05,
+      "loss": 0.2694,
+      "step": 74325
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.3124560498804557e-05,
+      "loss": 0.1335,
+      "step": 74330
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.3124091697529417e-05,
+      "loss": 0.1436,
+      "step": 74335
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.312362289625428e-05,
+      "loss": 0.2363,
+      "step": 74340
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.312315409497914e-05,
+      "loss": 0.2065,
+      "step": 74345
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.3122685293704e-05,
+      "loss": 0.0453,
+      "step": 74350
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.3122216492428862e-05,
+      "loss": 0.0222,
+      "step": 74355
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.3121747691153722e-05,
+      "loss": 0.0301,
+      "step": 74360
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.3121278889878582e-05,
+      "loss": 0.0662,
+      "step": 74365
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.3120810088603441e-05,
+      "loss": 0.0514,
+      "step": 74370
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.3120341287328301e-05,
+      "loss": 0.0178,
+      "step": 74375
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.3119872486053163e-05,
+      "loss": 0.1477,
+      "step": 74380
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.3119403684778023e-05,
+      "loss": 0.0924,
+      "step": 74385
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.3118934883502885e-05,
+      "loss": 0.1522,
+      "step": 74390
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.3118466082227746e-05,
+      "loss": 0.4497,
+      "step": 74395
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.3117997280952606e-05,
+      "loss": 0.0668,
+      "step": 74400
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.3117528479677466e-05,
+      "loss": 0.0368,
+      "step": 74405
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.3117059678402326e-05,
+      "loss": 0.0469,
+      "step": 74410
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.3116590877127186e-05,
+      "loss": 0.0359,
+      "step": 74415
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.3116122075852048e-05,
+      "loss": 0.0854,
+      "step": 74420
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.3115653274576908e-05,
+      "loss": 0.125,
+      "step": 74425
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.3115184473301767e-05,
+      "loss": 0.09,
+      "step": 74430
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.311471567202663e-05,
+      "loss": 0.1239,
+      "step": 74435
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.311424687075149e-05,
+      "loss": 0.2436,
+      "step": 74440
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.311377806947635e-05,
+      "loss": 0.3589,
+      "step": 74445
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.311330926820121e-05,
+      "loss": 0.0651,
+      "step": 74450
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.311284046692607e-05,
+      "loss": 0.0261,
+      "step": 74455
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.3112371665650932e-05,
+      "loss": 0.024,
+      "step": 74460
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.3111902864375792e-05,
+      "loss": 0.0978,
+      "step": 74465
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.3111434063100652e-05,
+      "loss": 0.112,
+      "step": 74470
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.3110965261825512e-05,
+      "loss": 0.0872,
+      "step": 74475
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.3110496460550375e-05,
+      "loss": 0.0558,
+      "step": 74480
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.3110027659275235e-05,
+      "loss": 0.1574,
+      "step": 74485
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.3109558858000095e-05,
+      "loss": 0.1574,
+      "step": 74490
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.3109090056724955e-05,
+      "loss": 0.2159,
+      "step": 74495
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.3108621255449817e-05,
+      "loss": 0.06,
+      "step": 74500
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.3108152454174677e-05,
+      "loss": 0.0309,
+      "step": 74505
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.3107683652899537e-05,
+      "loss": 0.0342,
+      "step": 74510
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.3107214851624397e-05,
+      "loss": 0.0972,
+      "step": 74515
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.3106746050349256e-05,
+      "loss": 0.0685,
+      "step": 74520
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.3106277249074118e-05,
+      "loss": 0.0872,
+      "step": 74525
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.310580844779898e-05,
+      "loss": 0.0839,
+      "step": 74530
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.310533964652384e-05,
+      "loss": 0.1234,
+      "step": 74535
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.3104870845248701e-05,
+      "loss": 0.2244,
+      "step": 74540
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.3104402043973561e-05,
+      "loss": 0.2721,
+      "step": 74545
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.3103933242698421e-05,
+      "loss": 0.0539,
+      "step": 74550
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.3103464441423281e-05,
+      "loss": 0.0219,
+      "step": 74555
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.3102995640148141e-05,
+      "loss": 0.0461,
+      "step": 74560
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.3102526838873003e-05,
+      "loss": 0.0784,
+      "step": 74565
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.3102058037597863e-05,
+      "loss": 0.0725,
+      "step": 74570
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.3101589236322724e-05,
+      "loss": 0.129,
+      "step": 74575
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.3101120435047586e-05,
+      "loss": 0.0321,
+      "step": 74580
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.3100651633772446e-05,
+      "loss": 0.0981,
+      "step": 74585
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.3100182832497306e-05,
+      "loss": 0.1701,
+      "step": 74590
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.3099714031222166e-05,
+      "loss": 0.1971,
+      "step": 74595
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.3099245229947026e-05,
+      "loss": 0.0555,
+      "step": 74600
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.3098776428671887e-05,
+      "loss": 0.0607,
+      "step": 74605
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.3098307627396747e-05,
+      "loss": 0.0498,
+      "step": 74610
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.3097838826121607e-05,
+      "loss": 0.0246,
+      "step": 74615
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.309737002484647e-05,
+      "loss": 0.0938,
+      "step": 74620
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.309690122357133e-05,
+      "loss": 0.0851,
+      "step": 74625
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.309643242229619e-05,
+      "loss": 0.1235,
+      "step": 74630
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.309596362102105e-05,
+      "loss": 0.1757,
+      "step": 74635
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.309549481974591e-05,
+      "loss": 0.2538,
+      "step": 74640
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.3095026018470772e-05,
+      "loss": 0.2441,
+      "step": 74645
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.3094557217195632e-05,
+      "loss": 0.0279,
+      "step": 74650
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.3094088415920492e-05,
+      "loss": 0.0447,
+      "step": 74655
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.3093619614645352e-05,
+      "loss": 0.011,
+      "step": 74660
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.3093150813370215e-05,
+      "loss": 0.0893,
+      "step": 74665
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.3092682012095075e-05,
+      "loss": 0.0987,
+      "step": 74670
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.3092213210819935e-05,
+      "loss": 0.1312,
+      "step": 74675
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.3091744409544795e-05,
+      "loss": 0.1104,
+      "step": 74680
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.3091275608269656e-05,
+      "loss": 0.1359,
+      "step": 74685
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.3090806806994516e-05,
+      "loss": 0.2558,
+      "step": 74690
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.3090338005719376e-05,
+      "loss": 0.2142,
+      "step": 74695
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.3089869204444236e-05,
+      "loss": 0.0664,
+      "step": 74700
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.3089400403169098e-05,
+      "loss": 0.0286,
+      "step": 74705
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.3088931601893958e-05,
+      "loss": 0.0856,
+      "step": 74710
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.308846280061882e-05,
+      "loss": 0.021,
+      "step": 74715
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.308799399934368e-05,
+      "loss": 0.0525,
+      "step": 74720
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.308752519806854e-05,
+      "loss": 0.047,
+      "step": 74725
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.30870563967934e-05,
+      "loss": 0.1208,
+      "step": 74730
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.308658759551826e-05,
+      "loss": 0.208,
+      "step": 74735
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.308611879424312e-05,
+      "loss": 0.182,
+      "step": 74740
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.3085649992967982e-05,
+      "loss": 0.2804,
+      "step": 74745
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.3085181191692842e-05,
+      "loss": 0.0816,
+      "step": 74750
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.3084712390417702e-05,
+      "loss": 0.0376,
+      "step": 74755
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.3084243589142564e-05,
+      "loss": 0.0442,
+      "step": 74760
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.3083774787867425e-05,
+      "loss": 0.0874,
+      "step": 74765
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.3083305986592285e-05,
+      "loss": 0.0598,
+      "step": 74770
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.3082837185317145e-05,
+      "loss": 0.0949,
+      "step": 74775
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.3082368384042005e-05,
+      "loss": 0.1243,
+      "step": 74780
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.3081899582766867e-05,
+      "loss": 0.1393,
+      "step": 74785
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.3081430781491727e-05,
+      "loss": 0.1766,
+      "step": 74790
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.3080961980216587e-05,
+      "loss": 0.329,
+      "step": 74795
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.3080493178941447e-05,
+      "loss": 0.0834,
+      "step": 74800
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.308002437766631e-05,
+      "loss": 0.0429,
+      "step": 74805
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.307955557639117e-05,
+      "loss": 0.0264,
+      "step": 74810
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.307908677511603e-05,
+      "loss": 0.0512,
+      "step": 74815
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.307861797384089e-05,
+      "loss": 0.1067,
+      "step": 74820
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.3078149172565751e-05,
+      "loss": 0.1192,
+      "step": 74825
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.3077680371290611e-05,
+      "loss": 0.0869,
+      "step": 74830
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.3077211570015471e-05,
+      "loss": 0.1571,
+      "step": 74835
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.3076742768740331e-05,
+      "loss": 0.214,
+      "step": 74840
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.3076273967465191e-05,
+      "loss": 0.1172,
+      "step": 74845
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.3075805166190053e-05,
+      "loss": 0.067,
+      "step": 74850
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.3075336364914914e-05,
+      "loss": 0.0654,
+      "step": 74855
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.3074867563639774e-05,
+      "loss": 0.0608,
+      "step": 74860
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.3074398762364636e-05,
+      "loss": 0.0573,
+      "step": 74865
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.3073929961089496e-05,
+      "loss": 0.0422,
+      "step": 74870
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.3073461159814356e-05,
+      "loss": 0.091,
+      "step": 74875
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.3072992358539216e-05,
+      "loss": 0.1209,
+      "step": 74880
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.3072523557264076e-05,
+      "loss": 0.1703,
+      "step": 74885
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.3072054755988937e-05,
+      "loss": 0.0818,
+      "step": 74890
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.3071585954713797e-05,
+      "loss": 0.2644,
+      "step": 74895
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.3071117153438659e-05,
+      "loss": 0.0688,
+      "step": 74900
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.307064835216352e-05,
+      "loss": 0.0201,
+      "step": 74905
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.307017955088838e-05,
+      "loss": 0.0338,
+      "step": 74910
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.306971074961324e-05,
+      "loss": 0.1382,
+      "step": 74915
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.30692419483381e-05,
+      "loss": 0.1653,
+      "step": 74920
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.306877314706296e-05,
+      "loss": 0.0666,
+      "step": 74925
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.3068304345787822e-05,
+      "loss": 0.0343,
+      "step": 74930
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.3067835544512682e-05,
+      "loss": 0.1865,
+      "step": 74935
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.3067366743237542e-05,
+      "loss": 0.2692,
+      "step": 74940
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.3066897941962405e-05,
+      "loss": 0.3542,
+      "step": 74945
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.3066429140687265e-05,
+      "loss": 0.0865,
+      "step": 74950
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.3065960339412125e-05,
+      "loss": 0.0346,
+      "step": 74955
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.3065491538136985e-05,
+      "loss": 0.0699,
+      "step": 74960
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.3065022736861845e-05,
+      "loss": 0.0342,
+      "step": 74965
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.3064553935586706e-05,
+      "loss": 0.1412,
+      "step": 74970
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.3064085134311566e-05,
+      "loss": 0.0652,
+      "step": 74975
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.3063616333036426e-05,
+      "loss": 0.1106,
+      "step": 74980
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.3063147531761286e-05,
+      "loss": 0.1842,
+      "step": 74985
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.306267873048615e-05,
+      "loss": 0.2778,
+      "step": 74990
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.306220992921101e-05,
+      "loss": 0.2805,
+      "step": 74995
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.306174112793587e-05,
+      "loss": 0.0736,
+      "step": 75000
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.306127232666073e-05,
+      "loss": 0.0197,
+      "step": 75005
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.3060803525385591e-05,
+      "loss": 0.0479,
+      "step": 75010
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.3060334724110451e-05,
+      "loss": 0.0528,
+      "step": 75015
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.305986592283531e-05,
+      "loss": 0.065,
+      "step": 75020
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.305939712156017e-05,
+      "loss": 0.1103,
+      "step": 75025
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.305892832028503e-05,
+      "loss": 0.0833,
+      "step": 75030
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.3058459519009892e-05,
+      "loss": 0.0872,
+      "step": 75035
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.3057990717734754e-05,
+      "loss": 0.1221,
+      "step": 75040
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.3057521916459614e-05,
+      "loss": 0.2611,
+      "step": 75045
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.3057053115184475e-05,
+      "loss": 0.0998,
+      "step": 75050
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.3056584313909335e-05,
+      "loss": 0.031,
+      "step": 75055
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.3056115512634195e-05,
+      "loss": 0.0343,
+      "step": 75060
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.3055646711359055e-05,
+      "loss": 0.037,
+      "step": 75065
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.3055177910083915e-05,
+      "loss": 0.186,
+      "step": 75070
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.3054709108808777e-05,
+      "loss": 0.0453,
+      "step": 75075
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.3054240307533637e-05,
+      "loss": 0.1268,
+      "step": 75080
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.3053771506258498e-05,
+      "loss": 0.1189,
+      "step": 75085
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.305330270498336e-05,
+      "loss": 0.1851,
+      "step": 75090
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.305283390370822e-05,
+      "loss": 0.3663,
+      "step": 75095
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.305236510243308e-05,
+      "loss": 0.087,
+      "step": 75100
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.305189630115794e-05,
+      "loss": 0.015,
+      "step": 75105
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.30514274998828e-05,
+      "loss": 0.0165,
+      "step": 75110
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.3050958698607661e-05,
+      "loss": 0.0363,
+      "step": 75115
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.3050489897332521e-05,
+      "loss": 0.0847,
+      "step": 75120
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.3050021096057381e-05,
+      "loss": 0.0863,
+      "step": 75125
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.3049552294782245e-05,
+      "loss": 0.0849,
+      "step": 75130
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.3049083493507105e-05,
+      "loss": 0.1342,
+      "step": 75135
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.3048614692231964e-05,
+      "loss": 0.0638,
+      "step": 75140
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.3048145890956824e-05,
+      "loss": 0.181,
+      "step": 75145
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.3047677089681684e-05,
+      "loss": 0.1007,
+      "step": 75150
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.3047208288406546e-05,
+      "loss": 0.0194,
+      "step": 75155
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.3046739487131406e-05,
+      "loss": 0.0726,
+      "step": 75160
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.3046270685856266e-05,
+      "loss": 0.0531,
+      "step": 75165
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.3045801884581126e-05,
+      "loss": 0.0701,
+      "step": 75170
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.3045333083305987e-05,
+      "loss": 0.1159,
+      "step": 75175
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.3044864282030849e-05,
+      "loss": 0.0942,
+      "step": 75180
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.3044395480755709e-05,
+      "loss": 0.1559,
+      "step": 75185
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.3043926679480569e-05,
+      "loss": 0.1268,
+      "step": 75190
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.304345787820543e-05,
+      "loss": 0.2342,
+      "step": 75195
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.304298907693029e-05,
+      "loss": 0.1179,
+      "step": 75200
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.304252027565515e-05,
+      "loss": 0.0162,
+      "step": 75205
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.304205147438001e-05,
+      "loss": 0.0417,
+      "step": 75210
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.3041582673104872e-05,
+      "loss": 0.0826,
+      "step": 75215
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.3041113871829732e-05,
+      "loss": 0.0919,
+      "step": 75220
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.3040645070554593e-05,
+      "loss": 0.0402,
+      "step": 75225
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.3040176269279453e-05,
+      "loss": 0.1075,
+      "step": 75230
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.3039707468004315e-05,
+      "loss": 0.1924,
+      "step": 75235
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.3039238666729175e-05,
+      "loss": 0.2085,
+      "step": 75240
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.3038769865454035e-05,
+      "loss": 0.2829,
+      "step": 75245
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.3038301064178895e-05,
+      "loss": 0.0622,
+      "step": 75250
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.3037832262903756e-05,
+      "loss": 0.0114,
+      "step": 75255
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.3037363461628616e-05,
+      "loss": 0.0967,
+      "step": 75260
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.3036894660353476e-05,
+      "loss": 0.0325,
+      "step": 75265
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.3036425859078338e-05,
+      "loss": 0.1324,
+      "step": 75270
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.30359570578032e-05,
+      "loss": 0.0891,
+      "step": 75275
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.303548825652806e-05,
+      "loss": 0.0565,
+      "step": 75280
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.303501945525292e-05,
+      "loss": 0.1619,
+      "step": 75285
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.303455065397778e-05,
+      "loss": 0.2103,
+      "step": 75290
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.3034081852702641e-05,
+      "loss": 0.2858,
+      "step": 75295
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.3033613051427501e-05,
+      "loss": 0.0829,
+      "step": 75300
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.3033144250152361e-05,
+      "loss": 0.0536,
+      "step": 75305
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.303267544887722e-05,
+      "loss": 0.0318,
+      "step": 75310
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.3032206647602084e-05,
+      "loss": 0.0717,
+      "step": 75315
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.3031737846326944e-05,
+      "loss": 0.0579,
+      "step": 75320
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.3031269045051804e-05,
+      "loss": 0.0892,
+      "step": 75325
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.3030800243776664e-05,
+      "loss": 0.072,
+      "step": 75330
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.3030331442501526e-05,
+      "loss": 0.1365,
+      "step": 75335
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.3029862641226386e-05,
+      "loss": 0.1741,
+      "step": 75340
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.3029393839951245e-05,
+      "loss": 0.32,
+      "step": 75345
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.3028925038676105e-05,
+      "loss": 0.0796,
+      "step": 75350
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.3028456237400965e-05,
+      "loss": 0.0354,
+      "step": 75355
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.3027987436125827e-05,
+      "loss": 0.042,
+      "step": 75360
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.3027518634850689e-05,
+      "loss": 0.0825,
+      "step": 75365
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.3027049833575549e-05,
+      "loss": 0.1049,
+      "step": 75370
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.302658103230041e-05,
+      "loss": 0.0567,
+      "step": 75375
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.302611223102527e-05,
+      "loss": 0.134,
+      "step": 75380
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.302564342975013e-05,
+      "loss": 0.1164,
+      "step": 75385
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.302517462847499e-05,
+      "loss": 0.1301,
+      "step": 75390
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.302470582719985e-05,
+      "loss": 0.2434,
+      "step": 75395
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.3024237025924711e-05,
+      "loss": 0.0633,
+      "step": 75400
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.3023768224649571e-05,
+      "loss": 0.0375,
+      "step": 75405
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.3023299423374433e-05,
+      "loss": 0.1196,
+      "step": 75410
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.3022830622099295e-05,
+      "loss": 0.0867,
+      "step": 75415
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.3022361820824155e-05,
+      "loss": 0.057,
+      "step": 75420
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.3021893019549015e-05,
+      "loss": 0.0412,
+      "step": 75425
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.3021424218273874e-05,
+      "loss": 0.1075,
+      "step": 75430
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.3020955416998734e-05,
+      "loss": 0.063,
+      "step": 75435
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.3020486615723596e-05,
+      "loss": 0.1088,
+      "step": 75440
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.3020017814448456e-05,
+      "loss": 0.2778,
+      "step": 75445
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.3019549013173316e-05,
+      "loss": 0.0748,
+      "step": 75450
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.301908021189818e-05,
+      "loss": 0.0526,
+      "step": 75455
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.301861141062304e-05,
+      "loss": 0.0055,
+      "step": 75460
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.3018142609347899e-05,
+      "loss": 0.028,
+      "step": 75465
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.3017673808072759e-05,
+      "loss": 0.0265,
+      "step": 75470
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.3017205006797619e-05,
+      "loss": 0.0494,
+      "step": 75475
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.301673620552248e-05,
+      "loss": 0.0785,
+      "step": 75480
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.301626740424734e-05,
+      "loss": 0.0594,
+      "step": 75485
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.30157986029722e-05,
+      "loss": 0.2856,
+      "step": 75490
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.301532980169706e-05,
+      "loss": 0.279,
+      "step": 75495
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.301486100042192e-05,
+      "loss": 0.0904,
+      "step": 75500
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.3014392199146784e-05,
+      "loss": 0.0202,
+      "step": 75505
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.3013923397871644e-05,
+      "loss": 0.096,
+      "step": 75510
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.3013454596596504e-05,
+      "loss": 0.0634,
+      "step": 75515
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.3012985795321365e-05,
+      "loss": 0.0727,
+      "step": 75520
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.3012516994046225e-05,
+      "loss": 0.1254,
+      "step": 75525
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.3012048192771085e-05,
+      "loss": 0.0975,
+      "step": 75530
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.3011579391495945e-05,
+      "loss": 0.0998,
+      "step": 75535
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.3011110590220805e-05,
+      "loss": 0.1179,
+      "step": 75540
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.3010641788945667e-05,
+      "loss": 0.2451,
+      "step": 75545
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.3010172987670528e-05,
+      "loss": 0.0274,
+      "step": 75550
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.3009704186395388e-05,
+      "loss": 0.0173,
+      "step": 75555
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.300923538512025e-05,
+      "loss": 0.0474,
+      "step": 75560
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.300876658384511e-05,
+      "loss": 0.0392,
+      "step": 75565
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.300829778256997e-05,
+      "loss": 0.0561,
+      "step": 75570
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.300782898129483e-05,
+      "loss": 0.0775,
+      "step": 75575
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.300736018001969e-05,
+      "loss": 0.1528,
+      "step": 75580
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.3006891378744551e-05,
+      "loss": 0.0843,
+      "step": 75585
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.3006422577469411e-05,
+      "loss": 0.2861,
+      "step": 75590
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.3005953776194273e-05,
+      "loss": 0.2198,
+      "step": 75595
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.3005484974919134e-05,
+      "loss": 0.0706,
+      "step": 75600
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.3005016173643994e-05,
+      "loss": 0.0442,
+      "step": 75605
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.3004547372368854e-05,
+      "loss": 0.0733,
+      "step": 75610
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.3004078571093714e-05,
+      "loss": 0.0251,
+      "step": 75615
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.3003609769818574e-05,
+      "loss": 0.0101,
+      "step": 75620
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.3003140968543436e-05,
+      "loss": 0.0295,
+      "step": 75625
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.3002672167268296e-05,
+      "loss": 0.0772,
+      "step": 75630
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.3002203365993155e-05,
+      "loss": 0.1783,
+      "step": 75635
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.3001734564718019e-05,
+      "loss": 0.296,
+      "step": 75640
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.3001265763442879e-05,
+      "loss": 0.1765,
+      "step": 75645
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.3000796962167739e-05,
+      "loss": 0.0566,
+      "step": 75650
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.3000328160892599e-05,
+      "loss": 0.0553,
+      "step": 75655
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.2999859359617459e-05,
+      "loss": 0.0437,
+      "step": 75660
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.299939055834232e-05,
+      "loss": 0.0497,
+      "step": 75665
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.299892175706718e-05,
+      "loss": 0.0262,
+      "step": 75670
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.299845295579204e-05,
+      "loss": 0.0857,
+      "step": 75675
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.29979841545169e-05,
+      "loss": 0.0881,
+      "step": 75680
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.2997515353241762e-05,
+      "loss": 0.1327,
+      "step": 75685
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.2997046551966623e-05,
+      "loss": 0.2564,
+      "step": 75690
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.2996577750691483e-05,
+      "loss": 0.2501,
+      "step": 75695
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.2996108949416343e-05,
+      "loss": 0.0766,
+      "step": 75700
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.2995640148141205e-05,
+      "loss": 0.034,
+      "step": 75705
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.2995171346866065e-05,
+      "loss": 0.0137,
+      "step": 75710
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.2994702545590925e-05,
+      "loss": 0.0617,
+      "step": 75715
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.2994233744315785e-05,
+      "loss": 0.1114,
+      "step": 75720
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.2993764943040646e-05,
+      "loss": 0.0929,
+      "step": 75725
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.2993296141765506e-05,
+      "loss": 0.1268,
+      "step": 75730
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.2992827340490368e-05,
+      "loss": 0.1356,
+      "step": 75735
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.299235853921523e-05,
+      "loss": 0.1289,
+      "step": 75740
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.299188973794009e-05,
+      "loss": 0.2894,
+      "step": 75745
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.299142093666495e-05,
+      "loss": 0.1238,
+      "step": 75750
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.2990952135389809e-05,
+      "loss": 0.0415,
+      "step": 75755
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.2990483334114669e-05,
+      "loss": 0.0233,
+      "step": 75760
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.299001453283953e-05,
+      "loss": 0.0836,
+      "step": 75765
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.298954573156439e-05,
+      "loss": 0.0405,
+      "step": 75770
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.298907693028925e-05,
+      "loss": 0.0434,
+      "step": 75775
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.2988608129014114e-05,
+      "loss": 0.0612,
+      "step": 75780
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.2988139327738974e-05,
+      "loss": 0.1175,
+      "step": 75785
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.2987670526463834e-05,
+      "loss": 0.1582,
+      "step": 75790
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.2987201725188694e-05,
+      "loss": 0.2338,
+      "step": 75795
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.2986732923913554e-05,
+      "loss": 0.0745,
+      "step": 75800
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.2986264122638415e-05,
+      "loss": 0.0183,
+      "step": 75805
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.2985795321363275e-05,
+      "loss": 0.0273,
+      "step": 75810
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.2985326520088135e-05,
+      "loss": 0.0882,
+      "step": 75815
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.2984857718812995e-05,
+      "loss": 0.1038,
+      "step": 75820
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.2984388917537855e-05,
+      "loss": 0.077,
+      "step": 75825
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.2983920116262718e-05,
+      "loss": 0.1234,
+      "step": 75830
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.2983451314987578e-05,
+      "loss": 0.0779,
+      "step": 75835
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.2982982513712438e-05,
+      "loss": 0.1526,
+      "step": 75840
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.29825137124373e-05,
+      "loss": 0.2533,
+      "step": 75845
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.298204491116216e-05,
+      "loss": 0.033,
+      "step": 75850
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.298157610988702e-05,
+      "loss": 0.062,
+      "step": 75855
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.298110730861188e-05,
+      "loss": 0.0556,
+      "step": 75860
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.298063850733674e-05,
+      "loss": 0.076,
+      "step": 75865
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.2980169706061601e-05,
+      "loss": 0.0307,
+      "step": 75870
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.2979700904786463e-05,
+      "loss": 0.094,
+      "step": 75875
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.2979232103511323e-05,
+      "loss": 0.1644,
+      "step": 75880
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.2978763302236184e-05,
+      "loss": 0.159,
+      "step": 75885
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.2978294500961044e-05,
+      "loss": 0.1273,
+      "step": 75890
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.2977825699685904e-05,
+      "loss": 0.2573,
+      "step": 75895
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.2977356898410764e-05,
+      "loss": 0.0301,
+      "step": 75900
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.2976888097135624e-05,
+      "loss": 0.0274,
+      "step": 75905
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.2976419295860486e-05,
+      "loss": 0.0311,
+      "step": 75910
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.2975950494585346e-05,
+      "loss": 0.062,
+      "step": 75915
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.2975481693310207e-05,
+      "loss": 0.0581,
+      "step": 75920
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.2975012892035069e-05,
+      "loss": 0.0784,
+      "step": 75925
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.2974544090759929e-05,
+      "loss": 0.1502,
+      "step": 75930
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.2974075289484789e-05,
+      "loss": 0.1541,
+      "step": 75935
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.2973606488209649e-05,
+      "loss": 0.2041,
+      "step": 75940
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.2973137686934509e-05,
+      "loss": 0.3387,
+      "step": 75945
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.297266888565937e-05,
+      "loss": 0.0406,
+      "step": 75950
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.297220008438423e-05,
+      "loss": 0.0361,
+      "step": 75955
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.297173128310909e-05,
+      "loss": 0.0568,
+      "step": 75960
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.2971262481833953e-05,
+      "loss": 0.1504,
+      "step": 75965
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.2970793680558813e-05,
+      "loss": 0.1142,
+      "step": 75970
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.2970324879283673e-05,
+      "loss": 0.1309,
+      "step": 75975
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.2969856078008533e-05,
+      "loss": 0.1008,
+      "step": 75980
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.2969387276733393e-05,
+      "loss": 0.1449,
+      "step": 75985
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.2968918475458255e-05,
+      "loss": 0.2658,
+      "step": 75990
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.2968449674183115e-05,
+      "loss": 0.1993,
+      "step": 75995
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.2967980872907975e-05,
+      "loss": 0.1139,
+      "step": 76000
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.2967512071632835e-05,
+      "loss": 0.0226,
+      "step": 76005
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.2967043270357695e-05,
+      "loss": 0.0666,
+      "step": 76010
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.2966574469082558e-05,
+      "loss": 0.0206,
+      "step": 76015
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.2966105667807418e-05,
+      "loss": 0.08,
+      "step": 76020
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.2965636866532278e-05,
+      "loss": 0.1427,
+      "step": 76025
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.296516806525714e-05,
+      "loss": 0.1729,
+      "step": 76030
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.2964699263982e-05,
+      "loss": 0.1243,
+      "step": 76035
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.296423046270686e-05,
+      "loss": 0.2969,
+      "step": 76040
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.296376166143172e-05,
+      "loss": 0.3342,
+      "step": 76045
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.296329286015658e-05,
+      "loss": 0.0422,
+      "step": 76050
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.296282405888144e-05,
+      "loss": 0.0714,
+      "step": 76055
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.2962355257606302e-05,
+      "loss": 0.0715,
+      "step": 76060
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.2961886456331162e-05,
+      "loss": 0.0332,
+      "step": 76065
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.2961417655056024e-05,
+      "loss": 0.0604,
+      "step": 76070
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.2960948853780884e-05,
+      "loss": 0.0826,
+      "step": 76075
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.2960480052505744e-05,
+      "loss": 0.0887,
+      "step": 76080
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.2960011251230604e-05,
+      "loss": 0.1112,
+      "step": 76085
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.2959542449955465e-05,
+      "loss": 0.1512,
+      "step": 76090
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.2959073648680325e-05,
+      "loss": 0.221,
+      "step": 76095
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.2958604847405185e-05,
+      "loss": 0.0429,
+      "step": 76100
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.2958136046130047e-05,
+      "loss": 0.0337,
+      "step": 76105
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.2957667244854908e-05,
+      "loss": 0.0281,
+      "step": 76110
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.2957198443579768e-05,
+      "loss": 0.068,
+      "step": 76115
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.2956729642304628e-05,
+      "loss": 0.0651,
+      "step": 76120
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.2956260841029488e-05,
+      "loss": 0.0569,
+      "step": 76125
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.295579203975435e-05,
+      "loss": 0.1717,
+      "step": 76130
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.295532323847921e-05,
+      "loss": 0.1179,
+      "step": 76135
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.295485443720407e-05,
+      "loss": 0.27,
+      "step": 76140
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.295438563592893e-05,
+      "loss": 0.2105,
+      "step": 76145
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.295391683465379e-05,
+      "loss": 0.0964,
+      "step": 76150
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.2953448033378653e-05,
+      "loss": 0.0165,
+      "step": 76155
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.2952979232103513e-05,
+      "loss": 0.0608,
+      "step": 76160
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.2952510430828373e-05,
+      "loss": 0.0416,
+      "step": 76165
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.2952041629553234e-05,
+      "loss": 0.0805,
+      "step": 76170
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.2951572828278094e-05,
+      "loss": 0.0718,
+      "step": 76175
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.2951104027002954e-05,
+      "loss": 0.1264,
+      "step": 76180
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.2950635225727814e-05,
+      "loss": 0.1554,
+      "step": 76185
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.2950166424452674e-05,
+      "loss": 0.262,
+      "step": 76190
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.2949697623177536e-05,
+      "loss": 0.3103,
+      "step": 76195
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.2949228821902397e-05,
+      "loss": 0.0947,
+      "step": 76200
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.2948760020627257e-05,
+      "loss": 0.042,
+      "step": 76205
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.2948291219352119e-05,
+      "loss": 0.0349,
+      "step": 76210
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.2947822418076979e-05,
+      "loss": 0.0472,
+      "step": 76215
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.2947353616801839e-05,
+      "loss": 0.0736,
+      "step": 76220
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.2946884815526699e-05,
+      "loss": 0.1034,
+      "step": 76225
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.2946416014251559e-05,
+      "loss": 0.1762,
+      "step": 76230
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.294594721297642e-05,
+      "loss": 0.1403,
+      "step": 76235
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.294547841170128e-05,
+      "loss": 0.1301,
+      "step": 76240
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.2945009610426142e-05,
+      "loss": 0.3283,
+      "step": 76245
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.2944540809151004e-05,
+      "loss": 0.0479,
+      "step": 76250
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.2944072007875863e-05,
+      "loss": 0.014,
+      "step": 76255
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.2943603206600723e-05,
+      "loss": 0.0383,
+      "step": 76260
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.2943134405325583e-05,
+      "loss": 0.0522,
+      "step": 76265
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.2942665604050443e-05,
+      "loss": 0.1356,
+      "step": 76270
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.2942196802775305e-05,
+      "loss": 0.0687,
+      "step": 76275
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.2941728001500165e-05,
+      "loss": 0.0764,
+      "step": 76280
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.2941259200225025e-05,
+      "loss": 0.1817,
+      "step": 76285
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.2940790398949888e-05,
+      "loss": 0.2576,
+      "step": 76290
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.2940321597674748e-05,
+      "loss": 0.2009,
+      "step": 76295
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.2939852796399608e-05,
+      "loss": 0.0719,
+      "step": 76300
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.2939383995124468e-05,
+      "loss": 0.0553,
+      "step": 76305
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.2938915193849328e-05,
+      "loss": 0.0314,
+      "step": 76310
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.293844639257419e-05,
+      "loss": 0.0829,
+      "step": 76315
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.293797759129905e-05,
+      "loss": 0.121,
+      "step": 76320
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.293750879002391e-05,
+      "loss": 0.071,
+      "step": 76325
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.293703998874877e-05,
+      "loss": 0.0498,
+      "step": 76330
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.293657118747363e-05,
+      "loss": 0.0843,
+      "step": 76335
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.2936102386198493e-05,
+      "loss": 0.1836,
+      "step": 76340
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.2935633584923352e-05,
+      "loss": 0.2321,
+      "step": 76345
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.2935164783648212e-05,
+      "loss": 0.0305,
+      "step": 76350
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.2934695982373074e-05,
+      "loss": 0.0301,
+      "step": 76355
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.2934227181097934e-05,
+      "loss": 0.0454,
+      "step": 76360
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.2933758379822794e-05,
+      "loss": 0.0474,
+      "step": 76365
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.2933289578547654e-05,
+      "loss": 0.0194,
+      "step": 76370
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.2932820777272514e-05,
+      "loss": 0.0325,
+      "step": 76375
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.2932351975997375e-05,
+      "loss": 0.074,
+      "step": 76380
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.2931883174722237e-05,
+      "loss": 0.134,
+      "step": 76385
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.2931414373447097e-05,
+      "loss": 0.1937,
+      "step": 76390
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.2930945572171959e-05,
+      "loss": 0.2208,
+      "step": 76395
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.2930476770896819e-05,
+      "loss": 0.073,
+      "step": 76400
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.2930007969621678e-05,
+      "loss": 0.0244,
+      "step": 76405
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.2929539168346538e-05,
+      "loss": 0.0703,
+      "step": 76410
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.2929070367071398e-05,
+      "loss": 0.0698,
+      "step": 76415
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.292860156579626e-05,
+      "loss": 0.0817,
+      "step": 76420
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.292813276452112e-05,
+      "loss": 0.0615,
+      "step": 76425
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.2927663963245981e-05,
+      "loss": 0.0691,
+      "step": 76430
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.2927195161970843e-05,
+      "loss": 0.0809,
+      "step": 76435
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.2926726360695703e-05,
+      "loss": 0.2511,
+      "step": 76440
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.2926257559420563e-05,
+      "loss": 0.2983,
+      "step": 76445
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.2925788758145423e-05,
+      "loss": 0.073,
+      "step": 76450
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.2925319956870283e-05,
+      "loss": 0.0145,
+      "step": 76455
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.2924851155595144e-05,
+      "loss": 0.0306,
+      "step": 76460
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.2924382354320004e-05,
+      "loss": 0.0594,
+      "step": 76465
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.2923913553044864e-05,
+      "loss": 0.056,
+      "step": 76470
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.2923444751769724e-05,
+      "loss": 0.0702,
+      "step": 76475
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.2922975950494588e-05,
+      "loss": 0.039,
+      "step": 76480
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.2922507149219448e-05,
+      "loss": 0.1921,
+      "step": 76485
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.2922038347944307e-05,
+      "loss": 0.1554,
+      "step": 76490
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.2921569546669167e-05,
+      "loss": 0.2371,
+      "step": 76495
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.2921100745394029e-05,
+      "loss": 0.0921,
+      "step": 76500
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.2920631944118889e-05,
+      "loss": 0.0387,
+      "step": 76505
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.2920163142843749e-05,
+      "loss": 0.0425,
+      "step": 76510
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.2919694341568609e-05,
+      "loss": 0.0556,
+      "step": 76515
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.291922554029347e-05,
+      "loss": 0.0308,
+      "step": 76520
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.2918756739018332e-05,
+      "loss": 0.0584,
+      "step": 76525
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.2918287937743192e-05,
+      "loss": 0.1801,
+      "step": 76530
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.2917819136468052e-05,
+      "loss": 0.0882,
+      "step": 76535
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.2917350335192914e-05,
+      "loss": 0.1778,
+      "step": 76540
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.2916881533917774e-05,
+      "loss": 0.3989,
+      "step": 76545
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.2916412732642633e-05,
+      "loss": 0.0859,
+      "step": 76550
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.2915943931367493e-05,
+      "loss": 0.0368,
+      "step": 76555
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.2915475130092355e-05,
+      "loss": 0.0566,
+      "step": 76560
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.2915006328817215e-05,
+      "loss": 0.081,
+      "step": 76565
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.2914537527542077e-05,
+      "loss": 0.0375,
+      "step": 76570
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.2914068726266937e-05,
+      "loss": 0.0694,
+      "step": 76575
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.2913599924991798e-05,
+      "loss": 0.1095,
+      "step": 76580
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.2913131123716658e-05,
+      "loss": 0.1153,
+      "step": 76585
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.2912662322441518e-05,
+      "loss": 0.1803,
+      "step": 76590
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.2912193521166378e-05,
+      "loss": 0.2579,
+      "step": 76595
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.291172471989124e-05,
+      "loss": 0.1152,
+      "step": 76600
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.29112559186161e-05,
+      "loss": 0.0289,
+      "step": 76605
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.291078711734096e-05,
+      "loss": 0.0535,
+      "step": 76610
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.291031831606582e-05,
+      "loss": 0.1614,
+      "step": 76615
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.2909849514790683e-05,
+      "loss": 0.0627,
+      "step": 76620
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.2909380713515543e-05,
+      "loss": 0.0397,
+      "step": 76625
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.2908911912240403e-05,
+      "loss": 0.1691,
+      "step": 76630
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.2908443110965262e-05,
+      "loss": 0.0838,
+      "step": 76635
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.2907974309690124e-05,
+      "loss": 0.1414,
+      "step": 76640
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.2907505508414984e-05,
+      "loss": 0.2367,
+      "step": 76645
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.2907036707139844e-05,
+      "loss": 0.1015,
+      "step": 76650
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.2906567905864704e-05,
+      "loss": 0.0206,
+      "step": 76655
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.2906099104589564e-05,
+      "loss": 0.0277,
+      "step": 76660
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.2905630303314427e-05,
+      "loss": 0.059,
+      "step": 76665
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.2905161502039287e-05,
+      "loss": 0.0672,
+      "step": 76670
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.2904692700764147e-05,
+      "loss": 0.0522,
+      "step": 76675
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.2904223899489009e-05,
+      "loss": 0.1238,
+      "step": 76680
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.2903755098213869e-05,
+      "loss": 0.105,
+      "step": 76685
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.2903286296938729e-05,
+      "loss": 0.299,
+      "step": 76690
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.2902817495663588e-05,
+      "loss": 0.1564,
+      "step": 76695
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.2902348694388448e-05,
+      "loss": 0.0793,
+      "step": 76700
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.290187989311331e-05,
+      "loss": 0.0337,
+      "step": 76705
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.2901411091838172e-05,
+      "loss": 0.0111,
+      "step": 76710
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.2900942290563032e-05,
+      "loss": 0.0667,
+      "step": 76715
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.2900473489287893e-05,
+      "loss": 0.0295,
+      "step": 76720
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.2900004688012753e-05,
+      "loss": 0.0884,
+      "step": 76725
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.2899535886737613e-05,
+      "loss": 0.0905,
+      "step": 76730
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.2899067085462473e-05,
+      "loss": 0.0383,
+      "step": 76735
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.2898598284187333e-05,
+      "loss": 0.1757,
+      "step": 76740
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.2898129482912195e-05,
+      "loss": 0.1919,
+      "step": 76745
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.2897660681637055e-05,
+      "loss": 0.0592,
+      "step": 76750
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.2897191880361916e-05,
+      "loss": 0.0298,
+      "step": 76755
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.2896723079086778e-05,
+      "loss": 0.0382,
+      "step": 76760
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.2896254277811638e-05,
+      "loss": 0.028,
+      "step": 76765
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.2895785476536498e-05,
+      "loss": 0.0667,
+      "step": 76770
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.2895316675261358e-05,
+      "loss": 0.101,
+      "step": 76775
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.2894847873986218e-05,
+      "loss": 0.0995,
+      "step": 76780
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.2894379072711079e-05,
+      "loss": 0.1493,
+      "step": 76785
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.2893910271435939e-05,
+      "loss": 0.1684,
+      "step": 76790
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.2893441470160799e-05,
+      "loss": 0.2884,
+      "step": 76795
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.2892972668885659e-05,
+      "loss": 0.0547,
+      "step": 76800
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.2892503867610522e-05,
+      "loss": 0.0472,
+      "step": 76805
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.2892035066335382e-05,
+      "loss": 0.0213,
+      "step": 76810
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.2891566265060242e-05,
+      "loss": 0.0917,
+      "step": 76815
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.2891097463785102e-05,
+      "loss": 0.0812,
+      "step": 76820
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.2890628662509964e-05,
+      "loss": 0.1001,
+      "step": 76825
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.2890159861234824e-05,
+      "loss": 0.0974,
+      "step": 76830
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.2889691059959684e-05,
+      "loss": 0.1771,
+      "step": 76835
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.2889222258684543e-05,
+      "loss": 0.3194,
+      "step": 76840
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.2888753457409403e-05,
+      "loss": 0.1879,
+      "step": 76845
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.2888284656134267e-05,
+      "loss": 0.0425,
+      "step": 76850
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.2887815854859127e-05,
+      "loss": 0.022,
+      "step": 76855
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.2887347053583987e-05,
+      "loss": 0.037,
+      "step": 76860
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.2886878252308848e-05,
+      "loss": 0.0621,
+      "step": 76865
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.2886409451033708e-05,
+      "loss": 0.1295,
+      "step": 76870
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.2885940649758568e-05,
+      "loss": 0.0795,
+      "step": 76875
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.2885471848483428e-05,
+      "loss": 0.1052,
+      "step": 76880
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.2885003047208288e-05,
+      "loss": 0.1687,
+      "step": 76885
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.288453424593315e-05,
+      "loss": 0.2261,
+      "step": 76890
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.2884065444658011e-05,
+      "loss": 0.2531,
+      "step": 76895
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.2883596643382871e-05,
+      "loss": 0.0715,
+      "step": 76900
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.2883127842107733e-05,
+      "loss": 0.0353,
+      "step": 76905
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.2882659040832593e-05,
+      "loss": 0.0076,
+      "step": 76910
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.2882190239557453e-05,
+      "loss": 0.0836,
+      "step": 76915
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.2881721438282313e-05,
+      "loss": 0.0912,
+      "step": 76920
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.2881252637007173e-05,
+      "loss": 0.1204,
+      "step": 76925
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.2880783835732034e-05,
+      "loss": 0.1609,
+      "step": 76930
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.2880315034456894e-05,
+      "loss": 0.0685,
+      "step": 76935
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.2879846233181754e-05,
+      "loss": 0.0957,
+      "step": 76940
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.2879377431906617e-05,
+      "loss": 0.3234,
+      "step": 76945
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.2878908630631477e-05,
+      "loss": 0.0517,
+      "step": 76950
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.2878439829356337e-05,
+      "loss": 0.0045,
+      "step": 76955
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.2877971028081197e-05,
+      "loss": 0.0288,
+      "step": 76960
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.2877502226806057e-05,
+      "loss": 0.0282,
+      "step": 76965
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.2877033425530919e-05,
+      "loss": 0.0993,
+      "step": 76970
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.2876564624255779e-05,
+      "loss": 0.0922,
+      "step": 76975
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.2876095822980639e-05,
+      "loss": 0.0956,
+      "step": 76980
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.2875627021705498e-05,
+      "loss": 0.17,
+      "step": 76985
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.2875158220430362e-05,
+      "loss": 0.2218,
+      "step": 76990
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.2874689419155222e-05,
+      "loss": 0.2137,
+      "step": 76995
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.2874220617880082e-05,
+      "loss": 0.0566,
+      "step": 77000
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.2873751816604942e-05,
+      "loss": 0.052,
+      "step": 77005
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.2873283015329803e-05,
+      "loss": 0.0441,
+      "step": 77010
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.2872814214054663e-05,
+      "loss": 0.0196,
+      "step": 77015
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.2872345412779523e-05,
+      "loss": 0.0477,
+      "step": 77020
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.2871876611504383e-05,
+      "loss": 0.1028,
+      "step": 77025
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.2871407810229245e-05,
+      "loss": 0.1279,
+      "step": 77030
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.2870939008954106e-05,
+      "loss": 0.0714,
+      "step": 77035
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.2870470207678966e-05,
+      "loss": 0.1664,
+      "step": 77040
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.2870001406403826e-05,
+      "loss": 0.1847,
+      "step": 77045
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.2869532605128688e-05,
+      "loss": 0.0399,
+      "step": 77050
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.2869063803853548e-05,
+      "loss": 0.0232,
+      "step": 77055
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.2868595002578408e-05,
+      "loss": 0.0392,
+      "step": 77060
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.2868126201303268e-05,
+      "loss": 0.028,
+      "step": 77065
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.286765740002813e-05,
+      "loss": 0.1343,
+      "step": 77070
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.286718859875299e-05,
+      "loss": 0.0538,
+      "step": 77075
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.286671979747785e-05,
+      "loss": 0.1227,
+      "step": 77080
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.286625099620271e-05,
+      "loss": 0.1368,
+      "step": 77085
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.2865782194927572e-05,
+      "loss": 0.151,
+      "step": 77090
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.2865313393652432e-05,
+      "loss": 0.3466,
+      "step": 77095
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.2864844592377292e-05,
+      "loss": 0.0476,
+      "step": 77100
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.2864375791102152e-05,
+      "loss": 0.0579,
+      "step": 77105
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.2863906989827014e-05,
+      "loss": 0.0507,
+      "step": 77110
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.2863438188551874e-05,
+      "loss": 0.058,
+      "step": 77115
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.2862969387276734e-05,
+      "loss": 0.0483,
+      "step": 77120
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.2862500586001594e-05,
+      "loss": 0.0714,
+      "step": 77125
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.2862031784726457e-05,
+      "loss": 0.133,
+      "step": 77130
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.2861562983451317e-05,
+      "loss": 0.0765,
+      "step": 77135
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.2861094182176177e-05,
+      "loss": 0.1949,
+      "step": 77140
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.2860625380901037e-05,
+      "loss": 0.2952,
+      "step": 77145
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.2860156579625898e-05,
+      "loss": 0.0776,
+      "step": 77150
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.2859687778350758e-05,
+      "loss": 0.0083,
+      "step": 77155
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.2859218977075618e-05,
+      "loss": 0.0133,
+      "step": 77160
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.2858750175800478e-05,
+      "loss": 0.014,
+      "step": 77165
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.2858281374525338e-05,
+      "loss": 0.0443,
+      "step": 77170
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.2857812573250201e-05,
+      "loss": 0.0292,
+      "step": 77175
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.2857343771975061e-05,
+      "loss": 0.0948,
+      "step": 77180
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.2856874970699921e-05,
+      "loss": 0.0503,
+      "step": 77185
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.2856406169424783e-05,
+      "loss": 0.3029,
+      "step": 77190
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.2855937368149643e-05,
+      "loss": 0.2604,
+      "step": 77195
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.2855468566874503e-05,
+      "loss": 0.0711,
+      "step": 77200
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.2854999765599363e-05,
+      "loss": 0.0367,
+      "step": 77205
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.2854530964324223e-05,
+      "loss": 0.0417,
+      "step": 77210
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.2854062163049084e-05,
+      "loss": 0.0567,
+      "step": 77215
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.2853593361773946e-05,
+      "loss": 0.0547,
+      "step": 77220
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.2853124560498806e-05,
+      "loss": 0.0556,
+      "step": 77225
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.2852655759223667e-05,
+      "loss": 0.0986,
+      "step": 77230
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.2852186957948527e-05,
+      "loss": 0.1588,
+      "step": 77235
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.2851718156673387e-05,
+      "loss": 0.1353,
+      "step": 77240
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.2851249355398247e-05,
+      "loss": 0.1648,
+      "step": 77245
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.2850780554123107e-05,
+      "loss": 0.0773,
+      "step": 77250
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.2850311752847969e-05,
+      "loss": 0.0213,
+      "step": 77255
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.2849842951572829e-05,
+      "loss": 0.023,
+      "step": 77260
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.2849374150297689e-05,
+      "loss": 0.0577,
+      "step": 77265
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.2848905349022552e-05,
+      "loss": 0.0631,
+      "step": 77270
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.2848436547747412e-05,
+      "loss": 0.0435,
+      "step": 77275
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.2847967746472272e-05,
+      "loss": 0.1016,
+      "step": 77280
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.2847498945197132e-05,
+      "loss": 0.1129,
+      "step": 77285
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.2847030143921992e-05,
+      "loss": 0.1439,
+      "step": 77290
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.2846561342646853e-05,
+      "loss": 0.2883,
+      "step": 77295
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.2846092541371713e-05,
+      "loss": 0.0785,
+      "step": 77300
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.2845623740096573e-05,
+      "loss": 0.0401,
+      "step": 77305
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.2845154938821433e-05,
+      "loss": 0.0234,
+      "step": 77310
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.2844686137546296e-05,
+      "loss": 0.0456,
+      "step": 77315
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.2844217336271156e-05,
+      "loss": 0.0422,
+      "step": 77320
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.2843748534996016e-05,
+      "loss": 0.035,
+      "step": 77325
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.2843279733720876e-05,
+      "loss": 0.1707,
+      "step": 77330
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.2842810932445738e-05,
+      "loss": 0.1678,
+      "step": 77335
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.2842342131170598e-05,
+      "loss": 0.191,
+      "step": 77340
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.2841873329895458e-05,
+      "loss": 0.1558,
+      "step": 77345
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.2841404528620318e-05,
+      "loss": 0.0502,
+      "step": 77350
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.2840935727345178e-05,
+      "loss": 0.0446,
+      "step": 77355
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.2840466926070041e-05,
+      "loss": 0.0883,
+      "step": 77360
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.2839998124794901e-05,
+      "loss": 0.0766,
+      "step": 77365
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.283952932351976e-05,
+      "loss": 0.0502,
+      "step": 77370
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.2839060522244622e-05,
+      "loss": 0.0774,
+      "step": 77375
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.2838591720969482e-05,
+      "loss": 0.1192,
+      "step": 77380
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.2838122919694342e-05,
+      "loss": 0.0739,
+      "step": 77385
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.2837654118419202e-05,
+      "loss": 0.2167,
+      "step": 77390
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.2837185317144062e-05,
+      "loss": 0.2906,
+      "step": 77395
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.2836716515868924e-05,
+      "loss": 0.0427,
+      "step": 77400
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.2836247714593785e-05,
+      "loss": 0.0139,
+      "step": 77405
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.2835778913318645e-05,
+      "loss": 0.0063,
+      "step": 77410
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.2835310112043507e-05,
+      "loss": 0.0543,
+      "step": 77415
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.2834841310768367e-05,
+      "loss": 0.0994,
+      "step": 77420
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.2834372509493227e-05,
+      "loss": 0.1773,
+      "step": 77425
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.2833903708218087e-05,
+      "loss": 0.1949,
+      "step": 77430
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.2833434906942947e-05,
+      "loss": 0.1184,
+      "step": 77435
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.2832966105667808e-05,
+      "loss": 0.2374,
+      "step": 77440
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.2832497304392668e-05,
+      "loss": 0.2472,
+      "step": 77445
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.2832028503117528e-05,
+      "loss": 0.0869,
+      "step": 77450
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.2831559701842392e-05,
+      "loss": 0.018,
+      "step": 77455
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.2831090900567251e-05,
+      "loss": 0.028,
+      "step": 77460
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.2830622099292111e-05,
+      "loss": 0.0759,
+      "step": 77465
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.2830153298016971e-05,
+      "loss": 0.0593,
+      "step": 77470
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.2829684496741833e-05,
+      "loss": 0.1196,
+      "step": 77475
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.2829215695466693e-05,
+      "loss": 0.0831,
+      "step": 77480
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.2828746894191553e-05,
+      "loss": 0.2338,
+      "step": 77485
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.2828278092916413e-05,
+      "loss": 0.2097,
+      "step": 77490
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.2827809291641273e-05,
+      "loss": 0.2988,
+      "step": 77495
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.2827340490366136e-05,
+      "loss": 0.0616,
+      "step": 77500
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.2826871689090996e-05,
+      "loss": 0.0468,
+      "step": 77505
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.2826402887815856e-05,
+      "loss": 0.0371,
+      "step": 77510
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.2825934086540718e-05,
+      "loss": 0.0759,
+      "step": 77515
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.2825465285265577e-05,
+      "loss": 0.0393,
+      "step": 77520
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.2824996483990437e-05,
+      "loss": 0.0323,
+      "step": 77525
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.2824527682715297e-05,
+      "loss": 0.0613,
+      "step": 77530
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.2824058881440157e-05,
+      "loss": 0.182,
+      "step": 77535
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.2823590080165019e-05,
+      "loss": 0.1234,
+      "step": 77540
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.282312127888988e-05,
+      "loss": 0.2522,
+      "step": 77545
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.282265247761474e-05,
+      "loss": 0.0885,
+      "step": 77550
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.2822183676339602e-05,
+      "loss": 0.0178,
+      "step": 77555
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.2821714875064462e-05,
+      "loss": 0.0603,
+      "step": 77560
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.2821246073789322e-05,
+      "loss": 0.0397,
+      "step": 77565
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.2820777272514182e-05,
+      "loss": 0.0488,
+      "step": 77570
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.2820308471239042e-05,
+      "loss": 0.0395,
+      "step": 77575
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.2819839669963903e-05,
+      "loss": 0.0458,
+      "step": 77580
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.2819370868688763e-05,
+      "loss": 0.1616,
+      "step": 77585
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.2818902067413623e-05,
+      "loss": 0.1039,
+      "step": 77590
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.2818433266138487e-05,
+      "loss": 0.1839,
+      "step": 77595
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.2817964464863347e-05,
+      "loss": 0.0934,
+      "step": 77600
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.2817495663588207e-05,
+      "loss": 0.0272,
+      "step": 77605
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.2817026862313066e-05,
+      "loss": 0.042,
+      "step": 77610
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.2816558061037926e-05,
+      "loss": 0.0479,
+      "step": 77615
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.2816089259762788e-05,
+      "loss": 0.0599,
+      "step": 77620
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.2815620458487648e-05,
+      "loss": 0.0756,
+      "step": 77625
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.2815151657212508e-05,
+      "loss": 0.1672,
+      "step": 77630
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.2814682855937368e-05,
+      "loss": 0.0426,
+      "step": 77635
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.2814214054662231e-05,
+      "loss": 0.1304,
+      "step": 77640
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.2813745253387091e-05,
+      "loss": 0.2211,
+      "step": 77645
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.2813276452111951e-05,
+      "loss": 0.0842,
+      "step": 77650
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.2812807650836811e-05,
+      "loss": 0.041,
+      "step": 77655
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.2812338849561673e-05,
+      "loss": 0.0409,
+      "step": 77660
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.2811870048286532e-05,
+      "loss": 0.0172,
+      "step": 77665
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.2811401247011392e-05,
+      "loss": 0.0704,
+      "step": 77670
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.2810932445736252e-05,
+      "loss": 0.1615,
+      "step": 77675
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.2810463644461112e-05,
+      "loss": 0.1176,
+      "step": 77680
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.2809994843185976e-05,
+      "loss": 0.158,
+      "step": 77685
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.2809526041910836e-05,
+      "loss": 0.2321,
+      "step": 77690
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.2809057240635695e-05,
+      "loss": 0.2896,
+      "step": 77695
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.2808588439360557e-05,
+      "loss": 0.0511,
+      "step": 77700
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.2808119638085417e-05,
+      "loss": 0.0223,
+      "step": 77705
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.2807650836810277e-05,
+      "loss": 0.0048,
+      "step": 77710
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.2807182035535137e-05,
+      "loss": 0.0474,
+      "step": 77715
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.2806713234259997e-05,
+      "loss": 0.1519,
+      "step": 77720
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.2806244432984858e-05,
+      "loss": 0.0975,
+      "step": 77725
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.280577563170972e-05,
+      "loss": 0.0945,
+      "step": 77730
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.280530683043458e-05,
+      "loss": 0.1527,
+      "step": 77735
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.2804838029159442e-05,
+      "loss": 0.27,
+      "step": 77740
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.2804369227884302e-05,
+      "loss": 0.3309,
+      "step": 77745
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.2803900426609162e-05,
+      "loss": 0.045,
+      "step": 77750
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.2803431625334021e-05,
+      "loss": 0.0078,
+      "step": 77755
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.2802962824058881e-05,
+      "loss": 0.0441,
+      "step": 77760
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.2802494022783743e-05,
+      "loss": 0.0816,
+      "step": 77765
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.2802025221508603e-05,
+      "loss": 0.0397,
+      "step": 77770
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.2801556420233463e-05,
+      "loss": 0.1505,
+      "step": 77775
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.2801087618958326e-05,
+      "loss": 0.073,
+      "step": 77780
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.2800618817683186e-05,
+      "loss": 0.1564,
+      "step": 77785
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.2800150016408046e-05,
+      "loss": 0.176,
+      "step": 77790
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.2799681215132906e-05,
+      "loss": 0.2646,
+      "step": 77795
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.2799212413857766e-05,
+      "loss": 0.0586,
+      "step": 77800
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.2798743612582628e-05,
+      "loss": 0.0285,
+      "step": 77805
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.2798274811307488e-05,
+      "loss": 0.0784,
+      "step": 77810
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.2797806010032347e-05,
+      "loss": 0.0579,
+      "step": 77815
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.2797337208757207e-05,
+      "loss": 0.057,
+      "step": 77820
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.279686840748207e-05,
+      "loss": 0.0754,
+      "step": 77825
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.279639960620693e-05,
+      "loss": 0.0948,
+      "step": 77830
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.279593080493179e-05,
+      "loss": 0.1263,
+      "step": 77835
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.279546200365665e-05,
+      "loss": 0.1962,
+      "step": 77840
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.2794993202381512e-05,
+      "loss": 0.3922,
+      "step": 77845
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.2794524401106372e-05,
+      "loss": 0.0564,
+      "step": 77850
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.2794055599831232e-05,
+      "loss": 0.0262,
+      "step": 77855
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.2793586798556092e-05,
+      "loss": 0.0637,
+      "step": 77860
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.2793117997280954e-05,
+      "loss": 0.0372,
+      "step": 77865
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.2792649196005815e-05,
+      "loss": 0.0452,
+      "step": 77870
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.2792180394730675e-05,
+      "loss": 0.0574,
+      "step": 77875
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.2791711593455535e-05,
+      "loss": 0.0336,
+      "step": 77880
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.2791242792180397e-05,
+      "loss": 0.2095,
+      "step": 77885
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.2790773990905257e-05,
+      "loss": 0.2141,
+      "step": 77890
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.2790305189630117e-05,
+      "loss": 0.2808,
+      "step": 77895
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.2789836388354976e-05,
+      "loss": 0.0335,
+      "step": 77900
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.2789367587079838e-05,
+      "loss": 0.0293,
+      "step": 77905
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.2788898785804698e-05,
+      "loss": 0.0677,
+      "step": 77910
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.2788429984529558e-05,
+      "loss": 0.0894,
+      "step": 77915
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.278796118325442e-05,
+      "loss": 0.1004,
+      "step": 77920
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.2787492381979281e-05,
+      "loss": 0.1227,
+      "step": 77925
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.2787023580704141e-05,
+      "loss": 0.077,
+      "step": 77930
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.2786554779429001e-05,
+      "loss": 0.1198,
+      "step": 77935
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.2786085978153861e-05,
+      "loss": 0.166,
+      "step": 77940
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.2785617176878723e-05,
+      "loss": 0.2562,
+      "step": 77945
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.2785148375603583e-05,
+      "loss": 0.0703,
+      "step": 77950
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.2784679574328443e-05,
+      "loss": 0.0304,
+      "step": 77955
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.2784210773053302e-05,
+      "loss": 0.0579,
+      "step": 77960
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.2783741971778166e-05,
+      "loss": 0.0867,
+      "step": 77965
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.2783273170503026e-05,
+      "loss": 0.0752,
+      "step": 77970
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.2782804369227886e-05,
+      "loss": 0.089,
+      "step": 77975
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.2782335567952746e-05,
+      "loss": 0.1287,
+      "step": 77980
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.2781866766677607e-05,
+      "loss": 0.1227,
+      "step": 77985
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.2781397965402467e-05,
+      "loss": 0.2063,
+      "step": 77990
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.2780929164127327e-05,
+      "loss": 0.2305,
+      "step": 77995
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.2780460362852187e-05,
+      "loss": 0.0413,
+      "step": 78000
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.2779991561577047e-05,
+      "loss": 0.0298,
+      "step": 78005
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.277952276030191e-05,
+      "loss": 0.0333,
+      "step": 78010
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.277905395902677e-05,
+      "loss": 0.0585,
+      "step": 78015
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.277858515775163e-05,
+      "loss": 0.1186,
+      "step": 78020
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.2778116356476492e-05,
+      "loss": 0.1,
+      "step": 78025
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.2777647555201352e-05,
+      "loss": 0.1195,
+      "step": 78030
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.2777178753926212e-05,
+      "loss": 0.1112,
+      "step": 78035
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.2776709952651072e-05,
+      "loss": 0.2569,
+      "step": 78040
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.2776241151375931e-05,
+      "loss": 0.2934,
+      "step": 78045
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.2775772350100793e-05,
+      "loss": 0.1156,
+      "step": 78050
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.2775303548825655e-05,
+      "loss": 0.0148,
+      "step": 78055
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.2774834747550515e-05,
+      "loss": 0.0416,
+      "step": 78060
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.2774365946275376e-05,
+      "loss": 0.0416,
+      "step": 78065
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.2773897145000236e-05,
+      "loss": 0.1044,
+      "step": 78070
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.2773428343725096e-05,
+      "loss": 0.041,
+      "step": 78075
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.2772959542449956e-05,
+      "loss": 0.1944,
+      "step": 78080
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.2772490741174816e-05,
+      "loss": 0.1567,
+      "step": 78085
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.2772021939899678e-05,
+      "loss": 0.1879,
+      "step": 78090
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.2771553138624538e-05,
+      "loss": 0.2258,
+      "step": 78095
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.2771084337349398e-05,
+      "loss": 0.067,
+      "step": 78100
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.2770615536074261e-05,
+      "loss": 0.0147,
+      "step": 78105
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.277014673479912e-05,
+      "loss": 0.0195,
+      "step": 78110
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.276967793352398e-05,
+      "loss": 0.0328,
+      "step": 78115
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.276920913224884e-05,
+      "loss": 0.0619,
+      "step": 78120
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.27687403309737e-05,
+      "loss": 0.0703,
+      "step": 78125
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.2768271529698562e-05,
+      "loss": 0.1786,
+      "step": 78130
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.2767802728423422e-05,
+      "loss": 0.0753,
+      "step": 78135
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.2767333927148282e-05,
+      "loss": 0.1554,
+      "step": 78140
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.2766865125873142e-05,
+      "loss": 0.2485,
+      "step": 78145
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.2766396324598005e-05,
+      "loss": 0.0698,
+      "step": 78150
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.2765927523322865e-05,
+      "loss": 0.0372,
+      "step": 78155
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.2765458722047725e-05,
+      "loss": 0.0206,
+      "step": 78160
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.2764989920772585e-05,
+      "loss": 0.082,
+      "step": 78165
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.2764521119497447e-05,
+      "loss": 0.0817,
+      "step": 78170
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.2764052318222307e-05,
+      "loss": 0.1483,
+      "step": 78175
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.2763583516947167e-05,
+      "loss": 0.1579,
+      "step": 78180
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.2763114715672027e-05,
+      "loss": 0.1345,
+      "step": 78185
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.2762645914396887e-05,
+      "loss": 0.1934,
+      "step": 78190
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.276217711312175e-05,
+      "loss": 0.2748,
+      "step": 78195
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.276170831184661e-05,
+      "loss": 0.0228,
+      "step": 78200
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.276123951057147e-05,
+      "loss": 0.0207,
+      "step": 78205
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.2760770709296331e-05,
+      "loss": 0.0314,
+      "step": 78210
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.2760301908021191e-05,
+      "loss": 0.0844,
+      "step": 78215
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.2759833106746051e-05,
+      "loss": 0.1211,
+      "step": 78220
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.2759364305470911e-05,
+      "loss": 0.0499,
+      "step": 78225
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.2758895504195771e-05,
+      "loss": 0.127,
+      "step": 78230
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.2758426702920633e-05,
+      "loss": 0.1436,
+      "step": 78235
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.2757957901645493e-05,
+      "loss": 0.3458,
+      "step": 78240
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.2757489100370354e-05,
+      "loss": 0.2878,
+      "step": 78245
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.2757020299095216e-05,
+      "loss": 0.063,
+      "step": 78250
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.2756551497820076e-05,
+      "loss": 0.0437,
+      "step": 78255
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.2756082696544936e-05,
+      "loss": 0.0231,
+      "step": 78260
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.2755613895269796e-05,
+      "loss": 0.0648,
+      "step": 78265
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.2755145093994656e-05,
+      "loss": 0.0842,
+      "step": 78270
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.2754676292719517e-05,
+      "loss": 0.0898,
+      "step": 78275
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.2754207491444377e-05,
+      "loss": 0.079,
+      "step": 78280
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.2753738690169237e-05,
+      "loss": 0.1467,
+      "step": 78285
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.27532698888941e-05,
+      "loss": 0.163,
+      "step": 78290
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.275280108761896e-05,
+      "loss": 0.357,
+      "step": 78295
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.275233228634382e-05,
+      "loss": 0.0504,
+      "step": 78300
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.275186348506868e-05,
+      "loss": 0.0187,
+      "step": 78305
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.275139468379354e-05,
+      "loss": 0.0454,
+      "step": 78310
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.2750925882518402e-05,
+      "loss": 0.0805,
+      "step": 78315
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.2750457081243262e-05,
+      "loss": 0.1055,
+      "step": 78320
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.2749988279968122e-05,
+      "loss": 0.1327,
+      "step": 78325
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.2749519478692982e-05,
+      "loss": 0.0689,
+      "step": 78330
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.2749050677417845e-05,
+      "loss": 0.2196,
+      "step": 78335
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.2748581876142705e-05,
+      "loss": 0.1498,
+      "step": 78340
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.2748113074867565e-05,
+      "loss": 0.1981,
+      "step": 78345
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.2747644273592425e-05,
+      "loss": 0.0679,
+      "step": 78350
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.2747175472317286e-05,
+      "loss": 0.0213,
+      "step": 78355
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.2746706671042146e-05,
+      "loss": 0.0227,
+      "step": 78360
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.2746237869767006e-05,
+      "loss": 0.0756,
+      "step": 78365
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.2745769068491866e-05,
+      "loss": 0.072,
+      "step": 78370
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.2745300267216728e-05,
+      "loss": 0.059,
+      "step": 78375
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.274483146594159e-05,
+      "loss": 0.1238,
+      "step": 78380
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.274436266466645e-05,
+      "loss": 0.1507,
+      "step": 78385
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.274389386339131e-05,
+      "loss": 0.1631,
+      "step": 78390
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.2743425062116171e-05,
+      "loss": 0.2719,
+      "step": 78395
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.274295626084103e-05,
+      "loss": 0.1075,
+      "step": 78400
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.274248745956589e-05,
+      "loss": 0.0285,
+      "step": 78405
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.274201865829075e-05,
+      "loss": 0.0663,
+      "step": 78410
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.2741549857015612e-05,
+      "loss": 0.041,
+      "step": 78415
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.2741081055740472e-05,
+      "loss": 0.0262,
+      "step": 78420
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.2740612254465332e-05,
+      "loss": 0.077,
+      "step": 78425
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.2740143453190194e-05,
+      "loss": 0.0287,
+      "step": 78430
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.2739674651915055e-05,
+      "loss": 0.1625,
+      "step": 78435
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.2739205850639915e-05,
+      "loss": 0.1884,
+      "step": 78440
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.2738737049364775e-05,
+      "loss": 0.2646,
+      "step": 78445
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.2738268248089635e-05,
+      "loss": 0.0529,
+      "step": 78450
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.2737799446814497e-05,
+      "loss": 0.0175,
+      "step": 78455
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.2737330645539357e-05,
+      "loss": 0.0781,
+      "step": 78460
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.2736861844264217e-05,
+      "loss": 0.0381,
+      "step": 78465
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.2736393042989077e-05,
+      "loss": 0.0914,
+      "step": 78470
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.273592424171394e-05,
+      "loss": 0.0668,
+      "step": 78475
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.27354554404388e-05,
+      "loss": 0.0598,
+      "step": 78480
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.273498663916366e-05,
+      "loss": 0.0972,
+      "step": 78485
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.273451783788852e-05,
+      "loss": 0.154,
+      "step": 78490
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.2734049036613381e-05,
+      "loss": 0.1806,
+      "step": 78495
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.2733580235338241e-05,
+      "loss": 0.083,
+      "step": 78500
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.2733111434063101e-05,
+      "loss": 0.0532,
+      "step": 78505
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.2732642632787961e-05,
+      "loss": 0.0337,
+      "step": 78510
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.2732173831512821e-05,
+      "loss": 0.1008,
+      "step": 78515
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.2731705030237684e-05,
+      "loss": 0.0331,
+      "step": 78520
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.2731236228962544e-05,
+      "loss": 0.0478,
+      "step": 78525
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.2730767427687404e-05,
+      "loss": 0.1999,
+      "step": 78530
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.2730298626412266e-05,
+      "loss": 0.0914,
+      "step": 78535
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.2729829825137126e-05,
+      "loss": 0.2158,
+      "step": 78540
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.2729361023861986e-05,
+      "loss": 0.2418,
+      "step": 78545
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.2728892222586846e-05,
+      "loss": 0.0747,
+      "step": 78550
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.2728423421311706e-05,
+      "loss": 0.0427,
+      "step": 78555
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.2727954620036567e-05,
+      "loss": 0.0648,
+      "step": 78560
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.2727485818761427e-05,
+      "loss": 0.0626,
+      "step": 78565
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.2727017017486289e-05,
+      "loss": 0.0518,
+      "step": 78570
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.272654821621115e-05,
+      "loss": 0.0454,
+      "step": 78575
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.272607941493601e-05,
+      "loss": 0.0809,
+      "step": 78580
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.272561061366087e-05,
+      "loss": 0.0751,
+      "step": 78585
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.272514181238573e-05,
+      "loss": 0.2592,
+      "step": 78590
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.272467301111059e-05,
+      "loss": 0.2784,
+      "step": 78595
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.2724204209835452e-05,
+      "loss": 0.0902,
+      "step": 78600
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.2723735408560312e-05,
+      "loss": 0.0245,
+      "step": 78605
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.2723266607285172e-05,
+      "loss": 0.0941,
+      "step": 78610
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.2722797806010035e-05,
+      "loss": 0.0411,
+      "step": 78615
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.2722329004734895e-05,
+      "loss": 0.0267,
+      "step": 78620
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.2721860203459755e-05,
+      "loss": 0.0835,
+      "step": 78625
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.2721391402184615e-05,
+      "loss": 0.0921,
+      "step": 78630
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.2720922600909475e-05,
+      "loss": 0.1474,
+      "step": 78635
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.2720453799634336e-05,
+      "loss": 0.2466,
+      "step": 78640
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.2719984998359196e-05,
+      "loss": 0.2737,
+      "step": 78645
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.2719516197084056e-05,
+      "loss": 0.0547,
+      "step": 78650
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.2719047395808916e-05,
+      "loss": 0.0126,
+      "step": 78655
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.271857859453378e-05,
+      "loss": 0.0942,
+      "step": 78660
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.271810979325864e-05,
+      "loss": 0.0636,
+      "step": 78665
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.27176409919835e-05,
+      "loss": 0.0685,
+      "step": 78670
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.271717219070836e-05,
+      "loss": 0.1465,
+      "step": 78675
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.2716703389433221e-05,
+      "loss": 0.1381,
+      "step": 78680
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.2716234588158081e-05,
+      "loss": 0.1399,
+      "step": 78685
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.2715765786882941e-05,
+      "loss": 0.1988,
+      "step": 78690
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.27152969856078e-05,
+      "loss": 0.3297,
+      "step": 78695
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.271482818433266e-05,
+      "loss": 0.0692,
+      "step": 78700
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.2714359383057524e-05,
+      "loss": 0.0334,
+      "step": 78705
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.2713890581782384e-05,
+      "loss": 0.0716,
+      "step": 78710
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.2713421780507244e-05,
+      "loss": 0.0738,
+      "step": 78715
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.2712952979232106e-05,
+      "loss": 0.0247,
+      "step": 78720
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.2712484177956965e-05,
+      "loss": 0.0789,
+      "step": 78725
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.2712015376681825e-05,
+      "loss": 0.0535,
+      "step": 78730
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.2711546575406685e-05,
+      "loss": 0.1207,
+      "step": 78735
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.2711077774131545e-05,
+      "loss": 0.084,
+      "step": 78740
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.2710608972856407e-05,
+      "loss": 0.2068,
+      "step": 78745
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.2710140171581267e-05,
+      "loss": 0.0803,
+      "step": 78750
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.2709671370306128e-05,
+      "loss": 0.0285,
+      "step": 78755
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.270920256903099e-05,
+      "loss": 0.0382,
+      "step": 78760
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.270873376775585e-05,
+      "loss": 0.0453,
+      "step": 78765
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.270826496648071e-05,
+      "loss": 0.0484,
+      "step": 78770
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.270779616520557e-05,
+      "loss": 0.0617,
+      "step": 78775
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.270732736393043e-05,
+      "loss": 0.0609,
+      "step": 78780
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.2706858562655291e-05,
+      "loss": 0.1081,
+      "step": 78785
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.2706389761380151e-05,
+      "loss": 0.2445,
+      "step": 78790
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.2705920960105011e-05,
+      "loss": 0.1503,
+      "step": 78795
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.2705452158829875e-05,
+      "loss": 0.1369,
+      "step": 78800
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.2704983357554735e-05,
+      "loss": 0.0178,
+      "step": 78805
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.2704514556279595e-05,
+      "loss": 0.0554,
+      "step": 78810
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.2704045755004454e-05,
+      "loss": 0.051,
+      "step": 78815
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.2703576953729314e-05,
+      "loss": 0.0841,
+      "step": 78820
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.2703108152454176e-05,
+      "loss": 0.0833,
+      "step": 78825
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.2702639351179036e-05,
+      "loss": 0.0539,
+      "step": 78830
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.2702170549903896e-05,
+      "loss": 0.1065,
+      "step": 78835
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.2701701748628756e-05,
+      "loss": 0.1182,
+      "step": 78840
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.2701232947353619e-05,
+      "loss": 0.1948,
+      "step": 78845
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.2700764146078479e-05,
+      "loss": 0.0758,
+      "step": 78850
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.2700295344803339e-05,
+      "loss": 0.0617,
+      "step": 78855
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.2699826543528199e-05,
+      "loss": 0.0756,
+      "step": 78860
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.269935774225306e-05,
+      "loss": 0.0447,
+      "step": 78865
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.269888894097792e-05,
+      "loss": 0.1191,
+      "step": 78870
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.269842013970278e-05,
+      "loss": 0.0886,
+      "step": 78875
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.269795133842764e-05,
+      "loss": 0.1098,
+      "step": 78880
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.2697482537152502e-05,
+      "loss": 0.0687,
+      "step": 78885
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.2697013735877362e-05,
+      "loss": 0.1674,
+      "step": 78890
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.2696544934602224e-05,
+      "loss": 0.1744,
+      "step": 78895
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.2696076133327085e-05,
+      "loss": 0.0752,
+      "step": 78900
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.2695607332051945e-05,
+      "loss": 0.0548,
+      "step": 78905
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.2695138530776805e-05,
+      "loss": 0.0126,
+      "step": 78910
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.2694669729501665e-05,
+      "loss": 0.0724,
+      "step": 78915
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.2694200928226525e-05,
+      "loss": 0.0485,
+      "step": 78920
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.2693732126951387e-05,
+      "loss": 0.1092,
+      "step": 78925
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.2693263325676246e-05,
+      "loss": 0.0903,
+      "step": 78930
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.2692794524401106e-05,
+      "loss": 0.1176,
+      "step": 78935
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.269232572312597e-05,
+      "loss": 0.1223,
+      "step": 78940
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.269185692185083e-05,
+      "loss": 0.2449,
+      "step": 78945
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.269138812057569e-05,
+      "loss": 0.0657,
+      "step": 78950
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.269091931930055e-05,
+      "loss": 0.0164,
+      "step": 78955
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.269045051802541e-05,
+      "loss": 0.0325,
+      "step": 78960
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.2689981716750271e-05,
+      "loss": 0.0615,
+      "step": 78965
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.2689512915475131e-05,
+      "loss": 0.0515,
+      "step": 78970
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.2689044114199991e-05,
+      "loss": 0.0701,
+      "step": 78975
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.2688575312924851e-05,
+      "loss": 0.1229,
+      "step": 78980
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.2688106511649714e-05,
+      "loss": 0.187,
+      "step": 78985
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.2687637710374574e-05,
+      "loss": 0.1972,
+      "step": 78990
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.2687168909099434e-05,
+      "loss": 0.3156,
+      "step": 78995
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.2686700107824294e-05,
+      "loss": 0.0696,
+      "step": 79000
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.2686231306549156e-05,
+      "loss": 0.0243,
+      "step": 79005
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.2685762505274016e-05,
+      "loss": 0.0274,
+      "step": 79010
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.2685293703998876e-05,
+      "loss": 0.0367,
+      "step": 79015
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.2684824902723735e-05,
+      "loss": 0.0755,
+      "step": 79020
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.2684356101448595e-05,
+      "loss": 0.0655,
+      "step": 79025
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.2683887300173459e-05,
+      "loss": 0.0875,
+      "step": 79030
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.2683418498898319e-05,
+      "loss": 0.0886,
+      "step": 79035
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.2682949697623179e-05,
+      "loss": 0.2241,
+      "step": 79040
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.268248089634804e-05,
+      "loss": 0.2467,
+      "step": 79045
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.26820120950729e-05,
+      "loss": 0.0494,
+      "step": 79050
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.268154329379776e-05,
+      "loss": 0.0358,
+      "step": 79055
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.268107449252262e-05,
+      "loss": 0.0871,
+      "step": 79060
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.268060569124748e-05,
+      "loss": 0.0837,
+      "step": 79065
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.2680136889972342e-05,
+      "loss": 0.137,
+      "step": 79070
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.2679668088697201e-05,
+      "loss": 0.1637,
+      "step": 79075
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.2679199287422063e-05,
+      "loss": 0.0536,
+      "step": 79080
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.2678730486146925e-05,
+      "loss": 0.1471,
+      "step": 79085
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.2678261684871785e-05,
+      "loss": 0.1969,
+      "step": 79090
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.2677792883596645e-05,
+      "loss": 0.266,
+      "step": 79095
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.2677324082321505e-05,
+      "loss": 0.1092,
+      "step": 79100
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.2676855281046364e-05,
+      "loss": 0.0263,
+      "step": 79105
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.2676386479771226e-05,
+      "loss": 0.0262,
+      "step": 79110
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.2675917678496086e-05,
+      "loss": 0.0766,
+      "step": 79115
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.2675448877220946e-05,
+      "loss": 0.0506,
+      "step": 79120
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.267498007594581e-05,
+      "loss": 0.0422,
+      "step": 79125
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.267451127467067e-05,
+      "loss": 0.0788,
+      "step": 79130
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.267404247339553e-05,
+      "loss": 0.1153,
+      "step": 79135
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.2673573672120389e-05,
+      "loss": 0.2412,
+      "step": 79140
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.2673104870845249e-05,
+      "loss": 0.289,
+      "step": 79145
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.267263606957011e-05,
+      "loss": 0.086,
+      "step": 79150
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.267216726829497e-05,
+      "loss": 0.0452,
+      "step": 79155
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.267169846701983e-05,
+      "loss": 0.0435,
+      "step": 79160
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.267122966574469e-05,
+      "loss": 0.0802,
+      "step": 79165
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.2670760864469554e-05,
+      "loss": 0.0239,
+      "step": 79170
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.2670292063194414e-05,
+      "loss": 0.1055,
+      "step": 79175
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.2669823261919274e-05,
+      "loss": 0.0681,
+      "step": 79180
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.2669354460644134e-05,
+      "loss": 0.1015,
+      "step": 79185
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.2668885659368995e-05,
+      "loss": 0.1638,
+      "step": 79190
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.2668416858093855e-05,
+      "loss": 0.2779,
+      "step": 79195
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.2667948056818715e-05,
+      "loss": 0.077,
+      "step": 79200
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.2667479255543575e-05,
+      "loss": 0.036,
+      "step": 79205
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.2667010454268435e-05,
+      "loss": 0.0236,
+      "step": 79210
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.2666541652993297e-05,
+      "loss": 0.0436,
+      "step": 79215
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.2666072851718158e-05,
+      "loss": 0.0387,
+      "step": 79220
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.2665604050443018e-05,
+      "loss": 0.0819,
+      "step": 79225
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.266513524916788e-05,
+      "loss": 0.0543,
+      "step": 79230
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.266466644789274e-05,
+      "loss": 0.1324,
+      "step": 79235
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.26641976466176e-05,
+      "loss": 0.1139,
+      "step": 79240
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.266372884534246e-05,
+      "loss": 0.2049,
+      "step": 79245
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.2663260044067321e-05,
+      "loss": 0.0223,
+      "step": 79250
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.2662791242792181e-05,
+      "loss": 0.0313,
+      "step": 79255
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.2662322441517041e-05,
+      "loss": 0.0249,
+      "step": 79260
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.2661853640241903e-05,
+      "loss": 0.0203,
+      "step": 79265
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.2661384838966764e-05,
+      "loss": 0.0849,
+      "step": 79270
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.2660916037691624e-05,
+      "loss": 0.1066,
+      "step": 79275
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.2660447236416484e-05,
+      "loss": 0.1808,
+      "step": 79280
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.2659978435141344e-05,
+      "loss": 0.1255,
+      "step": 79285
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.2659509633866206e-05,
+      "loss": 0.1438,
+      "step": 79290
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.2659040832591066e-05,
+      "loss": 0.1604,
+      "step": 79295
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.2658572031315926e-05,
+      "loss": 0.0171,
+      "step": 79300
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.2658103230040786e-05,
+      "loss": 0.0412,
+      "step": 79305
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.2657634428765649e-05,
+      "loss": 0.0508,
+      "step": 79310
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.2657165627490509e-05,
+      "loss": 0.086,
+      "step": 79315
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.2656696826215369e-05,
+      "loss": 0.0974,
+      "step": 79320
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.2656228024940229e-05,
+      "loss": 0.0887,
+      "step": 79325
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.265575922366509e-05,
+      "loss": 0.1053,
+      "step": 79330
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.265529042238995e-05,
+      "loss": 0.1274,
+      "step": 79335
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.265482162111481e-05,
+      "loss": 0.1013,
+      "step": 79340
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.265435281983967e-05,
+      "loss": 0.2872,
+      "step": 79345
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.265388401856453e-05,
+      "loss": 0.043,
+      "step": 79350
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.2653415217289393e-05,
+      "loss": 0.0644,
+      "step": 79355
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.2652946416014253e-05,
+      "loss": 0.0232,
+      "step": 79360
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.2652477614739113e-05,
+      "loss": 0.0363,
+      "step": 79365
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.2652008813463975e-05,
+      "loss": 0.0281,
+      "step": 79370
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.2651540012188835e-05,
+      "loss": 0.0818,
+      "step": 79375
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.2651071210913695e-05,
+      "loss": 0.0831,
+      "step": 79380
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.2650602409638555e-05,
+      "loss": 0.2475,
+      "step": 79385
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.2650133608363415e-05,
+      "loss": 0.2168,
+      "step": 79390
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.2649664807088276e-05,
+      "loss": 0.2529,
+      "step": 79395
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.2649196005813136e-05,
+      "loss": 0.0632,
+      "step": 79400
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.2648727204537998e-05,
+      "loss": 0.0312,
+      "step": 79405
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.264825840326286e-05,
+      "loss": 0.046,
+      "step": 79410
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.264778960198772e-05,
+      "loss": 0.0752,
+      "step": 79415
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.264732080071258e-05,
+      "loss": 0.0744,
+      "step": 79420
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.264685199943744e-05,
+      "loss": 0.0209,
+      "step": 79425
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.2646383198162299e-05,
+      "loss": 0.1069,
+      "step": 79430
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.264591439688716e-05,
+      "loss": 0.1733,
+      "step": 79435
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.264544559561202e-05,
+      "loss": 0.2361,
+      "step": 79440
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.264497679433688e-05,
+      "loss": 0.2983,
+      "step": 79445
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.2644507993061744e-05,
+      "loss": 0.0935,
+      "step": 79450
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.2644039191786604e-05,
+      "loss": 0.0397,
+      "step": 79455
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.2643570390511464e-05,
+      "loss": 0.0362,
+      "step": 79460
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.2643101589236324e-05,
+      "loss": 0.0528,
+      "step": 79465
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.2642632787961184e-05,
+      "loss": 0.0264,
+      "step": 79470
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.2642163986686045e-05,
+      "loss": 0.0674,
+      "step": 79475
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.2641695185410905e-05,
+      "loss": 0.0648,
+      "step": 79480
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.2641226384135765e-05,
+      "loss": 0.1282,
+      "step": 79485
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.2640757582860625e-05,
+      "loss": 0.1517,
+      "step": 79490
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.2640288781585488e-05,
+      "loss": 0.2751,
+      "step": 79495
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.2639819980310348e-05,
+      "loss": 0.0242,
+      "step": 79500
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.2639351179035208e-05,
+      "loss": 0.0099,
+      "step": 79505
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.2638882377760068e-05,
+      "loss": 0.0235,
+      "step": 79510
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.263841357648493e-05,
+      "loss": 0.0483,
+      "step": 79515
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.263794477520979e-05,
+      "loss": 0.015,
+      "step": 79520
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.263747597393465e-05,
+      "loss": 0.0684,
+      "step": 79525
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.263700717265951e-05,
+      "loss": 0.1226,
+      "step": 79530
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.263653837138437e-05,
+      "loss": 0.1239,
+      "step": 79535
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.2636069570109231e-05,
+      "loss": 0.0897,
+      "step": 79540
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.2635600768834093e-05,
+      "loss": 0.3738,
+      "step": 79545
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.2635131967558953e-05,
+      "loss": 0.0834,
+      "step": 79550
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.2634663166283814e-05,
+      "loss": 0.0141,
+      "step": 79555
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.2634194365008674e-05,
+      "loss": 0.0216,
+      "step": 79560
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.2633725563733534e-05,
+      "loss": 0.0372,
+      "step": 79565
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.2633256762458394e-05,
+      "loss": 0.0566,
+      "step": 79570
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.2632787961183254e-05,
+      "loss": 0.0272,
+      "step": 79575
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.2632319159908116e-05,
+      "loss": 0.0713,
+      "step": 79580
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.2631850358632976e-05,
+      "loss": 0.1718,
+      "step": 79585
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.2631381557357837e-05,
+      "loss": 0.1198,
+      "step": 79590
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.2630912756082699e-05,
+      "loss": 0.2892,
+      "step": 79595
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.2630443954807559e-05,
+      "loss": 0.0324,
+      "step": 79600
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.2629975153532419e-05,
+      "loss": 0.0319,
+      "step": 79605
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.2629506352257279e-05,
+      "loss": 0.0529,
+      "step": 79610
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.2629037550982139e-05,
+      "loss": 0.015,
+      "step": 79615
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.2628568749707e-05,
+      "loss": 0.0752,
+      "step": 79620
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.262809994843186e-05,
+      "loss": 0.062,
+      "step": 79625
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.262763114715672e-05,
+      "loss": 0.1813,
+      "step": 79630
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.2627162345881584e-05,
+      "loss": 0.1121,
+      "step": 79635
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.2626693544606443e-05,
+      "loss": 0.1794,
+      "step": 79640
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.2626224743331303e-05,
+      "loss": 0.2176,
+      "step": 79645
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.2625755942056163e-05,
+      "loss": 0.0534,
+      "step": 79650
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.2625287140781023e-05,
+      "loss": 0.0285,
+      "step": 79655
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.2624818339505885e-05,
+      "loss": 0.0811,
+      "step": 79660
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.2624349538230745e-05,
+      "loss": 0.0518,
+      "step": 79665
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.2623880736955605e-05,
+      "loss": 0.0688,
+      "step": 79670
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.2623411935680465e-05,
+      "loss": 0.0886,
+      "step": 79675
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.2622943134405328e-05,
+      "loss": 0.0969,
+      "step": 79680
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.2622474333130188e-05,
+      "loss": 0.053,
+      "step": 79685
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.2622005531855048e-05,
+      "loss": 0.3041,
+      "step": 79690
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.2621536730579908e-05,
+      "loss": 0.2311,
+      "step": 79695
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.262106792930477e-05,
+      "loss": 0.0817,
+      "step": 79700
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.262059912802963e-05,
+      "loss": 0.0394,
+      "step": 79705
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.262013032675449e-05,
+      "loss": 0.0818,
+      "step": 79710
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.261966152547935e-05,
+      "loss": 0.0913,
+      "step": 79715
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.2619192724204211e-05,
+      "loss": 0.0738,
+      "step": 79720
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.261872392292907e-05,
+      "loss": 0.0892,
+      "step": 79725
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.2618255121653932e-05,
+      "loss": 0.1014,
+      "step": 79730
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.2617786320378792e-05,
+      "loss": 0.1061,
+      "step": 79735
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.2617317519103654e-05,
+      "loss": 0.2978,
+      "step": 79740
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.2616848717828514e-05,
+      "loss": 0.2318,
+      "step": 79745
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.2616379916553374e-05,
+      "loss": 0.0564,
+      "step": 79750
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.2615911115278234e-05,
+      "loss": 0.0441,
+      "step": 79755
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.2615442314003095e-05,
+      "loss": 0.0578,
+      "step": 79760
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.2614973512727955e-05,
+      "loss": 0.0798,
+      "step": 79765
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.2614504711452815e-05,
+      "loss": 0.0579,
+      "step": 79770
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.2614035910177677e-05,
+      "loss": 0.0825,
+      "step": 79775
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.2613567108902539e-05,
+      "loss": 0.0441,
+      "step": 79780
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.2613098307627398e-05,
+      "loss": 0.2085,
+      "step": 79785
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.2612629506352258e-05,
+      "loss": 0.2451,
+      "step": 79790
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.2612160705077118e-05,
+      "loss": 0.1445,
+      "step": 79795
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.261169190380198e-05,
+      "loss": 0.0607,
+      "step": 79800
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.261122310252684e-05,
+      "loss": 0.0031,
+      "step": 79805
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.26107543012517e-05,
+      "loss": 0.0603,
+      "step": 79810
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.261028549997656e-05,
+      "loss": 0.0963,
+      "step": 79815
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.2609816698701423e-05,
+      "loss": 0.1376,
+      "step": 79820
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.2609347897426283e-05,
+      "loss": 0.0389,
+      "step": 79825
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.2608879096151143e-05,
+      "loss": 0.0209,
+      "step": 79830
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.2608410294876003e-05,
+      "loss": 0.057,
+      "step": 79835
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.2607941493600865e-05,
+      "loss": 0.2354,
+      "step": 79840
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.2607472692325724e-05,
+      "loss": 0.2524,
+      "step": 79845
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.2607003891050584e-05,
+      "loss": 0.0598,
+      "step": 79850
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.2606535089775444e-05,
+      "loss": 0.0671,
+      "step": 79855
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.2606066288500304e-05,
+      "loss": 0.0201,
+      "step": 79860
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.2605597487225166e-05,
+      "loss": 0.0904,
+      "step": 79865
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.2605128685950027e-05,
+      "loss": 0.0632,
+      "step": 79870
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.2604659884674887e-05,
+      "loss": 0.062,
+      "step": 79875
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.2604191083399749e-05,
+      "loss": 0.0882,
+      "step": 79880
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.2603722282124609e-05,
+      "loss": 0.1525,
+      "step": 79885
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.2603253480849469e-05,
+      "loss": 0.1475,
+      "step": 79890
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.2602784679574329e-05,
+      "loss": 0.3101,
+      "step": 79895
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.2602315878299189e-05,
+      "loss": 0.0707,
+      "step": 79900
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.260184707702405e-05,
+      "loss": 0.0436,
+      "step": 79905
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.260137827574891e-05,
+      "loss": 0.0728,
+      "step": 79910
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.2600909474473772e-05,
+      "loss": 0.0497,
+      "step": 79915
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.2600440673198634e-05,
+      "loss": 0.0326,
+      "step": 79920
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.2599971871923494e-05,
+      "loss": 0.0354,
+      "step": 79925
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.2599503070648353e-05,
+      "loss": 0.1185,
+      "step": 79930
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.2599034269373213e-05,
+      "loss": 0.1294,
+      "step": 79935
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.2598565468098073e-05,
+      "loss": 0.1781,
+      "step": 79940
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.2598096666822935e-05,
+      "loss": 0.1797,
+      "step": 79945
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.2597627865547795e-05,
+      "loss": 0.0274,
+      "step": 79950
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.2597159064272655e-05,
+      "loss": 0.0551,
+      "step": 79955
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.2596690262997518e-05,
+      "loss": 0.0142,
+      "step": 79960
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.2596221461722378e-05,
+      "loss": 0.045,
+      "step": 79965
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.2595752660447238e-05,
+      "loss": 0.0673,
+      "step": 79970
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.2595283859172098e-05,
+      "loss": 0.1018,
+      "step": 79975
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.2594815057896958e-05,
+      "loss": 0.1354,
+      "step": 79980
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.259434625662182e-05,
+      "loss": 0.134,
+      "step": 79985
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.259387745534668e-05,
+      "loss": 0.2244,
+      "step": 79990
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.259340865407154e-05,
+      "loss": 0.2815,
+      "step": 79995
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.25929398527964e-05,
+      "loss": 0.0444,
+      "step": 80000
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.2592471051521263e-05,
+      "loss": 0.0804,
+      "step": 80005
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.2592002250246123e-05,
+      "loss": 0.0218,
+      "step": 80010
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.2591533448970983e-05,
+      "loss": 0.0604,
+      "step": 80015
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.2591064647695842e-05,
+      "loss": 0.0753,
+      "step": 80020
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.2590595846420704e-05,
+      "loss": 0.0844,
+      "step": 80025
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.2590127045145564e-05,
+      "loss": 0.0364,
+      "step": 80030
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.2589658243870424e-05,
+      "loss": 0.1538,
+      "step": 80035
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.2589189442595284e-05,
+      "loss": 0.1413,
+      "step": 80040
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2588720641320144e-05,
+      "loss": 0.2433,
+      "step": 80045
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2588251840045005e-05,
+      "loss": 0.0731,
+      "step": 80050
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2587783038769867e-05,
+      "loss": 0.0485,
+      "step": 80055
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2587314237494727e-05,
+      "loss": 0.0428,
+      "step": 80060
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2586845436219589e-05,
+      "loss": 0.0224,
+      "step": 80065
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2586376634944449e-05,
+      "loss": 0.0614,
+      "step": 80070
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2585907833669308e-05,
+      "loss": 0.0539,
+      "step": 80075
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2585439032394168e-05,
+      "loss": 0.1579,
+      "step": 80080
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2584970231119028e-05,
+      "loss": 0.1344,
+      "step": 80085
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.258450142984389e-05,
+      "loss": 0.1079,
+      "step": 80090
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.258403262856875e-05,
+      "loss": 0.3042,
+      "step": 80095
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2583563827293612e-05,
+      "loss": 0.0669,
+      "step": 80100
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2583095026018473e-05,
+      "loss": 0.0288,
+      "step": 80105
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2582626224743333e-05,
+      "loss": 0.0293,
+      "step": 80110
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2582157423468193e-05,
+      "loss": 0.0376,
+      "step": 80115
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2581688622193053e-05,
+      "loss": 0.0837,
+      "step": 80120
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2581219820917913e-05,
+      "loss": 0.0305,
+      "step": 80125
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2580751019642775e-05,
+      "loss": 0.0277,
+      "step": 80130
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2580282218367634e-05,
+      "loss": 0.1386,
+      "step": 80135
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2579813417092494e-05,
+      "loss": 0.2618,
+      "step": 80140
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2579344615817358e-05,
+      "loss": 0.3455,
+      "step": 80145
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2578875814542218e-05,
+      "loss": 0.0272,
+      "step": 80150
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2578407013267078e-05,
+      "loss": 0.0101,
+      "step": 80155
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2577938211991938e-05,
+      "loss": 0.0286,
+      "step": 80160
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2577469410716797e-05,
+      "loss": 0.0154,
+      "step": 80165
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2577000609441659e-05,
+      "loss": 0.0527,
+      "step": 80170
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2576531808166519e-05,
+      "loss": 0.054,
+      "step": 80175
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2576063006891379e-05,
+      "loss": 0.108,
+      "step": 80180
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2575594205616239e-05,
+      "loss": 0.1607,
+      "step": 80185
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.25751254043411e-05,
+      "loss": 0.2015,
+      "step": 80190
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2574656603065962e-05,
+      "loss": 0.2262,
+      "step": 80195
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2574187801790822e-05,
+      "loss": 0.0363,
+      "step": 80200
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2573719000515682e-05,
+      "loss": 0.0288,
+      "step": 80205
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2573250199240544e-05,
+      "loss": 0.0167,
+      "step": 80210
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2572781397965404e-05,
+      "loss": 0.0268,
+      "step": 80215
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2572312596690264e-05,
+      "loss": 0.0623,
+      "step": 80220
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2571843795415123e-05,
+      "loss": 0.097,
+      "step": 80225
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2571374994139985e-05,
+      "loss": 0.0578,
+      "step": 80230
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2570906192864845e-05,
+      "loss": 0.0878,
+      "step": 80235
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2570437391589707e-05,
+      "loss": 0.2988,
+      "step": 80240
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2569968590314567e-05,
+      "loss": 0.3496,
+      "step": 80245
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2569499789039428e-05,
+      "loss": 0.0512,
+      "step": 80250
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2569030987764288e-05,
+      "loss": 0.035,
+      "step": 80255
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2568562186489148e-05,
+      "loss": 0.0614,
+      "step": 80260
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2568093385214008e-05,
+      "loss": 0.0579,
+      "step": 80265
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.256762458393887e-05,
+      "loss": 0.0558,
+      "step": 80270
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.256715578266373e-05,
+      "loss": 0.1048,
+      "step": 80275
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.256668698138859e-05,
+      "loss": 0.0946,
+      "step": 80280
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2566218180113451e-05,
+      "loss": 0.1173,
+      "step": 80285
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2565749378838313e-05,
+      "loss": 0.192,
+      "step": 80290
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2565280577563173e-05,
+      "loss": 0.2514,
+      "step": 80295
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2564811776288033e-05,
+      "loss": 0.0597,
+      "step": 80300
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2564342975012893e-05,
+      "loss": 0.031,
+      "step": 80305
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2563874173737754e-05,
+      "loss": 0.0351,
+      "step": 80310
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2563405372462614e-05,
+      "loss": 0.0605,
+      "step": 80315
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2562936571187474e-05,
+      "loss": 0.0533,
+      "step": 80320
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2562467769912334e-05,
+      "loss": 0.096,
+      "step": 80325
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2561998968637197e-05,
+      "loss": 0.0566,
+      "step": 80330
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2561530167362057e-05,
+      "loss": 0.1548,
+      "step": 80335
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2561061366086917e-05,
+      "loss": 0.2723,
+      "step": 80340
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2560592564811777e-05,
+      "loss": 0.2603,
+      "step": 80345
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2560123763536639e-05,
+      "loss": 0.1099,
+      "step": 80350
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2559654962261499e-05,
+      "loss": 0.021,
+      "step": 80355
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2559186160986359e-05,
+      "loss": 0.0277,
+      "step": 80360
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2558717359711219e-05,
+      "loss": 0.0486,
+      "step": 80365
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2558248558436078e-05,
+      "loss": 0.0699,
+      "step": 80370
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.255777975716094e-05,
+      "loss": 0.0733,
+      "step": 80375
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2557310955885802e-05,
+      "loss": 0.0568,
+      "step": 80380
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2556842154610662e-05,
+      "loss": 0.0946,
+      "step": 80385
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2556373353335523e-05,
+      "loss": 0.1323,
+      "step": 80390
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2555904552060383e-05,
+      "loss": 0.2184,
+      "step": 80395
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2555435750785243e-05,
+      "loss": 0.0836,
+      "step": 80400
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2554966949510103e-05,
+      "loss": 0.0477,
+      "step": 80405
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2554498148234963e-05,
+      "loss": 0.0217,
+      "step": 80410
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2554029346959825e-05,
+      "loss": 0.0079,
+      "step": 80415
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2553560545684685e-05,
+      "loss": 0.0874,
+      "step": 80420
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2553091744409546e-05,
+      "loss": 0.0398,
+      "step": 80425
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2552622943134408e-05,
+      "loss": 0.0725,
+      "step": 80430
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2552154141859268e-05,
+      "loss": 0.1184,
+      "step": 80435
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2551685340584128e-05,
+      "loss": 0.1652,
+      "step": 80440
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2551216539308988e-05,
+      "loss": 0.3357,
+      "step": 80445
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2550747738033848e-05,
+      "loss": 0.0782,
+      "step": 80450
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.255027893675871e-05,
+      "loss": 0.0426,
+      "step": 80455
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2549810135483569e-05,
+      "loss": 0.015,
+      "step": 80460
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2549341334208429e-05,
+      "loss": 0.0465,
+      "step": 80465
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2548872532933292e-05,
+      "loss": 0.0607,
+      "step": 80470
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2548403731658152e-05,
+      "loss": 0.0239,
+      "step": 80475
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2547934930383012e-05,
+      "loss": 0.1087,
+      "step": 80480
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2547466129107872e-05,
+      "loss": 0.1056,
+      "step": 80485
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2546997327832732e-05,
+      "loss": 0.2304,
+      "step": 80490
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2546528526557594e-05,
+      "loss": 0.3079,
+      "step": 80495
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2546059725282454e-05,
+      "loss": 0.0593,
+      "step": 80500
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2545590924007314e-05,
+      "loss": 0.038,
+      "step": 80505
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2545122122732174e-05,
+      "loss": 0.0221,
+      "step": 80510
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2544653321457033e-05,
+      "loss": 0.0695,
+      "step": 80515
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2544184520181897e-05,
+      "loss": 0.0898,
+      "step": 80520
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2543715718906757e-05,
+      "loss": 0.0655,
+      "step": 80525
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2543246917631617e-05,
+      "loss": 0.0924,
+      "step": 80530
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2542778116356478e-05,
+      "loss": 0.1634,
+      "step": 80535
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2542309315081338e-05,
+      "loss": 0.2233,
+      "step": 80540
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2541840513806198e-05,
+      "loss": 0.2174,
+      "step": 80545
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2541371712531058e-05,
+      "loss": 0.0375,
+      "step": 80550
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2540902911255918e-05,
+      "loss": 0.0261,
+      "step": 80555
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.254043410998078e-05,
+      "loss": 0.03,
+      "step": 80560
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2539965308705641e-05,
+      "loss": 0.0588,
+      "step": 80565
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2539496507430501e-05,
+      "loss": 0.0313,
+      "step": 80570
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2539027706155363e-05,
+      "loss": 0.0533,
+      "step": 80575
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2538558904880223e-05,
+      "loss": 0.0913,
+      "step": 80580
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2538090103605083e-05,
+      "loss": 0.1477,
+      "step": 80585
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2537621302329943e-05,
+      "loss": 0.1816,
+      "step": 80590
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2537152501054803e-05,
+      "loss": 0.2674,
+      "step": 80595
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2536683699779664e-05,
+      "loss": 0.046,
+      "step": 80600
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2536214898504524e-05,
+      "loss": 0.0391,
+      "step": 80605
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2535746097229386e-05,
+      "loss": 0.0888,
+      "step": 80610
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2535277295954247e-05,
+      "loss": 0.0344,
+      "step": 80615
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2534808494679107e-05,
+      "loss": 0.0741,
+      "step": 80620
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2534339693403967e-05,
+      "loss": 0.0925,
+      "step": 80625
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2533870892128827e-05,
+      "loss": 0.1463,
+      "step": 80630
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2533402090853687e-05,
+      "loss": 0.1298,
+      "step": 80635
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2532933289578549e-05,
+      "loss": 0.1728,
+      "step": 80640
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2532464488303409e-05,
+      "loss": 0.4064,
+      "step": 80645
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2531995687028269e-05,
+      "loss": 0.0877,
+      "step": 80650
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2531526885753132e-05,
+      "loss": 0.039,
+      "step": 80655
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2531058084477992e-05,
+      "loss": 0.0609,
+      "step": 80660
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2530589283202852e-05,
+      "loss": 0.058,
+      "step": 80665
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2530120481927712e-05,
+      "loss": 0.0619,
+      "step": 80670
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2529651680652573e-05,
+      "loss": 0.1318,
+      "step": 80675
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2529182879377433e-05,
+      "loss": 0.0365,
+      "step": 80680
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2528714078102293e-05,
+      "loss": 0.1273,
+      "step": 80685
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2528245276827153e-05,
+      "loss": 0.0779,
+      "step": 80690
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2527776475552013e-05,
+      "loss": 0.2552,
+      "step": 80695
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2527307674276875e-05,
+      "loss": 0.0446,
+      "step": 80700
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2526838873001736e-05,
+      "loss": 0.014,
+      "step": 80705
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2526370071726596e-05,
+      "loss": 0.029,
+      "step": 80710
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2525901270451458e-05,
+      "loss": 0.0484,
+      "step": 80715
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2525432469176318e-05,
+      "loss": 0.0712,
+      "step": 80720
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2524963667901178e-05,
+      "loss": 0.1041,
+      "step": 80725
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2524494866626038e-05,
+      "loss": 0.1207,
+      "step": 80730
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2524026065350898e-05,
+      "loss": 0.1015,
+      "step": 80735
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.252355726407576e-05,
+      "loss": 0.161,
+      "step": 80740
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.252308846280062e-05,
+      "loss": 0.2087,
+      "step": 80745
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2522619661525481e-05,
+      "loss": 0.0644,
+      "step": 80750
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2522150860250342e-05,
+      "loss": 0.027,
+      "step": 80755
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2521682058975202e-05,
+      "loss": 0.0427,
+      "step": 80760
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2521213257700062e-05,
+      "loss": 0.0347,
+      "step": 80765
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2520744456424922e-05,
+      "loss": 0.0643,
+      "step": 80770
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2520275655149782e-05,
+      "loss": 0.078,
+      "step": 80775
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2519806853874644e-05,
+      "loss": 0.1366,
+      "step": 80780
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2519338052599504e-05,
+      "loss": 0.0901,
+      "step": 80785
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2518869251324364e-05,
+      "loss": 0.1419,
+      "step": 80790
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2518400450049227e-05,
+      "loss": 0.158,
+      "step": 80795
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2517931648774087e-05,
+      "loss": 0.0393,
+      "step": 80800
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2517462847498947e-05,
+      "loss": 0.0096,
+      "step": 80805
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2516994046223807e-05,
+      "loss": 0.0595,
+      "step": 80810
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2516525244948667e-05,
+      "loss": 0.0377,
+      "step": 80815
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2516056443673528e-05,
+      "loss": 0.1042,
+      "step": 80820
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2515587642398388e-05,
+      "loss": 0.0524,
+      "step": 80825
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2515118841123248e-05,
+      "loss": 0.1414,
+      "step": 80830
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2514650039848108e-05,
+      "loss": 0.1149,
+      "step": 80835
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2514181238572968e-05,
+      "loss": 0.2525,
+      "step": 80840
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2513712437297831e-05,
+      "loss": 0.3062,
+      "step": 80845
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2513243636022691e-05,
+      "loss": 0.0789,
+      "step": 80850
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2512774834747551e-05,
+      "loss": 0.0635,
+      "step": 80855
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2512306033472413e-05,
+      "loss": 0.0316,
+      "step": 80860
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2511837232197273e-05,
+      "loss": 0.0441,
+      "step": 80865
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2511368430922133e-05,
+      "loss": 0.0474,
+      "step": 80870
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2510899629646993e-05,
+      "loss": 0.0926,
+      "step": 80875
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2510430828371853e-05,
+      "loss": 0.1219,
+      "step": 80880
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2509962027096714e-05,
+      "loss": 0.105,
+      "step": 80885
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2509493225821576e-05,
+      "loss": 0.1523,
+      "step": 80890
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2509024424546436e-05,
+      "loss": 0.2556,
+      "step": 80895
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2508555623271297e-05,
+      "loss": 0.0617,
+      "step": 80900
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2508086821996157e-05,
+      "loss": 0.0293,
+      "step": 80905
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2507618020721017e-05,
+      "loss": 0.0318,
+      "step": 80910
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2507149219445877e-05,
+      "loss": 0.066,
+      "step": 80915
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2506680418170737e-05,
+      "loss": 0.0665,
+      "step": 80920
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2506211616895599e-05,
+      "loss": 0.0471,
+      "step": 80925
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2505742815620459e-05,
+      "loss": 0.0831,
+      "step": 80930
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.250527401434532e-05,
+      "loss": 0.1305,
+      "step": 80935
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2504805213070182e-05,
+      "loss": 0.2657,
+      "step": 80940
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2504336411795042e-05,
+      "loss": 0.2224,
+      "step": 80945
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2503867610519902e-05,
+      "loss": 0.0763,
+      "step": 80950
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2503398809244762e-05,
+      "loss": 0.0322,
+      "step": 80955
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2502930007969622e-05,
+      "loss": 0.0113,
+      "step": 80960
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2502461206694483e-05,
+      "loss": 0.0685,
+      "step": 80965
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2501992405419343e-05,
+      "loss": 0.0726,
+      "step": 80970
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2501523604144203e-05,
+      "loss": 0.0621,
+      "step": 80975
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2501054802869067e-05,
+      "loss": 0.1519,
+      "step": 80980
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2500586001593927e-05,
+      "loss": 0.1479,
+      "step": 80985
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2500117200318786e-05,
+      "loss": 0.177,
+      "step": 80990
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2499648399043646e-05,
+      "loss": 0.1956,
+      "step": 80995
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2499179597768506e-05,
+      "loss": 0.1411,
+      "step": 81000
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2498710796493368e-05,
+      "loss": 0.0162,
+      "step": 81005
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2498241995218228e-05,
+      "loss": 0.0299,
+      "step": 81010
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2497773193943088e-05,
+      "loss": 0.0735,
+      "step": 81015
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2497304392667948e-05,
+      "loss": 0.0619,
+      "step": 81020
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.249683559139281e-05,
+      "loss": 0.1804,
+      "step": 81025
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2496366790117671e-05,
+      "loss": 0.0614,
+      "step": 81030
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2495897988842531e-05,
+      "loss": 0.1022,
+      "step": 81035
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2495429187567391e-05,
+      "loss": 0.2208,
+      "step": 81040
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2494960386292253e-05,
+      "loss": 0.2193,
+      "step": 81045
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2494491585017112e-05,
+      "loss": 0.0922,
+      "step": 81050
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2494022783741972e-05,
+      "loss": 0.0116,
+      "step": 81055
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2493553982466832e-05,
+      "loss": 0.0212,
+      "step": 81060
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2493085181191694e-05,
+      "loss": 0.0263,
+      "step": 81065
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2492616379916554e-05,
+      "loss": 0.0611,
+      "step": 81070
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2492147578641415e-05,
+      "loss": 0.2149,
+      "step": 81075
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2491678777366275e-05,
+      "loss": 0.0776,
+      "step": 81080
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2491209976091137e-05,
+      "loss": 0.0936,
+      "step": 81085
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2490741174815997e-05,
+      "loss": 0.2098,
+      "step": 81090
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2490272373540857e-05,
+      "loss": 0.2427,
+      "step": 81095
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2489803572265717e-05,
+      "loss": 0.1095,
+      "step": 81100
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2489334770990578e-05,
+      "loss": 0.0136,
+      "step": 81105
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2488865969715438e-05,
+      "loss": 0.0428,
+      "step": 81110
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2488397168440298e-05,
+      "loss": 0.0439,
+      "step": 81115
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.248792836716516e-05,
+      "loss": 0.07,
+      "step": 81120
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2487459565890022e-05,
+      "loss": 0.0502,
+      "step": 81125
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2486990764614882e-05,
+      "loss": 0.1179,
+      "step": 81130
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2486521963339741e-05,
+      "loss": 0.1173,
+      "step": 81135
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2486053162064601e-05,
+      "loss": 0.1832,
+      "step": 81140
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2485584360789463e-05,
+      "loss": 0.3722,
+      "step": 81145
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2485115559514323e-05,
+      "loss": 0.0988,
+      "step": 81150
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2484646758239183e-05,
+      "loss": 0.0145,
+      "step": 81155
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2484177956964043e-05,
+      "loss": 0.0427,
+      "step": 81160
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2483709155688903e-05,
+      "loss": 0.0635,
+      "step": 81165
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2483240354413766e-05,
+      "loss": 0.0609,
+      "step": 81170
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2482771553138626e-05,
+      "loss": 0.0846,
+      "step": 81175
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2482302751863486e-05,
+      "loss": 0.1059,
+      "step": 81180
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2481833950588348e-05,
+      "loss": 0.1432,
+      "step": 81185
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2481365149313208e-05,
+      "loss": 0.1982,
+      "step": 81190
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2480896348038067e-05,
+      "loss": 0.3358,
+      "step": 81195
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2480427546762927e-05,
+      "loss": 0.0494,
+      "step": 81200
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2479958745487787e-05,
+      "loss": 0.0277,
+      "step": 81205
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2479489944212649e-05,
+      "loss": 0.0717,
+      "step": 81210
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.247902114293751e-05,
+      "loss": 0.0455,
+      "step": 81215
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.247855234166237e-05,
+      "loss": 0.1189,
+      "step": 81220
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2478083540387232e-05,
+      "loss": 0.0544,
+      "step": 81225
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2477614739112092e-05,
+      "loss": 0.1064,
+      "step": 81230
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2477145937836952e-05,
+      "loss": 0.1111,
+      "step": 81235
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2476677136561812e-05,
+      "loss": 0.1693,
+      "step": 81240
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2476208335286672e-05,
+      "loss": 0.278,
+      "step": 81245
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2475739534011534e-05,
+      "loss": 0.0397,
+      "step": 81250
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2475270732736393e-05,
+      "loss": 0.0198,
+      "step": 81255
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2474801931461255e-05,
+      "loss": 0.029,
+      "step": 81260
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2474333130186117e-05,
+      "loss": 0.035,
+      "step": 81265
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2473864328910977e-05,
+      "loss": 0.0945,
+      "step": 81270
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2473395527635837e-05,
+      "loss": 0.0477,
+      "step": 81275
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2472926726360696e-05,
+      "loss": 0.069,
+      "step": 81280
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2472457925085556e-05,
+      "loss": 0.1089,
+      "step": 81285
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2471989123810418e-05,
+      "loss": 0.1984,
+      "step": 81290
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2471520322535278e-05,
+      "loss": 0.3018,
+      "step": 81295
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2471051521260138e-05,
+      "loss": 0.0599,
+      "step": 81300
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2470582719985001e-05,
+      "loss": 0.0302,
+      "step": 81305
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2470113918709861e-05,
+      "loss": 0.0116,
+      "step": 81310
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2469645117434721e-05,
+      "loss": 0.0685,
+      "step": 81315
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2469176316159581e-05,
+      "loss": 0.1053,
+      "step": 81320
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2468707514884441e-05,
+      "loss": 0.1866,
+      "step": 81325
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2468238713609303e-05,
+      "loss": 0.0863,
+      "step": 81330
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2467769912334163e-05,
+      "loss": 0.0869,
+      "step": 81335
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2467301111059022e-05,
+      "loss": 0.1803,
+      "step": 81340
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2466832309783882e-05,
+      "loss": 0.4354,
+      "step": 81345
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2466363508508742e-05,
+      "loss": 0.0594,
+      "step": 81350
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2465894707233606e-05,
+      "loss": 0.0423,
+      "step": 81355
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2465425905958466e-05,
+      "loss": 0.0198,
+      "step": 81360
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2464957104683326e-05,
+      "loss": 0.0344,
+      "step": 81365
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2464488303408187e-05,
+      "loss": 0.0732,
+      "step": 81370
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2464019502133047e-05,
+      "loss": 0.0447,
+      "step": 81375
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2463550700857907e-05,
+      "loss": 0.1097,
+      "step": 81380
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2463081899582767e-05,
+      "loss": 0.1406,
+      "step": 81385
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2462613098307627e-05,
+      "loss": 0.1767,
+      "step": 81390
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2462144297032489e-05,
+      "loss": 0.3201,
+      "step": 81395
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.246167549575735e-05,
+      "loss": 0.0523,
+      "step": 81400
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.246120669448221e-05,
+      "loss": 0.012,
+      "step": 81405
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2460737893207072e-05,
+      "loss": 0.0234,
+      "step": 81410
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2460269091931932e-05,
+      "loss": 0.0656,
+      "step": 81415
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2459800290656792e-05,
+      "loss": 0.065,
+      "step": 81420
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2459331489381652e-05,
+      "loss": 0.1022,
+      "step": 81425
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2458862688106511e-05,
+      "loss": 0.0835,
+      "step": 81430
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2458393886831373e-05,
+      "loss": 0.0948,
+      "step": 81435
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2457925085556233e-05,
+      "loss": 0.1996,
+      "step": 81440
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2457456284281095e-05,
+      "loss": 0.2463,
+      "step": 81445
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2456987483005956e-05,
+      "loss": 0.0458,
+      "step": 81450
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2456518681730816e-05,
+      "loss": 0.0485,
+      "step": 81455
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2456049880455676e-05,
+      "loss": 0.0489,
+      "step": 81460
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2455581079180536e-05,
+      "loss": 0.051,
+      "step": 81465
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2455112277905396e-05,
+      "loss": 0.0888,
+      "step": 81470
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2454643476630258e-05,
+      "loss": 0.0656,
+      "step": 81475
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2454174675355118e-05,
+      "loss": 0.1618,
+      "step": 81480
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2453705874079977e-05,
+      "loss": 0.1987,
+      "step": 81485
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2453237072804837e-05,
+      "loss": 0.1733,
+      "step": 81490
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.24527682715297e-05,
+      "loss": 0.4266,
+      "step": 81495
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.245229947025456e-05,
+      "loss": 0.0668,
+      "step": 81500
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.245183066897942e-05,
+      "loss": 0.0266,
+      "step": 81505
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.245136186770428e-05,
+      "loss": 0.0385,
+      "step": 81510
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2450893066429142e-05,
+      "loss": 0.0189,
+      "step": 81515
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2450424265154002e-05,
+      "loss": 0.0433,
+      "step": 81520
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2449955463878862e-05,
+      "loss": 0.0716,
+      "step": 81525
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2449486662603722e-05,
+      "loss": 0.0516,
+      "step": 81530
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2449017861328584e-05,
+      "loss": 0.0534,
+      "step": 81535
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2448549060053445e-05,
+      "loss": 0.174,
+      "step": 81540
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.2448080258778305e-05,
+      "loss": 0.3213,
+      "step": 81545
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.2447611457503165e-05,
+      "loss": 0.2926,
+      "step": 81550
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.2447142656228027e-05,
+      "loss": 0.0328,
+      "step": 81555
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.2446673854952887e-05,
+      "loss": 0.042,
+      "step": 81560
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.2446205053677747e-05,
+      "loss": 0.0228,
+      "step": 81565
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.2445736252402607e-05,
+      "loss": 0.0259,
+      "step": 81570
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.2445267451127468e-05,
+      "loss": 0.0905,
+      "step": 81575
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.2444798649852328e-05,
+      "loss": 0.1253,
+      "step": 81580
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.244432984857719e-05,
+      "loss": 0.1169,
+      "step": 81585
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.244386104730205e-05,
+      "loss": 0.1853,
+      "step": 81590
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.2443392246026911e-05,
+      "loss": 0.2413,
+      "step": 81595
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.2442923444751771e-05,
+      "loss": 0.0897,
+      "step": 81600
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.2442454643476631e-05,
+      "loss": 0.0224,
+      "step": 81605
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.2441985842201491e-05,
+      "loss": 0.057,
+      "step": 81610
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.2441517040926353e-05,
+      "loss": 0.0515,
+      "step": 81615
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.2441048239651213e-05,
+      "loss": 0.073,
+      "step": 81620
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.2440579438376073e-05,
+      "loss": 0.0742,
+      "step": 81625
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.2440110637100934e-05,
+      "loss": 0.1327,
+      "step": 81630
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.2439641835825796e-05,
+      "loss": 0.0831,
+      "step": 81635
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.2439173034550656e-05,
+      "loss": 0.1077,
+      "step": 81640
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.2438704233275516e-05,
+      "loss": 0.2796,
+      "step": 81645
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.2438235432000376e-05,
+      "loss": 0.0781,
+      "step": 81650
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.2437766630725237e-05,
+      "loss": 0.0172,
+      "step": 81655
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.2437297829450097e-05,
+      "loss": 0.0175,
+      "step": 81660
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.2436829028174957e-05,
+      "loss": 0.0653,
+      "step": 81665
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.2436360226899817e-05,
+      "loss": 0.1016,
+      "step": 81670
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.2435891425624677e-05,
+      "loss": 0.046,
+      "step": 81675
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.243542262434954e-05,
+      "loss": 0.1179,
+      "step": 81680
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.24349538230744e-05,
+      "loss": 0.039,
+      "step": 81685
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.243448502179926e-05,
+      "loss": 0.1893,
+      "step": 81690
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.2434016220524122e-05,
+      "loss": 0.2309,
+      "step": 81695
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.2433547419248982e-05,
+      "loss": 0.0438,
+      "step": 81700
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.2433078617973842e-05,
+      "loss": 0.0377,
+      "step": 81705
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.2432609816698702e-05,
+      "loss": 0.0477,
+      "step": 81710
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.2432141015423562e-05,
+      "loss": 0.0524,
+      "step": 81715
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.2431672214148423e-05,
+      "loss": 0.0866,
+      "step": 81720
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.2431203412873285e-05,
+      "loss": 0.081,
+      "step": 81725
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.2430734611598145e-05,
+      "loss": 0.1324,
+      "step": 81730
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.2430265810323006e-05,
+      "loss": 0.147,
+      "step": 81735
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.2429797009047866e-05,
+      "loss": 0.1628,
+      "step": 81740
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.2429328207772726e-05,
+      "loss": 0.2222,
+      "step": 81745
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.2428859406497586e-05,
+      "loss": 0.118,
+      "step": 81750
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.2428390605222446e-05,
+      "loss": 0.0129,
+      "step": 81755
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.2427921803947308e-05,
+      "loss": 0.0356,
+      "step": 81760
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.2427453002672168e-05,
+      "loss": 0.0453,
+      "step": 81765
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.242698420139703e-05,
+      "loss": 0.0307,
+      "step": 81770
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.2426515400121891e-05,
+      "loss": 0.0505,
+      "step": 81775
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.242604659884675e-05,
+      "loss": 0.0904,
+      "step": 81780
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.242557779757161e-05,
+      "loss": 0.1525,
+      "step": 81785
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.242510899629647e-05,
+      "loss": 0.202,
+      "step": 81790
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.242464019502133e-05,
+      "loss": 0.2554,
+      "step": 81795
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.2424171393746192e-05,
+      "loss": 0.0533,
+      "step": 81800
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.2423702592471052e-05,
+      "loss": 0.0764,
+      "step": 81805
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.2423233791195912e-05,
+      "loss": 0.0551,
+      "step": 81810
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.2422764989920772e-05,
+      "loss": 0.035,
+      "step": 81815
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.2422296188645635e-05,
+      "loss": 0.0921,
+      "step": 81820
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.2421827387370495e-05,
+      "loss": 0.0407,
+      "step": 81825
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.2421358586095355e-05,
+      "loss": 0.1333,
+      "step": 81830
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.2420889784820215e-05,
+      "loss": 0.1425,
+      "step": 81835
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.2420420983545077e-05,
+      "loss": 0.2069,
+      "step": 81840
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.2419952182269937e-05,
+      "loss": 0.2449,
+      "step": 81845
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.2419483380994797e-05,
+      "loss": 0.0492,
+      "step": 81850
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.2419014579719657e-05,
+      "loss": 0.0101,
+      "step": 81855
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.2418545778444517e-05,
+      "loss": 0.0477,
+      "step": 81860
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.241807697716938e-05,
+      "loss": 0.0073,
+      "step": 81865
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.241760817589424e-05,
+      "loss": 0.0802,
+      "step": 81870
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.24171393746191e-05,
+      "loss": 0.0941,
+      "step": 81875
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.2416670573343961e-05,
+      "loss": 0.0735,
+      "step": 81880
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.2416201772068821e-05,
+      "loss": 0.0528,
+      "step": 81885
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.2415732970793681e-05,
+      "loss": 0.1818,
+      "step": 81890
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.2415264169518541e-05,
+      "loss": 0.2368,
+      "step": 81895
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.2414795368243401e-05,
+      "loss": 0.0771,
+      "step": 81900
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.2414326566968263e-05,
+      "loss": 0.0333,
+      "step": 81905
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.2413857765693124e-05,
+      "loss": 0.0334,
+      "step": 81910
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.2413388964417984e-05,
+      "loss": 0.0567,
+      "step": 81915
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.2412920163142846e-05,
+      "loss": 0.0596,
+      "step": 81920
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.2412451361867706e-05,
+      "loss": 0.0773,
+      "step": 81925
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.2411982560592566e-05,
+      "loss": 0.0531,
+      "step": 81930
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.2411513759317426e-05,
+      "loss": 0.0679,
+      "step": 81935
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.2411044958042286e-05,
+      "loss": 0.2293,
+      "step": 81940
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.2410576156767147e-05,
+      "loss": 0.1537,
+      "step": 81945
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.2410107355492007e-05,
+      "loss": 0.0234,
+      "step": 81950
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.2409638554216869e-05,
+      "loss": 0.0393,
+      "step": 81955
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.240916975294173e-05,
+      "loss": 0.0569,
+      "step": 81960
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.240870095166659e-05,
+      "loss": 0.0688,
+      "step": 81965
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.240823215039145e-05,
+      "loss": 0.033,
+      "step": 81970
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.240776334911631e-05,
+      "loss": 0.1131,
+      "step": 81975
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.240729454784117e-05,
+      "loss": 0.0823,
+      "step": 81980
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.2406825746566032e-05,
+      "loss": 0.0743,
+      "step": 81985
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.2406356945290892e-05,
+      "loss": 0.1694,
+      "step": 81990
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.2405888144015752e-05,
+      "loss": 0.3186,
+      "step": 81995
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.2405419342740612e-05,
+      "loss": 0.0837,
+      "step": 82000
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.2404950541465475e-05,
+      "loss": 0.0543,
+      "step": 82005
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.2404481740190335e-05,
+      "loss": 0.0624,
+      "step": 82010
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.2404012938915195e-05,
+      "loss": 0.0256,
+      "step": 82015
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.2403544137640055e-05,
+      "loss": 0.089,
+      "step": 82020
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.2403075336364916e-05,
+      "loss": 0.0715,
+      "step": 82025
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.2402606535089776e-05,
+      "loss": 0.166,
+      "step": 82030
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.2402137733814636e-05,
+      "loss": 0.1849,
+      "step": 82035
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.2401668932539496e-05,
+      "loss": 0.2395,
+      "step": 82040
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.2401200131264358e-05,
+      "loss": 0.2466,
+      "step": 82045
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.240073132998922e-05,
+      "loss": 0.0628,
+      "step": 82050
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.240026252871408e-05,
+      "loss": 0.0546,
+      "step": 82055
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.2399793727438941e-05,
+      "loss": 0.0379,
+      "step": 82060
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.2399324926163801e-05,
+      "loss": 0.0424,
+      "step": 82065
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.2398856124888661e-05,
+      "loss": 0.0487,
+      "step": 82070
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.239838732361352e-05,
+      "loss": 0.1114,
+      "step": 82075
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.239791852233838e-05,
+      "loss": 0.06,
+      "step": 82080
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.2397449721063242e-05,
+      "loss": 0.1358,
+      "step": 82085
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.2396980919788102e-05,
+      "loss": 0.2307,
+      "step": 82090
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.2396512118512964e-05,
+      "loss": 0.3083,
+      "step": 82095
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.2396043317237826e-05,
+      "loss": 0.0341,
+      "step": 82100
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.2395574515962685e-05,
+      "loss": 0.0243,
+      "step": 82105
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.2395105714687545e-05,
+      "loss": 0.0231,
+      "step": 82110
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.2394636913412405e-05,
+      "loss": 0.0442,
+      "step": 82115
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.2394168112137265e-05,
+      "loss": 0.1169,
+      "step": 82120
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.2393699310862127e-05,
+      "loss": 0.1093,
+      "step": 82125
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.2393230509586987e-05,
+      "loss": 0.1125,
+      "step": 82130
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.2392761708311847e-05,
+      "loss": 0.1146,
+      "step": 82135
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.2392292907036707e-05,
+      "loss": 0.2026,
+      "step": 82140
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.239182410576157e-05,
+      "loss": 0.3897,
+      "step": 82145
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.239135530448643e-05,
+      "loss": 0.0805,
+      "step": 82150
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.239088650321129e-05,
+      "loss": 0.037,
+      "step": 82155
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.239041770193615e-05,
+      "loss": 0.0358,
+      "step": 82160
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.2389948900661011e-05,
+      "loss": 0.019,
+      "step": 82165
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.2389480099385871e-05,
+      "loss": 0.0678,
+      "step": 82170
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.2389011298110731e-05,
+      "loss": 0.1076,
+      "step": 82175
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.2388542496835591e-05,
+      "loss": 0.1069,
+      "step": 82180
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.2388073695560451e-05,
+      "loss": 0.1264,
+      "step": 82185
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.2387604894285315e-05,
+      "loss": 0.1963,
+      "step": 82190
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.2387136093010174e-05,
+      "loss": 0.3654,
+      "step": 82195
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.2386667291735034e-05,
+      "loss": 0.0684,
+      "step": 82200
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.2386198490459896e-05,
+      "loss": 0.0057,
+      "step": 82205
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.2385729689184756e-05,
+      "loss": 0.0854,
+      "step": 82210
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.2385260887909616e-05,
+      "loss": 0.069,
+      "step": 82215
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.2384792086634476e-05,
+      "loss": 0.0872,
+      "step": 82220
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.2384323285359336e-05,
+      "loss": 0.0495,
+      "step": 82225
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.2383854484084197e-05,
+      "loss": 0.1551,
+      "step": 82230
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.2383385682809059e-05,
+      "loss": 0.1237,
+      "step": 82235
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.2382916881533919e-05,
+      "loss": 0.327,
+      "step": 82240
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.238244808025878e-05,
+      "loss": 0.292,
+      "step": 82245
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.238197927898364e-05,
+      "loss": 0.0647,
+      "step": 82250
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.23815104777085e-05,
+      "loss": 0.0344,
+      "step": 82255
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.238104167643336e-05,
+      "loss": 0.0264,
+      "step": 82260
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.238057287515822e-05,
+      "loss": 0.0361,
+      "step": 82265
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.2380104073883082e-05,
+      "loss": 0.0335,
+      "step": 82270
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.2379635272607942e-05,
+      "loss": 0.0492,
+      "step": 82275
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.2379166471332804e-05,
+      "loss": 0.1007,
+      "step": 82280
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.2378697670057665e-05,
+      "loss": 0.106,
+      "step": 82285
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.2378228868782525e-05,
+      "loss": 0.235,
+      "step": 82290
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.2377760067507385e-05,
+      "loss": 0.2,
+      "step": 82295
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.2377291266232245e-05,
+      "loss": 0.0597,
+      "step": 82300
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.2376822464957105e-05,
+      "loss": 0.0244,
+      "step": 82305
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.2376353663681966e-05,
+      "loss": 0.0544,
+      "step": 82310
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.2375884862406826e-05,
+      "loss": 0.0364,
+      "step": 82315
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.2375416061131686e-05,
+      "loss": 0.1265,
+      "step": 82320
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.2374947259856546e-05,
+      "loss": 0.1188,
+      "step": 82325
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.237447845858141e-05,
+      "loss": 0.1075,
+      "step": 82330
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.237400965730627e-05,
+      "loss": 0.0742,
+      "step": 82335
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.237354085603113e-05,
+      "loss": 0.3229,
+      "step": 82340
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.237307205475599e-05,
+      "loss": 0.3322,
+      "step": 82345
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.2372603253480851e-05,
+      "loss": 0.0564,
+      "step": 82350
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.2372134452205711e-05,
+      "loss": 0.0833,
+      "step": 82355
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.2371665650930571e-05,
+      "loss": 0.0352,
+      "step": 82360
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.237119684965543e-05,
+      "loss": 0.0388,
+      "step": 82365
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.237072804838029e-05,
+      "loss": 0.0282,
+      "step": 82370
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.2370259247105154e-05,
+      "loss": 0.0479,
+      "step": 82375
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.2369790445830014e-05,
+      "loss": 0.1388,
+      "step": 82380
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.2369321644554874e-05,
+      "loss": 0.0462,
+      "step": 82385
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.2368852843279736e-05,
+      "loss": 0.1379,
+      "step": 82390
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.2368384042004596e-05,
+      "loss": 0.2816,
+      "step": 82395
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.2367915240729455e-05,
+      "loss": 0.072,
+      "step": 82400
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.2367446439454315e-05,
+      "loss": 0.0364,
+      "step": 82405
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.2366977638179177e-05,
+      "loss": 0.05,
+      "step": 82410
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.2366508836904037e-05,
+      "loss": 0.0285,
+      "step": 82415
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.2366040035628899e-05,
+      "loss": 0.058,
+      "step": 82420
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.2365571234353759e-05,
+      "loss": 0.0935,
+      "step": 82425
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.236510243307862e-05,
+      "loss": 0.0923,
+      "step": 82430
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.236463363180348e-05,
+      "loss": 0.1319,
+      "step": 82435
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.236416483052834e-05,
+      "loss": 0.081,
+      "step": 82440
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.23636960292532e-05,
+      "loss": 0.2415,
+      "step": 82445
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.2363227227978062e-05,
+      "loss": 0.0919,
+      "step": 82450
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.2362758426702922e-05,
+      "loss": 0.0045,
+      "step": 82455
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.2362289625427781e-05,
+      "loss": 0.0395,
+      "step": 82460
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.2361820824152641e-05,
+      "loss": 0.0473,
+      "step": 82465
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.2361352022877505e-05,
+      "loss": 0.0098,
+      "step": 82470
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.2360883221602365e-05,
+      "loss": 0.1075,
+      "step": 82475
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.2360414420327225e-05,
+      "loss": 0.0496,
+      "step": 82480
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.2359945619052084e-05,
+      "loss": 0.135,
+      "step": 82485
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.2359476817776946e-05,
+      "loss": 0.1991,
+      "step": 82490
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.2359008016501806e-05,
+      "loss": 0.1782,
+      "step": 82495
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.2358539215226666e-05,
+      "loss": 0.0896,
+      "step": 82500
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.2358070413951526e-05,
+      "loss": 0.0338,
+      "step": 82505
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.2357601612676386e-05,
+      "loss": 0.033,
+      "step": 82510
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.235713281140125e-05,
+      "loss": 0.0812,
+      "step": 82515
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.2356664010126109e-05,
+      "loss": 0.0989,
+      "step": 82520
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.2356195208850969e-05,
+      "loss": 0.0882,
+      "step": 82525
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.235572640757583e-05,
+      "loss": 0.0474,
+      "step": 82530
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.235525760630069e-05,
+      "loss": 0.0846,
+      "step": 82535
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.235478880502555e-05,
+      "loss": 0.1838,
+      "step": 82540
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.235432000375041e-05,
+      "loss": 0.1798,
+      "step": 82545
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.235385120247527e-05,
+      "loss": 0.0859,
+      "step": 82550
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.2353382401200132e-05,
+      "loss": 0.0166,
+      "step": 82555
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.2352913599924994e-05,
+      "loss": 0.0137,
+      "step": 82560
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.2352444798649854e-05,
+      "loss": 0.013,
+      "step": 82565
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.2351975997374715e-05,
+      "loss": 0.046,
+      "step": 82570
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.2351507196099575e-05,
+      "loss": 0.0529,
+      "step": 82575
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.2351038394824435e-05,
+      "loss": 0.0893,
+      "step": 82580
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.2350569593549295e-05,
+      "loss": 0.1083,
+      "step": 82585
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.2350100792274155e-05,
+      "loss": 0.1114,
+      "step": 82590
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.2349631990999017e-05,
+      "loss": 0.3898,
+      "step": 82595
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.2349163189723877e-05,
+      "loss": 0.0617,
+      "step": 82600
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.2348694388448738e-05,
+      "loss": 0.025,
+      "step": 82605
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.23482255871736e-05,
+      "loss": 0.0496,
+      "step": 82610
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.234775678589846e-05,
+      "loss": 0.0378,
+      "step": 82615
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.234728798462332e-05,
+      "loss": 0.1224,
+      "step": 82620
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.234681918334818e-05,
+      "loss": 0.1153,
+      "step": 82625
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.234635038207304e-05,
+      "loss": 0.0989,
+      "step": 82630
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.2345881580797901e-05,
+      "loss": 0.1872,
+      "step": 82635
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.2345412779522761e-05,
+      "loss": 0.2531,
+      "step": 82640
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.2344943978247621e-05,
+      "loss": 0.2522,
+      "step": 82645
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.2344475176972481e-05,
+      "loss": 0.0415,
+      "step": 82650
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.2344006375697344e-05,
+      "loss": 0.0609,
+      "step": 82655
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.2343537574422204e-05,
+      "loss": 0.0157,
+      "step": 82660
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.2343068773147064e-05,
+      "loss": 0.1067,
+      "step": 82665
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.2342599971871924e-05,
+      "loss": 0.1152,
+      "step": 82670
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.2342131170596786e-05,
+      "loss": 0.0615,
+      "step": 82675
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.2341662369321646e-05,
+      "loss": 0.1448,
+      "step": 82680
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.2341193568046506e-05,
+      "loss": 0.1699,
+      "step": 82685
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.2340724766771365e-05,
+      "loss": 0.2683,
+      "step": 82690
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.2340255965496225e-05,
+      "loss": 0.2239,
+      "step": 82695
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.2339787164221089e-05,
+      "loss": 0.0372,
+      "step": 82700
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.2339318362945949e-05,
+      "loss": 0.02,
+      "step": 82705
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.2338849561670809e-05,
+      "loss": 0.0412,
+      "step": 82710
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.233838076039567e-05,
+      "loss": 0.0322,
+      "step": 82715
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.233791195912053e-05,
+      "loss": 0.0874,
+      "step": 82720
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.233744315784539e-05,
+      "loss": 0.0607,
+      "step": 82725
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.233697435657025e-05,
+      "loss": 0.13,
+      "step": 82730
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.233650555529511e-05,
+      "loss": 0.1419,
+      "step": 82735
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.2336036754019972e-05,
+      "loss": 0.1799,
+      "step": 82740
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.2335567952744833e-05,
+      "loss": 0.1997,
+      "step": 82745
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.2335099151469693e-05,
+      "loss": 0.054,
+      "step": 82750
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.2334630350194555e-05,
+      "loss": 0.0319,
+      "step": 82755
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.2334161548919415e-05,
+      "loss": 0.0091,
+      "step": 82760
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.2333692747644275e-05,
+      "loss": 0.0929,
+      "step": 82765
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.2333223946369135e-05,
+      "loss": 0.052,
+      "step": 82770
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.2332755145093995e-05,
+      "loss": 0.0458,
+      "step": 82775
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.2332286343818856e-05,
+      "loss": 0.06,
+      "step": 82780
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.2331817542543716e-05,
+      "loss": 0.2001,
+      "step": 82785
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.2331348741268576e-05,
+      "loss": 0.198,
+      "step": 82790
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.233087993999344e-05,
+      "loss": 0.249,
+      "step": 82795
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.23304111387183e-05,
+      "loss": 0.0389,
+      "step": 82800
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.232994233744316e-05,
+      "loss": 0.03,
+      "step": 82805
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.2329473536168019e-05,
+      "loss": 0.0357,
+      "step": 82810
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.2329004734892879e-05,
+      "loss": 0.0694,
+      "step": 82815
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.232853593361774e-05,
+      "loss": 0.0476,
+      "step": 82820
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.23280671323426e-05,
+      "loss": 0.1148,
+      "step": 82825
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.232759833106746e-05,
+      "loss": 0.1288,
+      "step": 82830
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.232712952979232e-05,
+      "loss": 0.0922,
+      "step": 82835
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.2326660728517184e-05,
+      "loss": 0.1951,
+      "step": 82840
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.2326191927242044e-05,
+      "loss": 0.2239,
+      "step": 82845
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.2325723125966904e-05,
+      "loss": 0.0755,
+      "step": 82850
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.2325254324691764e-05,
+      "loss": 0.033,
+      "step": 82855
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.2324785523416625e-05,
+      "loss": 0.055,
+      "step": 82860
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.2324316722141485e-05,
+      "loss": 0.0388,
+      "step": 82865
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.2323847920866345e-05,
+      "loss": 0.114,
+      "step": 82870
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.2323379119591205e-05,
+      "loss": 0.0769,
+      "step": 82875
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.2322910318316067e-05,
+      "loss": 0.112,
+      "step": 82880
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.2322441517040928e-05,
+      "loss": 0.164,
+      "step": 82885
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.2321972715765788e-05,
+      "loss": 0.3269,
+      "step": 82890
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.2321503914490648e-05,
+      "loss": 0.4316,
+      "step": 82895
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.232103511321551e-05,
+      "loss": 0.0552,
+      "step": 82900
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.232056631194037e-05,
+      "loss": 0.0282,
+      "step": 82905
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.232009751066523e-05,
+      "loss": 0.0203,
+      "step": 82910
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.231962870939009e-05,
+      "loss": 0.0557,
+      "step": 82915
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.2319159908114951e-05,
+      "loss": 0.0393,
+      "step": 82920
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.2318691106839811e-05,
+      "loss": 0.0684,
+      "step": 82925
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.2318222305564673e-05,
+      "loss": 0.0716,
+      "step": 82930
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.2317753504289533e-05,
+      "loss": 0.1523,
+      "step": 82935
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.2317284703014394e-05,
+      "loss": 0.1399,
+      "step": 82940
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.2316815901739254e-05,
+      "loss": 0.1503,
+      "step": 82945
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.2316347100464114e-05,
+      "loss": 0.0584,
+      "step": 82950
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.2315878299188974e-05,
+      "loss": 0.0096,
+      "step": 82955
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.2315409497913836e-05,
+      "loss": 0.0223,
+      "step": 82960
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.2314940696638696e-05,
+      "loss": 0.0296,
+      "step": 82965
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.2314471895363556e-05,
+      "loss": 0.0519,
+      "step": 82970
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.2314003094088416e-05,
+      "loss": 0.0427,
+      "step": 82975
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.2313534292813279e-05,
+      "loss": 0.0689,
+      "step": 82980
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.2313065491538139e-05,
+      "loss": 0.1305,
+      "step": 82985
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.2312596690262999e-05,
+      "loss": 0.1061,
+      "step": 82990
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.2312127888987859e-05,
+      "loss": 0.2652,
+      "step": 82995
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.231165908771272e-05,
+      "loss": 0.0835,
+      "step": 83000
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.231119028643758e-05,
+      "loss": 0.0156,
+      "step": 83005
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.231072148516244e-05,
+      "loss": 0.0279,
+      "step": 83010
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.23102526838873e-05,
+      "loss": 0.0313,
+      "step": 83015
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.230978388261216e-05,
+      "loss": 0.0731,
+      "step": 83020
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.2309315081337023e-05,
+      "loss": 0.1185,
+      "step": 83025
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.2308846280061883e-05,
+      "loss": 0.089,
+      "step": 83030
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.2308377478786743e-05,
+      "loss": 0.1278,
+      "step": 83035
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.2307908677511605e-05,
+      "loss": 0.1421,
+      "step": 83040
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.2307439876236465e-05,
+      "loss": 0.2779,
+      "step": 83045
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.2306971074961325e-05,
+      "loss": 0.0081,
+      "step": 83050
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.2306502273686185e-05,
+      "loss": 0.0025,
+      "step": 83055
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.2306033472411045e-05,
+      "loss": 0.0285,
+      "step": 83060
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.2305564671135906e-05,
+      "loss": 0.0414,
+      "step": 83065
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.2305095869860768e-05,
+      "loss": 0.0701,
+      "step": 83070
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.2304627068585628e-05,
+      "loss": 0.069,
+      "step": 83075
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.230415826731049e-05,
+      "loss": 0.0509,
+      "step": 83080
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.230368946603535e-05,
+      "loss": 0.1334,
+      "step": 83085
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.230322066476021e-05,
+      "loss": 0.1531,
+      "step": 83090
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.230275186348507e-05,
+      "loss": 0.227,
+      "step": 83095
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.230228306220993e-05,
+      "loss": 0.0466,
+      "step": 83100
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.230181426093479e-05,
+      "loss": 0.0232,
+      "step": 83105
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.230134545965965e-05,
+      "loss": 0.0455,
+      "step": 83110
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.230087665838451e-05,
+      "loss": 0.015,
+      "step": 83115
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.2300407857109374e-05,
+      "loss": 0.0395,
+      "step": 83120
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.2299939055834234e-05,
+      "loss": 0.1822,
+      "step": 83125
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.2299470254559094e-05,
+      "loss": 0.0729,
+      "step": 83130
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.2299001453283954e-05,
+      "loss": 0.0764,
+      "step": 83135
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.2298532652008814e-05,
+      "loss": 0.1371,
+      "step": 83140
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.2298063850733675e-05,
+      "loss": 0.3928,
+      "step": 83145
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.2297595049458535e-05,
+      "loss": 0.0693,
+      "step": 83150
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.2297126248183395e-05,
+      "loss": 0.0335,
+      "step": 83155
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.2296657446908255e-05,
+      "loss": 0.007,
+      "step": 83160
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.2296188645633118e-05,
+      "loss": 0.027,
+      "step": 83165
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.2295719844357978e-05,
+      "loss": 0.0918,
+      "step": 83170
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.2295251043082838e-05,
+      "loss": 0.0589,
+      "step": 83175
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.2294782241807698e-05,
+      "loss": 0.1043,
+      "step": 83180
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.229431344053256e-05,
+      "loss": 0.0707,
+      "step": 83185
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.229384463925742e-05,
+      "loss": 0.2569,
+      "step": 83190
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.229337583798228e-05,
+      "loss": 0.2281,
+      "step": 83195
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.229290703670714e-05,
+      "loss": 0.0738,
+      "step": 83200
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.2292438235432e-05,
+      "loss": 0.0241,
+      "step": 83205
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.2291969434156863e-05,
+      "loss": 0.0525,
+      "step": 83210
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.2291500632881723e-05,
+      "loss": 0.0872,
+      "step": 83215
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.2291031831606583e-05,
+      "loss": 0.0468,
+      "step": 83220
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.2290563030331444e-05,
+      "loss": 0.0272,
+      "step": 83225
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.2290094229056304e-05,
+      "loss": 0.085,
+      "step": 83230
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.2289625427781164e-05,
+      "loss": 0.1253,
+      "step": 83235
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.2289156626506024e-05,
+      "loss": 0.235,
+      "step": 83240
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.2288687825230884e-05,
+      "loss": 0.2809,
+      "step": 83245
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.2288219023955746e-05,
+      "loss": 0.0401,
+      "step": 83250
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.2287750222680607e-05,
+      "loss": 0.0347,
+      "step": 83255
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.2287281421405467e-05,
+      "loss": 0.056,
+      "step": 83260
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.2286812620130329e-05,
+      "loss": 0.0216,
+      "step": 83265
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.2286343818855189e-05,
+      "loss": 0.0674,
+      "step": 83270
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.2285875017580049e-05,
+      "loss": 0.0733,
+      "step": 83275
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.2285406216304909e-05,
+      "loss": 0.0262,
+      "step": 83280
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.2284937415029769e-05,
+      "loss": 0.1099,
+      "step": 83285
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.228446861375463e-05,
+      "loss": 0.1443,
+      "step": 83290
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.228399981247949e-05,
+      "loss": 0.1946,
+      "step": 83295
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.228353101120435e-05,
+      "loss": 0.0785,
+      "step": 83300
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.2283062209929214e-05,
+      "loss": 0.033,
+      "step": 83305
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.2282593408654074e-05,
+      "loss": 0.0262,
+      "step": 83310
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.2282124607378933e-05,
+      "loss": 0.0774,
+      "step": 83315
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.2281655806103793e-05,
+      "loss": 0.0314,
+      "step": 83320
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.2281187004828653e-05,
+      "loss": 0.0695,
+      "step": 83325
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.2280718203553515e-05,
+      "loss": 0.0794,
+      "step": 83330
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.2280249402278375e-05,
+      "loss": 0.0691,
+      "step": 83335
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.2279780601003235e-05,
+      "loss": 0.0894,
+      "step": 83340
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.2279311799728095e-05,
+      "loss": 0.2373,
+      "step": 83345
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.2278842998452958e-05,
+      "loss": 0.0555,
+      "step": 83350
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.2278374197177818e-05,
+      "loss": 0.0623,
+      "step": 83355
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.2277905395902678e-05,
+      "loss": 0.1059,
+      "step": 83360
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.2277436594627538e-05,
+      "loss": 0.1103,
+      "step": 83365
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.22769677933524e-05,
+      "loss": 0.0721,
+      "step": 83370
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.227649899207726e-05,
+      "loss": 0.0832,
+      "step": 83375
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.227603019080212e-05,
+      "loss": 0.0922,
+      "step": 83380
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.227556138952698e-05,
+      "loss": 0.1668,
+      "step": 83385
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.2275092588251841e-05,
+      "loss": 0.1856,
+      "step": 83390
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.2274623786976703e-05,
+      "loss": 0.2563,
+      "step": 83395
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.2274154985701562e-05,
+      "loss": 0.0489,
+      "step": 83400
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.2273686184426422e-05,
+      "loss": 0.0425,
+      "step": 83405
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.2273217383151284e-05,
+      "loss": 0.0183,
+      "step": 83410
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.2272748581876144e-05,
+      "loss": 0.028,
+      "step": 83415
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.2272279780601004e-05,
+      "loss": 0.074,
+      "step": 83420
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.2271810979325864e-05,
+      "loss": 0.1143,
+      "step": 83425
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.2271342178050725e-05,
+      "loss": 0.063,
+      "step": 83430
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.2270873376775585e-05,
+      "loss": 0.1443,
+      "step": 83435
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.2270404575500445e-05,
+      "loss": 0.1416,
+      "step": 83440
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.2269935774225307e-05,
+      "loss": 0.3292,
+      "step": 83445
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.2269466972950169e-05,
+      "loss": 0.0906,
+      "step": 83450
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.2268998171675029e-05,
+      "loss": 0.025,
+      "step": 83455
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.2268529370399888e-05,
+      "loss": 0.0695,
+      "step": 83460
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.2268060569124748e-05,
+      "loss": 0.0739,
+      "step": 83465
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.226759176784961e-05,
+      "loss": 0.0687,
+      "step": 83470
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.226712296657447e-05,
+      "loss": 0.1008,
+      "step": 83475
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.226665416529933e-05,
+      "loss": 0.1099,
+      "step": 83480
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.226618536402419e-05,
+      "loss": 0.1334,
+      "step": 83485
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.2265716562749053e-05,
+      "loss": 0.2016,
+      "step": 83490
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.2265247761473913e-05,
+      "loss": 0.3243,
+      "step": 83495
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.2264778960198773e-05,
+      "loss": 0.0194,
+      "step": 83500
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.2264310158923633e-05,
+      "loss": 0.0295,
+      "step": 83505
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.2263841357648495e-05,
+      "loss": 0.0918,
+      "step": 83510
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.2263372556373354e-05,
+      "loss": 0.1133,
+      "step": 83515
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.2262903755098214e-05,
+      "loss": 0.039,
+      "step": 83520
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.2262434953823074e-05,
+      "loss": 0.0482,
+      "step": 83525
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.2261966152547934e-05,
+      "loss": 0.0764,
+      "step": 83530
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.2261497351272798e-05,
+      "loss": 0.1765,
+      "step": 83535
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.2261028549997658e-05,
+      "loss": 0.152,
+      "step": 83540
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.2260559748722517e-05,
+      "loss": 0.294,
+      "step": 83545
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.2260090947447379e-05,
+      "loss": 0.0708,
+      "step": 83550
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.2259622146172239e-05,
+      "loss": 0.0091,
+      "step": 83555
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.2259153344897099e-05,
+      "loss": 0.0575,
+      "step": 83560
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.2258684543621959e-05,
+      "loss": 0.0657,
+      "step": 83565
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.2258215742346819e-05,
+      "loss": 0.0315,
+      "step": 83570
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.225774694107168e-05,
+      "loss": 0.2276,
+      "step": 83575
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.2257278139796542e-05,
+      "loss": 0.0769,
+      "step": 83580
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.2256809338521402e-05,
+      "loss": 0.2337,
+      "step": 83585
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.2256340537246264e-05,
+      "loss": 0.1424,
+      "step": 83590
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.2255871735971124e-05,
+      "loss": 0.2379,
+      "step": 83595
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.2255402934695984e-05,
+      "loss": 0.0489,
+      "step": 83600
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.2254934133420843e-05,
+      "loss": 0.0474,
+      "step": 83605
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.2254465332145703e-05,
+      "loss": 0.0358,
+      "step": 83610
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.2253996530870565e-05,
+      "loss": 0.044,
+      "step": 83615
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.2253527729595425e-05,
+      "loss": 0.0684,
+      "step": 83620
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.2253058928320285e-05,
+      "loss": 0.0767,
+      "step": 83625
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.2252590127045148e-05,
+      "loss": 0.1371,
+      "step": 83630
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.2252121325770008e-05,
+      "loss": 0.1474,
+      "step": 83635
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.2251652524494868e-05,
+      "loss": 0.1975,
+      "step": 83640
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.2251183723219728e-05,
+      "loss": 0.1976,
+      "step": 83645
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.2250714921944588e-05,
+      "loss": 0.0165,
+      "step": 83650
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.225024612066945e-05,
+      "loss": 0.0297,
+      "step": 83655
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.224977731939431e-05,
+      "loss": 0.1577,
+      "step": 83660
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.224930851811917e-05,
+      "loss": 0.0528,
+      "step": 83665
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.224883971684403e-05,
+      "loss": 0.0448,
+      "step": 83670
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.2248370915568893e-05,
+      "loss": 0.0638,
+      "step": 83675
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.2247902114293753e-05,
+      "loss": 0.1076,
+      "step": 83680
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.2247433313018613e-05,
+      "loss": 0.1727,
+      "step": 83685
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2246964511743473e-05,
+      "loss": 0.1929,
+      "step": 83690
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2246495710468334e-05,
+      "loss": 0.2969,
+      "step": 83695
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2246026909193194e-05,
+      "loss": 0.0466,
+      "step": 83700
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2245558107918054e-05,
+      "loss": 0.048,
+      "step": 83705
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2245089306642914e-05,
+      "loss": 0.0381,
+      "step": 83710
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2244620505367774e-05,
+      "loss": 0.0691,
+      "step": 83715
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2244151704092637e-05,
+      "loss": 0.1224,
+      "step": 83720
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2243682902817497e-05,
+      "loss": 0.048,
+      "step": 83725
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2243214101542357e-05,
+      "loss": 0.1117,
+      "step": 83730
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2242745300267219e-05,
+      "loss": 0.1337,
+      "step": 83735
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2242276498992079e-05,
+      "loss": 0.3771,
+      "step": 83740
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2241807697716939e-05,
+      "loss": 0.3642,
+      "step": 83745
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2241338896441798e-05,
+      "loss": 0.0426,
+      "step": 83750
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2240870095166658e-05,
+      "loss": 0.0917,
+      "step": 83755
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.224040129389152e-05,
+      "loss": 0.0423,
+      "step": 83760
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.223993249261638e-05,
+      "loss": 0.0592,
+      "step": 83765
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2239463691341242e-05,
+      "loss": 0.0689,
+      "step": 83770
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2238994890066103e-05,
+      "loss": 0.0804,
+      "step": 83775
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2238526088790963e-05,
+      "loss": 0.1061,
+      "step": 83780
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2238057287515823e-05,
+      "loss": 0.1218,
+      "step": 83785
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2237588486240683e-05,
+      "loss": 0.1276,
+      "step": 83790
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2237119684965543e-05,
+      "loss": 0.2179,
+      "step": 83795
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2236650883690405e-05,
+      "loss": 0.0722,
+      "step": 83800
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2236182082415265e-05,
+      "loss": 0.0804,
+      "step": 83805
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2235713281140124e-05,
+      "loss": 0.0458,
+      "step": 83810
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2235244479864988e-05,
+      "loss": 0.067,
+      "step": 83815
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2234775678589848e-05,
+      "loss": 0.0403,
+      "step": 83820
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2234306877314708e-05,
+      "loss": 0.0922,
+      "step": 83825
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2233838076039568e-05,
+      "loss": 0.1323,
+      "step": 83830
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.223336927476443e-05,
+      "loss": 0.1212,
+      "step": 83835
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2232900473489289e-05,
+      "loss": 0.2292,
+      "step": 83840
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2232431672214149e-05,
+      "loss": 0.4067,
+      "step": 83845
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2231962870939009e-05,
+      "loss": 0.0513,
+      "step": 83850
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2231494069663869e-05,
+      "loss": 0.0135,
+      "step": 83855
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2231025268388732e-05,
+      "loss": 0.0269,
+      "step": 83860
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2230556467113592e-05,
+      "loss": 0.0542,
+      "step": 83865
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2230087665838452e-05,
+      "loss": 0.0683,
+      "step": 83870
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2229618864563314e-05,
+      "loss": 0.0926,
+      "step": 83875
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2229150063288174e-05,
+      "loss": 0.1678,
+      "step": 83880
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2228681262013034e-05,
+      "loss": 0.0925,
+      "step": 83885
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2228212460737894e-05,
+      "loss": 0.2065,
+      "step": 83890
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2227743659462753e-05,
+      "loss": 0.411,
+      "step": 83895
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2227274858187615e-05,
+      "loss": 0.0139,
+      "step": 83900
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.2226806056912477e-05,
+      "loss": 0.0444,
+      "step": 83905
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.2226337255637337e-05,
+      "loss": 0.0592,
+      "step": 83910
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.2225868454362198e-05,
+      "loss": 0.0458,
+      "step": 83915
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.2225399653087058e-05,
+      "loss": 0.0628,
+      "step": 83920
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.2224930851811918e-05,
+      "loss": 0.0943,
+      "step": 83925
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.2224462050536778e-05,
+      "loss": 0.0753,
+      "step": 83930
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.2223993249261638e-05,
+      "loss": 0.2237,
+      "step": 83935
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.22235244479865e-05,
+      "loss": 0.1878,
+      "step": 83940
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.222305564671136e-05,
+      "loss": 0.3213,
+      "step": 83945
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.222258684543622e-05,
+      "loss": 0.0989,
+      "step": 83950
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.2222118044161083e-05,
+      "loss": 0.033,
+      "step": 83955
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.2221649242885943e-05,
+      "loss": 0.0802,
+      "step": 83960
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.2221180441610803e-05,
+      "loss": 0.0214,
+      "step": 83965
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.2220711640335663e-05,
+      "loss": 0.0314,
+      "step": 83970
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.2220242839060523e-05,
+      "loss": 0.1221,
+      "step": 83975
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.2219774037785384e-05,
+      "loss": 0.1475,
+      "step": 83980
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.2219305236510244e-05,
+      "loss": 0.1117,
+      "step": 83985
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.2218836435235104e-05,
+      "loss": 0.3094,
+      "step": 83990
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.2218367633959964e-05,
+      "loss": 0.2421,
+      "step": 83995
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.2217898832684827e-05,
+      "loss": 0.0479,
+      "step": 84000
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.2217430031409687e-05,
+      "loss": 0.0401,
+      "step": 84005
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.2216961230134547e-05,
+      "loss": 0.0075,
+      "step": 84010
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.2216492428859407e-05,
+      "loss": 0.0173,
+      "step": 84015
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.2216023627584269e-05,
+      "loss": 0.0428,
+      "step": 84020
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.2215554826309129e-05,
+      "loss": 0.0481,
+      "step": 84025
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.2215086025033989e-05,
+      "loss": 0.0569,
+      "step": 84030
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.2214617223758849e-05,
+      "loss": 0.2363,
+      "step": 84035
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.2214148422483709e-05,
+      "loss": 0.1255,
+      "step": 84040
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.2213679621208572e-05,
+      "loss": 0.2421,
+      "step": 84045
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.2213210819933432e-05,
+      "loss": 0.0725,
+      "step": 84050
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.2212742018658292e-05,
+      "loss": 0.0362,
+      "step": 84055
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.2212273217383153e-05,
+      "loss": 0.0547,
+      "step": 84060
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.2211804416108013e-05,
+      "loss": 0.0224,
+      "step": 84065
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.2211335614832873e-05,
+      "loss": 0.0658,
+      "step": 84070
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.2210866813557733e-05,
+      "loss": 0.0599,
+      "step": 84075
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.2210398012282593e-05,
+      "loss": 0.0865,
+      "step": 84080
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.2209929211007455e-05,
+      "loss": 0.1166,
+      "step": 84085
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.2209460409732315e-05,
+      "loss": 0.1242,
+      "step": 84090
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.2208991608457176e-05,
+      "loss": 0.1641,
+      "step": 84095
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.2208522807182038e-05,
+      "loss": 0.0467,
+      "step": 84100
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.2208054005906898e-05,
+      "loss": 0.0167,
+      "step": 84105
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.2207585204631758e-05,
+      "loss": 0.0463,
+      "step": 84110
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.2207116403356618e-05,
+      "loss": 0.0289,
+      "step": 84115
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.2206647602081478e-05,
+      "loss": 0.0876,
+      "step": 84120
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.220617880080634e-05,
+      "loss": 0.0997,
+      "step": 84125
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.22057099995312e-05,
+      "loss": 0.0724,
+      "step": 84130
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.2205241198256059e-05,
+      "loss": 0.1029,
+      "step": 84135
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.2204772396980922e-05,
+      "loss": 0.1901,
+      "step": 84140
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.2204303595705782e-05,
+      "loss": 0.1756,
+      "step": 84145
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.2203834794430642e-05,
+      "loss": 0.0648,
+      "step": 84150
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.2203365993155502e-05,
+      "loss": 0.0376,
+      "step": 84155
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.2202897191880362e-05,
+      "loss": 0.0418,
+      "step": 84160
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.2202428390605224e-05,
+      "loss": 0.0209,
+      "step": 84165
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.2201959589330084e-05,
+      "loss": 0.0336,
+      "step": 84170
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.2201490788054944e-05,
+      "loss": 0.0429,
+      "step": 84175
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.2201021986779804e-05,
+      "loss": 0.2253,
+      "step": 84180
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.2200553185504667e-05,
+      "loss": 0.0823,
+      "step": 84185
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.2200084384229527e-05,
+      "loss": 0.2294,
+      "step": 84190
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.2199615582954387e-05,
+      "loss": 0.3767,
+      "step": 84195
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.2199146781679247e-05,
+      "loss": 0.0257,
+      "step": 84200
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.2198677980404108e-05,
+      "loss": 0.0578,
+      "step": 84205
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.2198209179128968e-05,
+      "loss": 0.096,
+      "step": 84210
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.2197740377853828e-05,
+      "loss": 0.0353,
+      "step": 84215
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.2197271576578688e-05,
+      "loss": 0.017,
+      "step": 84220
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.219680277530355e-05,
+      "loss": 0.0899,
+      "step": 84225
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.2196333974028411e-05,
+      "loss": 0.1363,
+      "step": 84230
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.2195865172753271e-05,
+      "loss": 0.1552,
+      "step": 84235
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.2195396371478131e-05,
+      "loss": 0.3038,
+      "step": 84240
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.2194927570202993e-05,
+      "loss": 0.1944,
+      "step": 84245
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.2194458768927853e-05,
+      "loss": 0.083,
+      "step": 84250
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.2193989967652713e-05,
+      "loss": 0.0862,
+      "step": 84255
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.2193521166377573e-05,
+      "loss": 0.0484,
+      "step": 84260
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.2193052365102434e-05,
+      "loss": 0.0431,
+      "step": 84265
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.2192583563827294e-05,
+      "loss": 0.0566,
+      "step": 84270
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.2192114762552154e-05,
+      "loss": 0.0449,
+      "step": 84275
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.2191645961277016e-05,
+      "loss": 0.1187,
+      "step": 84280
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.2191177160001877e-05,
+      "loss": 0.0942,
+      "step": 84285
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.2190708358726737e-05,
+      "loss": 0.1475,
+      "step": 84290
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.2190239557451597e-05,
+      "loss": 0.1864,
+      "step": 84295
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.2189770756176457e-05,
+      "loss": 0.0969,
+      "step": 84300
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.2189301954901319e-05,
+      "loss": 0.0383,
+      "step": 84305
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.2188833153626179e-05,
+      "loss": 0.0172,
+      "step": 84310
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.2188364352351039e-05,
+      "loss": 0.036,
+      "step": 84315
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.2187895551075899e-05,
+      "loss": 0.0517,
+      "step": 84320
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.2187426749800762e-05,
+      "loss": 0.0623,
+      "step": 84325
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.2186957948525622e-05,
+      "loss": 0.1455,
+      "step": 84330
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.2186489147250482e-05,
+      "loss": 0.0808,
+      "step": 84335
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.2186020345975342e-05,
+      "loss": 0.2054,
+      "step": 84340
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.2185551544700203e-05,
+      "loss": 0.2095,
+      "step": 84345
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.2185082743425063e-05,
+      "loss": 0.0996,
+      "step": 84350
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.2184613942149923e-05,
+      "loss": 0.016,
+      "step": 84355
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.2184145140874783e-05,
+      "loss": 0.0279,
+      "step": 84360
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.2183676339599643e-05,
+      "loss": 0.0278,
+      "step": 84365
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.2183207538324506e-05,
+      "loss": 0.0713,
+      "step": 84370
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.2182738737049366e-05,
+      "loss": 0.1478,
+      "step": 84375
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.2182269935774226e-05,
+      "loss": 0.0615,
+      "step": 84380
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.2181801134499088e-05,
+      "loss": 0.0657,
+      "step": 84385
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.2181332333223948e-05,
+      "loss": 0.1012,
+      "step": 84390
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.2180863531948808e-05,
+      "loss": 0.2686,
+      "step": 84395
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.2180394730673668e-05,
+      "loss": 0.06,
+      "step": 84400
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.2179925929398528e-05,
+      "loss": 0.0194,
+      "step": 84405
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.217945712812339e-05,
+      "loss": 0.0553,
+      "step": 84410
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.217898832684825e-05,
+      "loss": 0.096,
+      "step": 84415
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.2178519525573111e-05,
+      "loss": 0.0314,
+      "step": 84420
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.2178050724297973e-05,
+      "loss": 0.0546,
+      "step": 84425
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.2177581923022832e-05,
+      "loss": 0.1234,
+      "step": 84430
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.2177113121747692e-05,
+      "loss": 0.1008,
+      "step": 84435
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.2176644320472552e-05,
+      "loss": 0.0997,
+      "step": 84440
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.2176175519197412e-05,
+      "loss": 0.2987,
+      "step": 84445
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.2175706717922274e-05,
+      "loss": 0.0427,
+      "step": 84450
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.2175237916647134e-05,
+      "loss": 0.0185,
+      "step": 84455
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.2174769115371994e-05,
+      "loss": 0.0178,
+      "step": 84460
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.2174300314096857e-05,
+      "loss": 0.0259,
+      "step": 84465
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.2173831512821717e-05,
+      "loss": 0.0739,
+      "step": 84470
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.2173362711546577e-05,
+      "loss": 0.0694,
+      "step": 84475
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.2172893910271437e-05,
+      "loss": 0.1191,
+      "step": 84480
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.2172425108996297e-05,
+      "loss": 0.1201,
+      "step": 84485
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.2171956307721158e-05,
+      "loss": 0.1859,
+      "step": 84490
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.2171487506446018e-05,
+      "loss": 0.2759,
+      "step": 84495
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.2171018705170878e-05,
+      "loss": 0.0926,
+      "step": 84500
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.2170549903895738e-05,
+      "loss": 0.0321,
+      "step": 84505
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.2170081102620602e-05,
+      "loss": 0.0295,
+      "step": 84510
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.2169612301345462e-05,
+      "loss": 0.0777,
+      "step": 84515
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.2169143500070321e-05,
+      "loss": 0.0552,
+      "step": 84520
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.2168674698795181e-05,
+      "loss": 0.1077,
+      "step": 84525
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.2168205897520043e-05,
+      "loss": 0.1633,
+      "step": 84530
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.2167737096244903e-05,
+      "loss": 0.1222,
+      "step": 84535
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.2167268294969763e-05,
+      "loss": 0.1673,
+      "step": 84540
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.2166799493694623e-05,
+      "loss": 0.2728,
+      "step": 84545
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.2166330692419483e-05,
+      "loss": 0.0829,
+      "step": 84550
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.2165861891144346e-05,
+      "loss": 0.0173,
+      "step": 84555
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.2165393089869206e-05,
+      "loss": 0.0413,
+      "step": 84560
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.2164924288594066e-05,
+      "loss": 0.0687,
+      "step": 84565
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.2164455487318928e-05,
+      "loss": 0.0593,
+      "step": 84570
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.2163986686043787e-05,
+      "loss": 0.087,
+      "step": 84575
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.2163517884768647e-05,
+      "loss": 0.1026,
+      "step": 84580
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.2163049083493507e-05,
+      "loss": 0.102,
+      "step": 84585
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.2162580282218367e-05,
+      "loss": 0.2065,
+      "step": 84590
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.2162111480943229e-05,
+      "loss": 0.1685,
+      "step": 84595
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.2161642679668089e-05,
+      "loss": 0.1001,
+      "step": 84600
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.216117387839295e-05,
+      "loss": 0.0342,
+      "step": 84605
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.2160705077117812e-05,
+      "loss": 0.0525,
+      "step": 84610
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.2160236275842672e-05,
+      "loss": 0.0481,
+      "step": 84615
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.2159767474567532e-05,
+      "loss": 0.149,
+      "step": 84620
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.2159298673292392e-05,
+      "loss": 0.1458,
+      "step": 84625
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.2158829872017252e-05,
+      "loss": 0.1162,
+      "step": 84630
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.2158361070742113e-05,
+      "loss": 0.1173,
+      "step": 84635
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.2157892269466973e-05,
+      "loss": 0.1814,
+      "step": 84640
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.2157423468191833e-05,
+      "loss": 0.2595,
+      "step": 84645
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.2156954666916697e-05,
+      "loss": 0.0722,
+      "step": 84650
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.2156485865641557e-05,
+      "loss": 0.039,
+      "step": 84655
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.2156017064366417e-05,
+      "loss": 0.0383,
+      "step": 84660
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.2155548263091276e-05,
+      "loss": 0.1506,
+      "step": 84665
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.2155079461816136e-05,
+      "loss": 0.0744,
+      "step": 84670
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.2154610660540998e-05,
+      "loss": 0.0485,
+      "step": 84675
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.2154141859265858e-05,
+      "loss": 0.0927,
+      "step": 84680
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.2153673057990718e-05,
+      "loss": 0.1012,
+      "step": 84685
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.2153204256715578e-05,
+      "loss": 0.128,
+      "step": 84690
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.2152735455440441e-05,
+      "loss": 0.2211,
+      "step": 84695
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.2152266654165301e-05,
+      "loss": 0.098,
+      "step": 84700
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.2151797852890161e-05,
+      "loss": 0.0205,
+      "step": 84705
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.2151329051615021e-05,
+      "loss": 0.0545,
+      "step": 84710
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.2150860250339883e-05,
+      "loss": 0.0574,
+      "step": 84715
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.2150391449064743e-05,
+      "loss": 0.1062,
+      "step": 84720
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.2149922647789602e-05,
+      "loss": 0.034,
+      "step": 84725
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.2149453846514462e-05,
+      "loss": 0.0577,
+      "step": 84730
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.2148985045239324e-05,
+      "loss": 0.0859,
+      "step": 84735
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.2148516243964184e-05,
+      "loss": 0.1521,
+      "step": 84740
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.2148047442689046e-05,
+      "loss": 0.2548,
+      "step": 84745
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.2147578641413905e-05,
+      "loss": 0.0839,
+      "step": 84750
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.2147109840138767e-05,
+      "loss": 0.0285,
+      "step": 84755
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.2146641038863627e-05,
+      "loss": 0.0183,
+      "step": 84760
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.2146172237588487e-05,
+      "loss": 0.0776,
+      "step": 84765
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.2145703436313347e-05,
+      "loss": 0.0921,
+      "step": 84770
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.2145234635038209e-05,
+      "loss": 0.0571,
+      "step": 84775
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.2144765833763068e-05,
+      "loss": 0.1153,
+      "step": 84780
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.2144297032487928e-05,
+      "loss": 0.1005,
+      "step": 84785
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.214382823121279e-05,
+      "loss": 0.1477,
+      "step": 84790
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.2143359429937652e-05,
+      "loss": 0.2054,
+      "step": 84795
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.2142890628662512e-05,
+      "loss": 0.0746,
+      "step": 84800
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.2142421827387372e-05,
+      "loss": 0.0249,
+      "step": 84805
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.2141953026112231e-05,
+      "loss": 0.0197,
+      "step": 84810
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.2141484224837093e-05,
+      "loss": 0.0673,
+      "step": 84815
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.2141015423561953e-05,
+      "loss": 0.091,
+      "step": 84820
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.2140546622286813e-05,
+      "loss": 0.042,
+      "step": 84825
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.2140077821011673e-05,
+      "loss": 0.0678,
+      "step": 84830
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.2139609019736536e-05,
+      "loss": 0.1079,
+      "step": 84835
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.2139140218461396e-05,
+      "loss": 0.1308,
+      "step": 84840
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.2138671417186256e-05,
+      "loss": 0.1827,
+      "step": 84845
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.2138202615911116e-05,
+      "loss": 0.0779,
+      "step": 84850
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.2137733814635978e-05,
+      "loss": 0.0242,
+      "step": 84855
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.2137265013360838e-05,
+      "loss": 0.0863,
+      "step": 84860
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.2136796212085698e-05,
+      "loss": 0.0686,
+      "step": 84865
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.2136327410810557e-05,
+      "loss": 0.0695,
+      "step": 84870
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.2135858609535417e-05,
+      "loss": 0.0646,
+      "step": 84875
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.213538980826028e-05,
+      "loss": 0.0794,
+      "step": 84880
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.213492100698514e-05,
+      "loss": 0.0826,
+      "step": 84885
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.213445220571e-05,
+      "loss": 0.208,
+      "step": 84890
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.2133983404434862e-05,
+      "loss": 0.407,
+      "step": 84895
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.2133514603159722e-05,
+      "loss": 0.0326,
+      "step": 84900
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.2133045801884582e-05,
+      "loss": 0.0356,
+      "step": 84905
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.2132577000609442e-05,
+      "loss": 0.0131,
+      "step": 84910
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.2132108199334302e-05,
+      "loss": 0.0433,
+      "step": 84915
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.2131639398059164e-05,
+      "loss": 0.0459,
+      "step": 84920
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.2131170596784023e-05,
+      "loss": 0.0983,
+      "step": 84925
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.2130701795508885e-05,
+      "loss": 0.1348,
+      "step": 84930
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.2130232994233747e-05,
+      "loss": 0.1374,
+      "step": 84935
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.2129764192958607e-05,
+      "loss": 0.2749,
+      "step": 84940
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.2129295391683467e-05,
+      "loss": 0.2818,
+      "step": 84945
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.2128826590408327e-05,
+      "loss": 0.0536,
+      "step": 84950
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.2128357789133186e-05,
+      "loss": 0.0252,
+      "step": 84955
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.2127888987858048e-05,
+      "loss": 0.0497,
+      "step": 84960
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.2127420186582908e-05,
+      "loss": 0.0617,
+      "step": 84965
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.2126951385307768e-05,
+      "loss": 0.0436,
+      "step": 84970
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.2126482584032631e-05,
+      "loss": 0.073,
+      "step": 84975
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.2126013782757491e-05,
+      "loss": 0.1061,
+      "step": 84980
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.2125544981482351e-05,
+      "loss": 0.0889,
+      "step": 84985
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.2125076180207211e-05,
+      "loss": 0.1823,
+      "step": 84990
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.2124607378932071e-05,
+      "loss": 0.2849,
+      "step": 84995
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.2124138577656933e-05,
+      "loss": 0.0793,
+      "step": 85000
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.2123669776381793e-05,
+      "loss": 0.0345,
+      "step": 85005
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.2123200975106653e-05,
+      "loss": 0.0192,
+      "step": 85010
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.2122732173831512e-05,
+      "loss": 0.0744,
+      "step": 85015
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.2122263372556376e-05,
+      "loss": 0.034,
+      "step": 85020
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.2121794571281236e-05,
+      "loss": 0.06,
+      "step": 85025
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.2121325770006096e-05,
+      "loss": 0.0513,
+      "step": 85030
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.2120856968730956e-05,
+      "loss": 0.121,
+      "step": 85035
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.2120388167455817e-05,
+      "loss": 0.1783,
+      "step": 85040
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.2119919366180677e-05,
+      "loss": 0.2662,
+      "step": 85045
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.2119450564905537e-05,
+      "loss": 0.0381,
+      "step": 85050
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.2118981763630397e-05,
+      "loss": 0.0177,
+      "step": 85055
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.2118512962355257e-05,
+      "loss": 0.0451,
+      "step": 85060
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.2118044161080119e-05,
+      "loss": 0.0189,
+      "step": 85065
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.211757535980498e-05,
+      "loss": 0.0686,
+      "step": 85070
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.211710655852984e-05,
+      "loss": 0.0908,
+      "step": 85075
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.2116637757254702e-05,
+      "loss": 0.0855,
+      "step": 85080
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.2116168955979562e-05,
+      "loss": 0.1953,
+      "step": 85085
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.2115700154704422e-05,
+      "loss": 0.1753,
+      "step": 85090
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.2115231353429282e-05,
+      "loss": 0.2956,
+      "step": 85095
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.2114762552154142e-05,
+      "loss": 0.0359,
+      "step": 85100
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.2114293750879003e-05,
+      "loss": 0.0425,
+      "step": 85105
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.2113824949603863e-05,
+      "loss": 0.022,
+      "step": 85110
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.2113356148328725e-05,
+      "loss": 0.0516,
+      "step": 85115
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.2112887347053586e-05,
+      "loss": 0.0781,
+      "step": 85120
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.2112418545778446e-05,
+      "loss": 0.083,
+      "step": 85125
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.2111949744503306e-05,
+      "loss": 0.0752,
+      "step": 85130
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.2111480943228166e-05,
+      "loss": 0.0996,
+      "step": 85135
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.2111012141953026e-05,
+      "loss": 0.1423,
+      "step": 85140
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.2110543340677888e-05,
+      "loss": 0.2798,
+      "step": 85145
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.2110074539402748e-05,
+      "loss": 0.0164,
+      "step": 85150
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.2109605738127608e-05,
+      "loss": 0.0169,
+      "step": 85155
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.2109136936852471e-05,
+      "loss": 0.0546,
+      "step": 85160
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.210866813557733e-05,
+      "loss": 0.0847,
+      "step": 85165
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.210819933430219e-05,
+      "loss": 0.0182,
+      "step": 85170
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.210773053302705e-05,
+      "loss": 0.0705,
+      "step": 85175
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.210726173175191e-05,
+      "loss": 0.1267,
+      "step": 85180
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.2106792930476772e-05,
+      "loss": 0.165,
+      "step": 85185
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.2106324129201632e-05,
+      "loss": 0.2124,
+      "step": 85190
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.2105855327926492e-05,
+      "loss": 0.2334,
+      "step": 85195
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.2105386526651352e-05,
+      "loss": 0.0667,
+      "step": 85200
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.2104917725376215e-05,
+      "loss": 0.0356,
+      "step": 85205
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.2104448924101075e-05,
+      "loss": 0.0628,
+      "step": 85210
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.2103980122825935e-05,
+      "loss": 0.0778,
+      "step": 85215
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.2103511321550795e-05,
+      "loss": 0.0354,
+      "step": 85220
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.2103042520275657e-05,
+      "loss": 0.0464,
+      "step": 85225
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.2102573719000517e-05,
+      "loss": 0.0755,
+      "step": 85230
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.2102104917725377e-05,
+      "loss": 0.1194,
+      "step": 85235
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.2101636116450237e-05,
+      "loss": 0.1531,
+      "step": 85240
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.2101167315175098e-05,
+      "loss": 0.3248,
+      "step": 85245
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.2100698513899958e-05,
+      "loss": 0.1072,
+      "step": 85250
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.210022971262482e-05,
+      "loss": 0.016,
+      "step": 85255
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.2099760911349681e-05,
+      "loss": 0.0689,
+      "step": 85260
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.2099292110074541e-05,
+      "loss": 0.0442,
+      "step": 85265
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.2098823308799401e-05,
+      "loss": 0.0702,
+      "step": 85270
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.2098354507524261e-05,
+      "loss": 0.0641,
+      "step": 85275
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.2097885706249121e-05,
+      "loss": 0.1959,
+      "step": 85280
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.2097416904973983e-05,
+      "loss": 0.1061,
+      "step": 85285
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.2096948103698843e-05,
+      "loss": 0.1526,
+      "step": 85290
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.2096479302423703e-05,
+      "loss": 0.3181,
+      "step": 85295
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.2096010501148566e-05,
+      "loss": 0.0794,
+      "step": 85300
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.2095541699873426e-05,
+      "loss": 0.0377,
+      "step": 85305
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.2095072898598286e-05,
+      "loss": 0.0753,
+      "step": 85310
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.2094604097323146e-05,
+      "loss": 0.0725,
+      "step": 85315
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.2094135296048006e-05,
+      "loss": 0.0682,
+      "step": 85320
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.2093666494772867e-05,
+      "loss": 0.1105,
+      "step": 85325
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.2093197693497727e-05,
+      "loss": 0.091,
+      "step": 85330
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.2092728892222587e-05,
+      "loss": 0.1482,
+      "step": 85335
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.2092260090947447e-05,
+      "loss": 0.2113,
+      "step": 85340
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.209179128967231e-05,
+      "loss": 0.2973,
+      "step": 85345
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.209132248839717e-05,
+      "loss": 0.079,
+      "step": 85350
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.209085368712203e-05,
+      "loss": 0.0591,
+      "step": 85355
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.209038488584689e-05,
+      "loss": 0.0622,
+      "step": 85360
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.2089916084571752e-05,
+      "loss": 0.0578,
+      "step": 85365
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.2089447283296612e-05,
+      "loss": 0.0683,
+      "step": 85370
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.2088978482021472e-05,
+      "loss": 0.056,
+      "step": 85375
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.2088509680746332e-05,
+      "loss": 0.0709,
+      "step": 85380
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.2088040879471192e-05,
+      "loss": 0.2024,
+      "step": 85385
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.2087572078196053e-05,
+      "loss": 0.1626,
+      "step": 85390
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.2087103276920915e-05,
+      "loss": 0.2486,
+      "step": 85395
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.2086634475645775e-05,
+      "loss": 0.0634,
+      "step": 85400
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.2086165674370636e-05,
+      "loss": 0.0103,
+      "step": 85405
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.2085696873095496e-05,
+      "loss": 0.0332,
+      "step": 85410
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.2085228071820356e-05,
+      "loss": 0.0428,
+      "step": 85415
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.2084759270545216e-05,
+      "loss": 0.0711,
+      "step": 85420
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.2084290469270076e-05,
+      "loss": 0.0546,
+      "step": 85425
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.2083821667994938e-05,
+      "loss": 0.0558,
+      "step": 85430
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.2083352866719798e-05,
+      "loss": 0.1043,
+      "step": 85435
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.208288406544466e-05,
+      "loss": 0.1153,
+      "step": 85440
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.2082415264169521e-05,
+      "loss": 0.3828,
+      "step": 85445
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.2081946462894381e-05,
+      "loss": 0.0537,
+      "step": 85450
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.208147766161924e-05,
+      "loss": 0.0323,
+      "step": 85455
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.20810088603441e-05,
+      "loss": 0.0638,
+      "step": 85460
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.208054005906896e-05,
+      "loss": 0.0546,
+      "step": 85465
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.2080071257793822e-05,
+      "loss": 0.0969,
+      "step": 85470
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.2079602456518682e-05,
+      "loss": 0.0441,
+      "step": 85475
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.2079133655243542e-05,
+      "loss": 0.0426,
+      "step": 85480
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.2078664853968406e-05,
+      "loss": 0.207,
+      "step": 85485
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.2078196052693265e-05,
+      "loss": 0.1455,
+      "step": 85490
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.2077727251418125e-05,
+      "loss": 0.3497,
+      "step": 85495
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.2077258450142985e-05,
+      "loss": 0.0732,
+      "step": 85500
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.2076789648867845e-05,
+      "loss": 0.0068,
+      "step": 85505
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.2076320847592707e-05,
+      "loss": 0.0237,
+      "step": 85510
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.2075852046317567e-05,
+      "loss": 0.0273,
+      "step": 85515
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.2075383245042427e-05,
+      "loss": 0.0467,
+      "step": 85520
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.2074914443767287e-05,
+      "loss": 0.1015,
+      "step": 85525
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.2074445642492147e-05,
+      "loss": 0.08,
+      "step": 85530
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.207397684121701e-05,
+      "loss": 0.1203,
+      "step": 85535
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.207350803994187e-05,
+      "loss": 0.1713,
+      "step": 85540
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.207303923866673e-05,
+      "loss": 0.3559,
+      "step": 85545
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.2072570437391591e-05,
+      "loss": 0.0424,
+      "step": 85550
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.2072101636116451e-05,
+      "loss": 0.0211,
+      "step": 85555
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.2071632834841311e-05,
+      "loss": 0.0303,
+      "step": 85560
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.2071164033566171e-05,
+      "loss": 0.0616,
+      "step": 85565
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.2070695232291031e-05,
+      "loss": 0.0772,
+      "step": 85570
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.2070226431015893e-05,
+      "loss": 0.0717,
+      "step": 85575
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.2069757629740754e-05,
+      "loss": 0.104,
+      "step": 85580
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.2069288828465614e-05,
+      "loss": 0.1114,
+      "step": 85585
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.2068820027190476e-05,
+      "loss": 0.1281,
+      "step": 85590
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.2068351225915336e-05,
+      "loss": 0.3246,
+      "step": 85595
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.2067882424640196e-05,
+      "loss": 0.0502,
+      "step": 85600
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.2067413623365056e-05,
+      "loss": 0.0084,
+      "step": 85605
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.2066944822089917e-05,
+      "loss": 0.0513,
+      "step": 85610
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.2066476020814777e-05,
+      "loss": 0.0499,
+      "step": 85615
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.2066007219539637e-05,
+      "loss": 0.1073,
+      "step": 85620
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.2065538418264499e-05,
+      "loss": 0.1085,
+      "step": 85625
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.206506961698936e-05,
+      "loss": 0.0658,
+      "step": 85630
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.206460081571422e-05,
+      "loss": 0.1206,
+      "step": 85635
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.206413201443908e-05,
+      "loss": 0.2309,
+      "step": 85640
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.206366321316394e-05,
+      "loss": 0.2057,
+      "step": 85645
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.2063194411888802e-05,
+      "loss": 0.0522,
+      "step": 85650
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.2062725610613662e-05,
+      "loss": 0.0993,
+      "step": 85655
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.2062256809338522e-05,
+      "loss": 0.0346,
+      "step": 85660
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.2061788008063382e-05,
+      "loss": 0.0709,
+      "step": 85665
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.2061319206788245e-05,
+      "loss": 0.0602,
+      "step": 85670
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.2060850405513105e-05,
+      "loss": 0.0756,
+      "step": 85675
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.2060381604237965e-05,
+      "loss": 0.088,
+      "step": 85680
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.2059912802962825e-05,
+      "loss": 0.1051,
+      "step": 85685
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.2059444001687687e-05,
+      "loss": 0.2163,
+      "step": 85690
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.2058975200412546e-05,
+      "loss": 0.353,
+      "step": 85695
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.2058506399137406e-05,
+      "loss": 0.0459,
+      "step": 85700
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.2058037597862266e-05,
+      "loss": 0.0301,
+      "step": 85705
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.2057568796587126e-05,
+      "loss": 0.0574,
+      "step": 85710
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.2057099995311988e-05,
+      "loss": 0.0805,
+      "step": 85715
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.205663119403685e-05,
+      "loss": 0.1654,
+      "step": 85720
+    },
+    {
+      "epoch": 4.0,
+      "eval_cer": 0.01052703752799744,
+      "eval_loss": 0.09995592385530472,
+      "eval_runtime": 386.2,
+      "eval_samples_per_second": 49.327,
+      "eval_steps_per_second": 12.333,
+      "eval_wer": 0.08902480462113489,
+      "step": 85724
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.205616239276171e-05,
+      "loss": 0.4197,
+      "step": 85725
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.2055693591486571e-05,
+      "loss": 0.1284,
+      "step": 85730
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.2055224790211431e-05,
+      "loss": 0.0131,
+      "step": 85735
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.2054755988936291e-05,
+      "loss": 0.0433,
+      "step": 85740
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.2054287187661151e-05,
+      "loss": 0.0229,
+      "step": 85745
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.205381838638601e-05,
+      "loss": 0.0597,
+      "step": 85750
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.2053349585110872e-05,
+      "loss": 0.0687,
+      "step": 85755
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.2052880783835732e-05,
+      "loss": 0.0817,
+      "step": 85760
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.2052411982560594e-05,
+      "loss": 0.1638,
+      "step": 85765
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.2051943181285456e-05,
+      "loss": 0.1613,
+      "step": 85770
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.2051474380010316e-05,
+      "loss": 0.2871,
+      "step": 85775
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.2051005578735175e-05,
+      "loss": 0.0609,
+      "step": 85780
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.2050536777460035e-05,
+      "loss": 0.0114,
+      "step": 85785
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.2050067976184895e-05,
+      "loss": 0.0659,
+      "step": 85790
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.2049599174909757e-05,
+      "loss": 0.0619,
+      "step": 85795
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.2049130373634617e-05,
+      "loss": 0.0618,
+      "step": 85800
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.2048661572359477e-05,
+      "loss": 0.0857,
+      "step": 85805
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.204819277108434e-05,
+      "loss": 0.0764,
+      "step": 85810
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.20477239698092e-05,
+      "loss": 0.0716,
+      "step": 85815
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.204725516853406e-05,
+      "loss": 0.2912,
+      "step": 85820
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.204678636725892e-05,
+      "loss": 0.1991,
+      "step": 85825
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.204631756598378e-05,
+      "loss": 0.0563,
+      "step": 85830
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.2045848764708642e-05,
+      "loss": 0.0287,
+      "step": 85835
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.2045379963433501e-05,
+      "loss": 0.038,
+      "step": 85840
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.2044911162158361e-05,
+      "loss": 0.0949,
+      "step": 85845
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.2044442360883221e-05,
+      "loss": 0.0285,
+      "step": 85850
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.2043973559608081e-05,
+      "loss": 0.0395,
+      "step": 85855
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.2043504758332945e-05,
+      "loss": 0.116,
+      "step": 85860
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.2043035957057805e-05,
+      "loss": 0.1158,
+      "step": 85865
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.2042567155782664e-05,
+      "loss": 0.1392,
+      "step": 85870
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.2042098354507526e-05,
+      "loss": 0.1574,
+      "step": 85875
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.2041629553232386e-05,
+      "loss": 0.0739,
+      "step": 85880
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.2041160751957246e-05,
+      "loss": 0.0182,
+      "step": 85885
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.2040691950682106e-05,
+      "loss": 0.026,
+      "step": 85890
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.2040223149406966e-05,
+      "loss": 0.0572,
+      "step": 85895
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.2039754348131827e-05,
+      "loss": 0.0655,
+      "step": 85900
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.2039285546856689e-05,
+      "loss": 0.0803,
+      "step": 85905
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.2038816745581549e-05,
+      "loss": 0.1215,
+      "step": 85910
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.203834794430641e-05,
+      "loss": 0.2066,
+      "step": 85915
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.203787914303127e-05,
+      "loss": 0.1278,
+      "step": 85920
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.203741034175613e-05,
+      "loss": 0.4348,
+      "step": 85925
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.203694154048099e-05,
+      "loss": 0.0525,
+      "step": 85930
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.203647273920585e-05,
+      "loss": 0.0071,
+      "step": 85935
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.2036003937930712e-05,
+      "loss": 0.032,
+      "step": 85940
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.2035535136655572e-05,
+      "loss": 0.0231,
+      "step": 85945
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.2035066335380434e-05,
+      "loss": 0.0631,
+      "step": 85950
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.2034597534105295e-05,
+      "loss": 0.0885,
+      "step": 85955
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.2034128732830155e-05,
+      "loss": 0.0834,
+      "step": 85960
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.2033659931555015e-05,
+      "loss": 0.1309,
+      "step": 85965
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.2033191130279875e-05,
+      "loss": 0.1026,
+      "step": 85970
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.2032722329004735e-05,
+      "loss": 0.4151,
+      "step": 85975
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.2032253527729597e-05,
+      "loss": 0.1058,
+      "step": 85980
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.2031784726454456e-05,
+      "loss": 0.024,
+      "step": 85985
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.2031315925179316e-05,
+      "loss": 0.0085,
+      "step": 85990
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.203084712390418e-05,
+      "loss": 0.0561,
+      "step": 85995
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.203037832262904e-05,
+      "loss": 0.0677,
+      "step": 86000
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.20299095213539e-05,
+      "loss": 0.0989,
+      "step": 86005
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.202944072007876e-05,
+      "loss": 0.0582,
+      "step": 86010
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.202897191880362e-05,
+      "loss": 0.1444,
+      "step": 86015
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.2028503117528481e-05,
+      "loss": 0.1608,
+      "step": 86020
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.2028034316253341e-05,
+      "loss": 0.3524,
+      "step": 86025
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.2027565514978201e-05,
+      "loss": 0.0627,
+      "step": 86030
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.2027096713703061e-05,
+      "loss": 0.0291,
+      "step": 86035
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.2026627912427923e-05,
+      "loss": 0.0183,
+      "step": 86040
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.2026159111152784e-05,
+      "loss": 0.0537,
+      "step": 86045
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.2025690309877644e-05,
+      "loss": 0.0536,
+      "step": 86050
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.2025221508602504e-05,
+      "loss": 0.0536,
+      "step": 86055
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.2024752707327366e-05,
+      "loss": 0.1036,
+      "step": 86060
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.2024283906052226e-05,
+      "loss": 0.0556,
+      "step": 86065
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.2023815104777086e-05,
+      "loss": 0.2033,
+      "step": 86070
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.2023346303501945e-05,
+      "loss": 0.2803,
+      "step": 86075
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.2022877502226807e-05,
+      "loss": 0.0971,
+      "step": 86080
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.2022408700951667e-05,
+      "loss": 0.0376,
+      "step": 86085
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.2021939899676529e-05,
+      "loss": 0.0532,
+      "step": 86090
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.2021471098401389e-05,
+      "loss": 0.0816,
+      "step": 86095
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.202100229712625e-05,
+      "loss": 0.0376,
+      "step": 86100
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.202053349585111e-05,
+      "loss": 0.0546,
+      "step": 86105
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.202006469457597e-05,
+      "loss": 0.0291,
+      "step": 86110
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.201959589330083e-05,
+      "loss": 0.0721,
+      "step": 86115
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.2019127092025692e-05,
+      "loss": 0.1596,
+      "step": 86120
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.2018658290750552e-05,
+      "loss": 0.1853,
+      "step": 86125
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.2018189489475412e-05,
+      "loss": 0.0676,
+      "step": 86130
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.2017720688200273e-05,
+      "loss": 0.017,
+      "step": 86135
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.2017251886925135e-05,
+      "loss": 0.0247,
+      "step": 86140
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.2016783085649995e-05,
+      "loss": 0.023,
+      "step": 86145
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.2016314284374855e-05,
+      "loss": 0.0669,
+      "step": 86150
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.2015845483099715e-05,
+      "loss": 0.0811,
+      "step": 86155
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.2015376681824576e-05,
+      "loss": 0.147,
+      "step": 86160
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.2014907880549436e-05,
+      "loss": 0.1308,
+      "step": 86165
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.2014439079274296e-05,
+      "loss": 0.1823,
+      "step": 86170
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.2013970277999156e-05,
+      "loss": 0.2708,
+      "step": 86175
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.2013501476724016e-05,
+      "loss": 0.0748,
+      "step": 86180
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.201303267544888e-05,
+      "loss": 0.0506,
+      "step": 86185
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.201256387417374e-05,
+      "loss": 0.0432,
+      "step": 86190
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.2012095072898599e-05,
+      "loss": 0.0347,
+      "step": 86195
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.201162627162346e-05,
+      "loss": 0.0965,
+      "step": 86200
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.201115747034832e-05,
+      "loss": 0.0806,
+      "step": 86205
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.201068866907318e-05,
+      "loss": 0.0727,
+      "step": 86210
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.201021986779804e-05,
+      "loss": 0.0547,
+      "step": 86215
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.20097510665229e-05,
+      "loss": 0.1392,
+      "step": 86220
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.2009282265247762e-05,
+      "loss": 0.3181,
+      "step": 86225
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.2008813463972624e-05,
+      "loss": 0.0705,
+      "step": 86230
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.2008344662697484e-05,
+      "loss": 0.0555,
+      "step": 86235
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.2007875861422345e-05,
+      "loss": 0.0531,
+      "step": 86240
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.2007407060147205e-05,
+      "loss": 0.0358,
+      "step": 86245
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.2006938258872065e-05,
+      "loss": 0.0923,
+      "step": 86250
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.2006469457596925e-05,
+      "loss": 0.0677,
+      "step": 86255
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.2006000656321785e-05,
+      "loss": 0.0699,
+      "step": 86260
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.2005531855046647e-05,
+      "loss": 0.1537,
+      "step": 86265
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.2005063053771507e-05,
+      "loss": 0.195,
+      "step": 86270
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.2004594252496368e-05,
+      "loss": 0.2251,
+      "step": 86275
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.200412545122123e-05,
+      "loss": 0.0748,
+      "step": 86280
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.200365664994609e-05,
+      "loss": 0.0121,
+      "step": 86285
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.200318784867095e-05,
+      "loss": 0.0332,
+      "step": 86290
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.200271904739581e-05,
+      "loss": 0.0416,
+      "step": 86295
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.200225024612067e-05,
+      "loss": 0.0202,
+      "step": 86300
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.2001781444845531e-05,
+      "loss": 0.0398,
+      "step": 86305
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.2001312643570391e-05,
+      "loss": 0.041,
+      "step": 86310
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.2000843842295251e-05,
+      "loss": 0.1806,
+      "step": 86315
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.2000375041020114e-05,
+      "loss": 0.1158,
+      "step": 86320
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.1999906239744974e-05,
+      "loss": 0.1668,
+      "step": 86325
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.1999437438469834e-05,
+      "loss": 0.0735,
+      "step": 86330
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.1998968637194694e-05,
+      "loss": 0.0184,
+      "step": 86335
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.1998499835919554e-05,
+      "loss": 0.0445,
+      "step": 86340
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.1998031034644416e-05,
+      "loss": 0.023,
+      "step": 86345
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.1997562233369276e-05,
+      "loss": 0.0438,
+      "step": 86350
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.1997093432094136e-05,
+      "loss": 0.0444,
+      "step": 86355
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.1996624630818996e-05,
+      "loss": 0.0491,
+      "step": 86360
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.1996155829543855e-05,
+      "loss": 0.1171,
+      "step": 86365
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.1995687028268719e-05,
+      "loss": 0.1871,
+      "step": 86370
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.1995218226993579e-05,
+      "loss": 0.2566,
+      "step": 86375
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.1994749425718439e-05,
+      "loss": 0.0699,
+      "step": 86380
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.19942806244433e-05,
+      "loss": 0.0023,
+      "step": 86385
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.199381182316816e-05,
+      "loss": 0.0388,
+      "step": 86390
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.199334302189302e-05,
+      "loss": 0.065,
+      "step": 86395
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.199287422061788e-05,
+      "loss": 0.0478,
+      "step": 86400
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.199240541934274e-05,
+      "loss": 0.1402,
+      "step": 86405
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.1991936618067602e-05,
+      "loss": 0.0579,
+      "step": 86410
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.1991467816792463e-05,
+      "loss": 0.1218,
+      "step": 86415
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.1990999015517323e-05,
+      "loss": 0.1371,
+      "step": 86420
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.1990530214242185e-05,
+      "loss": 0.259,
+      "step": 86425
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.1990061412967045e-05,
+      "loss": 0.1506,
+      "step": 86430
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.1989592611691905e-05,
+      "loss": 0.0274,
+      "step": 86435
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.1989123810416765e-05,
+      "loss": 0.0322,
+      "step": 86440
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.1988655009141625e-05,
+      "loss": 0.0835,
+      "step": 86445
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.1988186207866486e-05,
+      "loss": 0.0417,
+      "step": 86450
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.1987717406591346e-05,
+      "loss": 0.0588,
+      "step": 86455
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.1987248605316208e-05,
+      "loss": 0.089,
+      "step": 86460
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.198677980404107e-05,
+      "loss": 0.1077,
+      "step": 86465
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.198631100276593e-05,
+      "loss": 0.1568,
+      "step": 86470
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.198584220149079e-05,
+      "loss": 0.4361,
+      "step": 86475
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.198537340021565e-05,
+      "loss": 0.1228,
+      "step": 86480
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.1984904598940509e-05,
+      "loss": 0.0356,
+      "step": 86485
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.198443579766537e-05,
+      "loss": 0.01,
+      "step": 86490
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.198396699639023e-05,
+      "loss": 0.0745,
+      "step": 86495
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.198349819511509e-05,
+      "loss": 0.0558,
+      "step": 86500
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.198302939383995e-05,
+      "loss": 0.0393,
+      "step": 86505
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.1982560592564814e-05,
+      "loss": 0.1151,
+      "step": 86510
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.1982091791289674e-05,
+      "loss": 0.1022,
+      "step": 86515
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.1981622990014534e-05,
+      "loss": 0.0857,
+      "step": 86520
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.1981154188739394e-05,
+      "loss": 0.3631,
+      "step": 86525
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.1980685387464255e-05,
+      "loss": 0.1152,
+      "step": 86530
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.1980216586189115e-05,
+      "loss": 0.0289,
+      "step": 86535
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.1979747784913975e-05,
+      "loss": 0.0444,
+      "step": 86540
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.1979278983638835e-05,
+      "loss": 0.0932,
+      "step": 86545
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.1978810182363697e-05,
+      "loss": 0.0505,
+      "step": 86550
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.1978341381088558e-05,
+      "loss": 0.0459,
+      "step": 86555
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.1977872579813418e-05,
+      "loss": 0.174,
+      "step": 86560
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.1977403778538278e-05,
+      "loss": 0.1497,
+      "step": 86565
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.197693497726314e-05,
+      "loss": 0.1939,
+      "step": 86570
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.1976466175988e-05,
+      "loss": 0.2241,
+      "step": 86575
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.197599737471286e-05,
+      "loss": 0.0899,
+      "step": 86580
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.197552857343772e-05,
+      "loss": 0.0443,
+      "step": 86585
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.1975059772162581e-05,
+      "loss": 0.0571,
+      "step": 86590
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.1974590970887441e-05,
+      "loss": 0.0522,
+      "step": 86595
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.1974122169612303e-05,
+      "loss": 0.0935,
+      "step": 86600
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.1973653368337163e-05,
+      "loss": 0.0466,
+      "step": 86605
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.1973184567062024e-05,
+      "loss": 0.117,
+      "step": 86610
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.1972715765786884e-05,
+      "loss": 0.1251,
+      "step": 86615
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.1972246964511744e-05,
+      "loss": 0.1509,
+      "step": 86620
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.1971778163236604e-05,
+      "loss": 0.1627,
+      "step": 86625
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.1971309361961466e-05,
+      "loss": 0.0744,
+      "step": 86630
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.1970840560686326e-05,
+      "loss": 0.043,
+      "step": 86635
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.1970371759411186e-05,
+      "loss": 0.0451,
+      "step": 86640
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.1969902958136047e-05,
+      "loss": 0.0292,
+      "step": 86645
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.1969434156860909e-05,
+      "loss": 0.0933,
+      "step": 86650
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.1968965355585769e-05,
+      "loss": 0.0681,
+      "step": 86655
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.1968496554310629e-05,
+      "loss": 0.0521,
+      "step": 86660
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.1968027753035489e-05,
+      "loss": 0.0842,
+      "step": 86665
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.196755895176035e-05,
+      "loss": 0.1892,
+      "step": 86670
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.196709015048521e-05,
+      "loss": 0.3711,
+      "step": 86675
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.196662134921007e-05,
+      "loss": 0.0664,
+      "step": 86680
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.196615254793493e-05,
+      "loss": 0.0181,
+      "step": 86685
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.196568374665979e-05,
+      "loss": 0.042,
+      "step": 86690
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.1965214945384653e-05,
+      "loss": 0.0395,
+      "step": 86695
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.1964746144109513e-05,
+      "loss": 0.0535,
+      "step": 86700
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.1964277342834373e-05,
+      "loss": 0.1296,
+      "step": 86705
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.1963808541559235e-05,
+      "loss": 0.0728,
+      "step": 86710
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.1963339740284095e-05,
+      "loss": 0.1151,
+      "step": 86715
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.1962870939008955e-05,
+      "loss": 0.1472,
+      "step": 86720
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.1962402137733815e-05,
+      "loss": 0.3731,
+      "step": 86725
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.1961933336458675e-05,
+      "loss": 0.0713,
+      "step": 86730
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.1961464535183536e-05,
+      "loss": 0.054,
+      "step": 86735
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.1960995733908398e-05,
+      "loss": 0.0754,
+      "step": 86740
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.1960526932633258e-05,
+      "loss": 0.0406,
+      "step": 86745
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.196005813135812e-05,
+      "loss": 0.0433,
+      "step": 86750
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.195958933008298e-05,
+      "loss": 0.0618,
+      "step": 86755
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.195912052880784e-05,
+      "loss": 0.1214,
+      "step": 86760
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.19586517275327e-05,
+      "loss": 0.0823,
+      "step": 86765
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.195818292625756e-05,
+      "loss": 0.2509,
+      "step": 86770
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.1957714124982421e-05,
+      "loss": 0.281,
+      "step": 86775
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.195724532370728e-05,
+      "loss": 0.0714,
+      "step": 86780
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.1956776522432142e-05,
+      "loss": 0.0125,
+      "step": 86785
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.1956307721157004e-05,
+      "loss": 0.0437,
+      "step": 86790
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.1955838919881864e-05,
+      "loss": 0.02,
+      "step": 86795
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.1955370118606724e-05,
+      "loss": 0.0802,
+      "step": 86800
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.1954901317331584e-05,
+      "loss": 0.0699,
+      "step": 86805
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.1954432516056444e-05,
+      "loss": 0.0943,
+      "step": 86810
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.1953963714781305e-05,
+      "loss": 0.0623,
+      "step": 86815
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.1953494913506165e-05,
+      "loss": 0.176,
+      "step": 86820
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.1953026112231025e-05,
+      "loss": 0.2118,
+      "step": 86825
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.1952557310955885e-05,
+      "loss": 0.0911,
+      "step": 86830
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.1952088509680749e-05,
+      "loss": 0.0413,
+      "step": 86835
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.1951619708405608e-05,
+      "loss": 0.0189,
+      "step": 86840
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.1951150907130468e-05,
+      "loss": 0.0124,
+      "step": 86845
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.1950682105855328e-05,
+      "loss": 0.0306,
+      "step": 86850
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.195021330458019e-05,
+      "loss": 0.0734,
+      "step": 86855
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.194974450330505e-05,
+      "loss": 0.0863,
+      "step": 86860
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.194927570202991e-05,
+      "loss": 0.191,
+      "step": 86865
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.194880690075477e-05,
+      "loss": 0.1266,
+      "step": 86870
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.194833809947963e-05,
+      "loss": 0.2328,
+      "step": 86875
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.1947869298204493e-05,
+      "loss": 0.09,
+      "step": 86880
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.1947400496929353e-05,
+      "loss": 0.0347,
+      "step": 86885
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.1946931695654213e-05,
+      "loss": 0.0266,
+      "step": 86890
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.1946462894379075e-05,
+      "loss": 0.053,
+      "step": 86895
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.1945994093103934e-05,
+      "loss": 0.0394,
+      "step": 86900
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.1945525291828794e-05,
+      "loss": 0.0295,
+      "step": 86905
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.1945056490553654e-05,
+      "loss": 0.0788,
+      "step": 86910
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.1944587689278514e-05,
+      "loss": 0.1111,
+      "step": 86915
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.1944118888003376e-05,
+      "loss": 0.1836,
+      "step": 86920
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.1943650086728238e-05,
+      "loss": 0.2362,
+      "step": 86925
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.1943181285453097e-05,
+      "loss": 0.1221,
+      "step": 86930
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.1942712484177959e-05,
+      "loss": 0.004,
+      "step": 86935
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.1942243682902819e-05,
+      "loss": 0.0897,
+      "step": 86940
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.1941774881627679e-05,
+      "loss": 0.0156,
+      "step": 86945
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.1941306080352539e-05,
+      "loss": 0.0265,
+      "step": 86950
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.1940837279077399e-05,
+      "loss": 0.125,
+      "step": 86955
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.194036847780226e-05,
+      "loss": 0.1269,
+      "step": 86960
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.193989967652712e-05,
+      "loss": 0.1611,
+      "step": 86965
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.1939430875251982e-05,
+      "loss": 0.2753,
+      "step": 86970
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.1938962073976844e-05,
+      "loss": 0.1691,
+      "step": 86975
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.1938493272701704e-05,
+      "loss": 0.0731,
+      "step": 86980
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.1938024471426563e-05,
+      "loss": 0.0174,
+      "step": 86985
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.1937555670151423e-05,
+      "loss": 0.0531,
+      "step": 86990
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.1937086868876285e-05,
+      "loss": 0.1098,
+      "step": 86995
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.1936618067601145e-05,
+      "loss": 0.0883,
+      "step": 87000
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.1936149266326005e-05,
+      "loss": 0.1077,
+      "step": 87005
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.1935680465050865e-05,
+      "loss": 0.0682,
+      "step": 87010
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.1935211663775725e-05,
+      "loss": 0.0773,
+      "step": 87015
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.1934742862500588e-05,
+      "loss": 0.2073,
+      "step": 87020
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.1934274061225448e-05,
+      "loss": 0.3685,
+      "step": 87025
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.1933805259950308e-05,
+      "loss": 0.0654,
+      "step": 87030
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.193333645867517e-05,
+      "loss": 0.0332,
+      "step": 87035
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.193286765740003e-05,
+      "loss": 0.0198,
+      "step": 87040
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.193239885612489e-05,
+      "loss": 0.0684,
+      "step": 87045
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.193193005484975e-05,
+      "loss": 0.0284,
+      "step": 87050
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.193146125357461e-05,
+      "loss": 0.0423,
+      "step": 87055
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.1930992452299471e-05,
+      "loss": 0.0619,
+      "step": 87060
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.1930523651024333e-05,
+      "loss": 0.157,
+      "step": 87065
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.1930054849749193e-05,
+      "loss": 0.119,
+      "step": 87070
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.1929586048474054e-05,
+      "loss": 0.2258,
+      "step": 87075
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.1929117247198914e-05,
+      "loss": 0.0805,
+      "step": 87080
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.1928648445923774e-05,
+      "loss": 0.0118,
+      "step": 87085
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.1928179644648634e-05,
+      "loss": 0.0255,
+      "step": 87090
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.1927710843373494e-05,
+      "loss": 0.0596,
+      "step": 87095
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.1927242042098356e-05,
+      "loss": 0.1062,
+      "step": 87100
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.1926773240823215e-05,
+      "loss": 0.0723,
+      "step": 87105
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.1926304439548077e-05,
+      "loss": 0.1052,
+      "step": 87110
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.1925835638272939e-05,
+      "loss": 0.1319,
+      "step": 87115
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.1925366836997799e-05,
+      "loss": 0.2386,
+      "step": 87120
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.1924898035722659e-05,
+      "loss": 0.4077,
+      "step": 87125
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.1924429234447519e-05,
+      "loss": 0.0996,
+      "step": 87130
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.1923960433172378e-05,
+      "loss": 0.0111,
+      "step": 87135
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.192349163189724e-05,
+      "loss": 0.0213,
+      "step": 87140
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.19230228306221e-05,
+      "loss": 0.0942,
+      "step": 87145
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.192255402934696e-05,
+      "loss": 0.0585,
+      "step": 87150
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.192208522807182e-05,
+      "loss": 0.09,
+      "step": 87155
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.1921616426796683e-05,
+      "loss": 0.1602,
+      "step": 87160
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.1921147625521543e-05,
+      "loss": 0.0582,
+      "step": 87165
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.1920678824246403e-05,
+      "loss": 0.1511,
+      "step": 87170
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.1920210022971263e-05,
+      "loss": 0.1997,
+      "step": 87175
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.1919741221696125e-05,
+      "loss": 0.1347,
+      "step": 87180
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.1919272420420985e-05,
+      "loss": 0.0469,
+      "step": 87185
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.1918803619145844e-05,
+      "loss": 0.0572,
+      "step": 87190
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.1918334817870704e-05,
+      "loss": 0.0742,
+      "step": 87195
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.1917866016595564e-05,
+      "loss": 0.0608,
+      "step": 87200
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.1917397215320428e-05,
+      "loss": 0.08,
+      "step": 87205
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.1916928414045288e-05,
+      "loss": 0.0855,
+      "step": 87210
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.1916459612770148e-05,
+      "loss": 0.0958,
+      "step": 87215
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.191599081149501e-05,
+      "loss": 0.2308,
+      "step": 87220
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.1915522010219869e-05,
+      "loss": 0.2714,
+      "step": 87225
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.1915053208944729e-05,
+      "loss": 0.0695,
+      "step": 87230
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.1914584407669589e-05,
+      "loss": 0.009,
+      "step": 87235
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.1914115606394449e-05,
+      "loss": 0.0141,
+      "step": 87240
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.191364680511931e-05,
+      "loss": 0.0225,
+      "step": 87245
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.1913178003844172e-05,
+      "loss": 0.0423,
+      "step": 87250
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.1912709202569032e-05,
+      "loss": 0.0474,
+      "step": 87255
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.1912240401293894e-05,
+      "loss": 0.0693,
+      "step": 87260
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.1911771600018754e-05,
+      "loss": 0.0688,
+      "step": 87265
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.1911302798743614e-05,
+      "loss": 0.1978,
+      "step": 87270
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.1910833997468474e-05,
+      "loss": 0.3268,
+      "step": 87275
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.1910365196193333e-05,
+      "loss": 0.1005,
+      "step": 87280
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.1909896394918195e-05,
+      "loss": 0.024,
+      "step": 87285
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.1909427593643055e-05,
+      "loss": 0.0411,
+      "step": 87290
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.1908958792367917e-05,
+      "loss": 0.0444,
+      "step": 87295
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.1908489991092778e-05,
+      "loss": 0.0589,
+      "step": 87300
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.1908021189817638e-05,
+      "loss": 0.0539,
+      "step": 87305
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.1907552388542498e-05,
+      "loss": 0.1432,
+      "step": 87310
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.1907083587267358e-05,
+      "loss": 0.0775,
+      "step": 87315
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.1906614785992218e-05,
+      "loss": 0.1898,
+      "step": 87320
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.190614598471708e-05,
+      "loss": 0.3632,
+      "step": 87325
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.190567718344194e-05,
+      "loss": 0.0627,
+      "step": 87330
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.19052083821668e-05,
+      "loss": 0.0112,
+      "step": 87335
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.190473958089166e-05,
+      "loss": 0.0243,
+      "step": 87340
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.1904270779616523e-05,
+      "loss": 0.0383,
+      "step": 87345
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.1903801978341383e-05,
+      "loss": 0.0393,
+      "step": 87350
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.1903333177066243e-05,
+      "loss": 0.0794,
+      "step": 87355
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.1902864375791103e-05,
+      "loss": 0.0875,
+      "step": 87360
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.1902395574515964e-05,
+      "loss": 0.0768,
+      "step": 87365
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.1901926773240824e-05,
+      "loss": 0.137,
+      "step": 87370
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.1901457971965684e-05,
+      "loss": 0.2204,
+      "step": 87375
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.1900989170690544e-05,
+      "loss": 0.0299,
+      "step": 87380
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.1900520369415406e-05,
+      "loss": 0.0461,
+      "step": 87385
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.1900051568140267e-05,
+      "loss": 0.0408,
+      "step": 87390
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.1899582766865127e-05,
+      "loss": 0.0258,
+      "step": 87395
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.1899113965589987e-05,
+      "loss": 0.0267,
+      "step": 87400
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.1898645164314849e-05,
+      "loss": 0.0925,
+      "step": 87405
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.1898176363039709e-05,
+      "loss": 0.0924,
+      "step": 87410
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.1897707561764569e-05,
+      "loss": 0.109,
+      "step": 87415
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.1897238760489429e-05,
+      "loss": 0.2402,
+      "step": 87420
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.189676995921429e-05,
+      "loss": 0.2597,
+      "step": 87425
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.189630115793915e-05,
+      "loss": 0.0844,
+      "step": 87430
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.1895832356664012e-05,
+      "loss": 0.0202,
+      "step": 87435
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.1895363555388872e-05,
+      "loss": 0.0664,
+      "step": 87440
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.1894894754113733e-05,
+      "loss": 0.0505,
+      "step": 87445
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.1894425952838593e-05,
+      "loss": 0.0556,
+      "step": 87450
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.1893957151563453e-05,
+      "loss": 0.1165,
+      "step": 87455
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.1893488350288313e-05,
+      "loss": 0.0669,
+      "step": 87460
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.1893019549013175e-05,
+      "loss": 0.1502,
+      "step": 87465
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.1892550747738035e-05,
+      "loss": 0.3517,
+      "step": 87470
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.1892081946462895e-05,
+      "loss": 0.2206,
+      "step": 87475
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.1891613145187755e-05,
+      "loss": 0.0826,
+      "step": 87480
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.1891144343912618e-05,
+      "loss": 0.0292,
+      "step": 87485
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.1890675542637478e-05,
+      "loss": 0.0348,
+      "step": 87490
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.1890206741362338e-05,
+      "loss": 0.0419,
+      "step": 87495
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.1889737940087198e-05,
+      "loss": 0.0687,
+      "step": 87500
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.188926913881206e-05,
+      "loss": 0.0166,
+      "step": 87505
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.188880033753692e-05,
+      "loss": 0.0892,
+      "step": 87510
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.1888331536261779e-05,
+      "loss": 0.1468,
+      "step": 87515
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.1887862734986639e-05,
+      "loss": 0.1371,
+      "step": 87520
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.1887393933711499e-05,
+      "loss": 0.229,
+      "step": 87525
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.1886925132436362e-05,
+      "loss": 0.1088,
+      "step": 87530
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.1886456331161222e-05,
+      "loss": 0.0123,
+      "step": 87535
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.1885987529886082e-05,
+      "loss": 0.0759,
+      "step": 87540
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.1885518728610944e-05,
+      "loss": 0.03,
+      "step": 87545
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.1885049927335804e-05,
+      "loss": 0.0509,
+      "step": 87550
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.1884581126060664e-05,
+      "loss": 0.073,
+      "step": 87555
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.1884112324785524e-05,
+      "loss": 0.1191,
+      "step": 87560
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.1883643523510384e-05,
+      "loss": 0.0545,
+      "step": 87565
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.1883174722235245e-05,
+      "loss": 0.0946,
+      "step": 87570
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.1882705920960107e-05,
+      "loss": 0.2369,
+      "step": 87575
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.1882237119684967e-05,
+      "loss": 0.052,
+      "step": 87580
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.1881768318409828e-05,
+      "loss": 0.0259,
+      "step": 87585
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.1881299517134688e-05,
+      "loss": 0.0106,
+      "step": 87590
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.1880830715859548e-05,
+      "loss": 0.0452,
+      "step": 87595
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.1880361914584408e-05,
+      "loss": 0.0208,
+      "step": 87600
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.1879893113309268e-05,
+      "loss": 0.0497,
+      "step": 87605
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.187942431203413e-05,
+      "loss": 0.0882,
+      "step": 87610
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.187895551075899e-05,
+      "loss": 0.17,
+      "step": 87615
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.1878486709483851e-05,
+      "loss": 0.1611,
+      "step": 87620
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.1878017908208713e-05,
+      "loss": 0.2782,
+      "step": 87625
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.1877549106933573e-05,
+      "loss": 0.0584,
+      "step": 87630
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.1877080305658433e-05,
+      "loss": 0.0585,
+      "step": 87635
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.1876611504383293e-05,
+      "loss": 0.0534,
+      "step": 87640
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.1876142703108153e-05,
+      "loss": 0.0818,
+      "step": 87645
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.1875673901833014e-05,
+      "loss": 0.0938,
+      "step": 87650
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.1875205100557874e-05,
+      "loss": 0.0645,
+      "step": 87655
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.1874736299282734e-05,
+      "loss": 0.058,
+      "step": 87660
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.1874267498007594e-05,
+      "loss": 0.1157,
+      "step": 87665
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.1873798696732457e-05,
+      "loss": 0.1999,
+      "step": 87670
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.1873329895457317e-05,
+      "loss": 0.3274,
+      "step": 87675
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.1872861094182177e-05,
+      "loss": 0.121,
+      "step": 87680
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.1872392292907037e-05,
+      "loss": 0.0206,
+      "step": 87685
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.1871923491631899e-05,
+      "loss": 0.0536,
+      "step": 87690
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.1871454690356759e-05,
+      "loss": 0.054,
+      "step": 87695
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.1870985889081619e-05,
+      "loss": 0.0702,
+      "step": 87700
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.1870517087806479e-05,
+      "loss": 0.0894,
+      "step": 87705
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.1870048286531339e-05,
+      "loss": 0.0561,
+      "step": 87710
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.1869579485256202e-05,
+      "loss": 0.1382,
+      "step": 87715
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.1869110683981062e-05,
+      "loss": 0.0828,
+      "step": 87720
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.1868641882705922e-05,
+      "loss": 0.2575,
+      "step": 87725
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.1868173081430783e-05,
+      "loss": 0.105,
+      "step": 87730
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.1867704280155643e-05,
+      "loss": 0.0239,
+      "step": 87735
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.1867235478880503e-05,
+      "loss": 0.031,
+      "step": 87740
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.1866766677605363e-05,
+      "loss": 0.0219,
+      "step": 87745
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.1866297876330223e-05,
+      "loss": 0.0578,
+      "step": 87750
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.1865829075055085e-05,
+      "loss": 0.0634,
+      "step": 87755
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.1865360273779946e-05,
+      "loss": 0.0727,
+      "step": 87760
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.1864891472504806e-05,
+      "loss": 0.1068,
+      "step": 87765
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.1864422671229668e-05,
+      "loss": 0.2169,
+      "step": 87770
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.1863953869954528e-05,
+      "loss": 0.3113,
+      "step": 87775
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.1863485068679388e-05,
+      "loss": 0.0885,
+      "step": 87780
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.1863016267404248e-05,
+      "loss": 0.0041,
+      "step": 87785
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.1862547466129108e-05,
+      "loss": 0.0168,
+      "step": 87790
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.186207866485397e-05,
+      "loss": 0.0328,
+      "step": 87795
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.186160986357883e-05,
+      "loss": 0.0349,
+      "step": 87800
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.186114106230369e-05,
+      "loss": 0.0933,
+      "step": 87805
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.1860672261028552e-05,
+      "loss": 0.0872,
+      "step": 87810
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.1860203459753412e-05,
+      "loss": 0.0645,
+      "step": 87815
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.1859734658478272e-05,
+      "loss": 0.0875,
+      "step": 87820
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.1859265857203132e-05,
+      "loss": 0.3078,
+      "step": 87825
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.1858797055927992e-05,
+      "loss": 0.0885,
+      "step": 87830
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.1858328254652854e-05,
+      "loss": 0.0082,
+      "step": 87835
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.1857859453377714e-05,
+      "loss": 0.0125,
+      "step": 87840
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.1857390652102574e-05,
+      "loss": 0.0745,
+      "step": 87845
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.1856921850827434e-05,
+      "loss": 0.0627,
+      "step": 87850
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.1856453049552297e-05,
+      "loss": 0.1043,
+      "step": 87855
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.1855984248277157e-05,
+      "loss": 0.1207,
+      "step": 87860
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.1855515447002017e-05,
+      "loss": 0.0784,
+      "step": 87865
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.1855046645726877e-05,
+      "loss": 0.0709,
+      "step": 87870
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.1854577844451738e-05,
+      "loss": 0.3188,
+      "step": 87875
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.1854109043176598e-05,
+      "loss": 0.097,
+      "step": 87880
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.1853640241901458e-05,
+      "loss": 0.0212,
+      "step": 87885
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.1853171440626318e-05,
+      "loss": 0.0323,
+      "step": 87890
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.185270263935118e-05,
+      "loss": 0.0622,
+      "step": 87895
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.1852233838076041e-05,
+      "loss": 0.0712,
+      "step": 87900
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.1851765036800901e-05,
+      "loss": 0.0905,
+      "step": 87905
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.1851296235525761e-05,
+      "loss": 0.1088,
+      "step": 87910
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.1850827434250623e-05,
+      "loss": 0.1603,
+      "step": 87915
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.1850358632975483e-05,
+      "loss": 0.1569,
+      "step": 87920
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.1849889831700343e-05,
+      "loss": 0.4106,
+      "step": 87925
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.1849421030425203e-05,
+      "loss": 0.0888,
+      "step": 87930
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.1848952229150064e-05,
+      "loss": 0.0147,
+      "step": 87935
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.1848483427874924e-05,
+      "loss": 0.0204,
+      "step": 87940
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.1848014626599786e-05,
+      "loss": 0.0484,
+      "step": 87945
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.1847545825324646e-05,
+      "loss": 0.0552,
+      "step": 87950
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.1847077024049508e-05,
+      "loss": 0.1071,
+      "step": 87955
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.1846608222774367e-05,
+      "loss": 0.0732,
+      "step": 87960
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.1846139421499227e-05,
+      "loss": 0.1531,
+      "step": 87965
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.1845670620224087e-05,
+      "loss": 0.1324,
+      "step": 87970
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.1845201818948949e-05,
+      "loss": 0.3174,
+      "step": 87975
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.1844733017673809e-05,
+      "loss": 0.0821,
+      "step": 87980
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.1844264216398669e-05,
+      "loss": 0.0447,
+      "step": 87985
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.1843795415123529e-05,
+      "loss": 0.0748,
+      "step": 87990
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.1843326613848392e-05,
+      "loss": 0.0336,
+      "step": 87995
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.1842857812573252e-05,
+      "loss": 0.0854,
+      "step": 88000
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.1842389011298112e-05,
+      "loss": 0.084,
+      "step": 88005
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.1841920210022972e-05,
+      "loss": 0.08,
+      "step": 88010
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.1841451408747833e-05,
+      "loss": 0.1339,
+      "step": 88015
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.1840982607472693e-05,
+      "loss": 0.1688,
+      "step": 88020
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.1840513806197553e-05,
+      "loss": 0.2253,
+      "step": 88025
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.1840045004922413e-05,
+      "loss": 0.0564,
+      "step": 88030
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.1839576203647273e-05,
+      "loss": 0.0427,
+      "step": 88035
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.1839107402372137e-05,
+      "loss": 0.0204,
+      "step": 88040
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.1838638601096996e-05,
+      "loss": 0.0069,
+      "step": 88045
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.1838169799821856e-05,
+      "loss": 0.0884,
+      "step": 88050
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.1837700998546718e-05,
+      "loss": 0.1743,
+      "step": 88055
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.1837232197271578e-05,
+      "loss": 0.0626,
+      "step": 88060
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.1836763395996438e-05,
+      "loss": 0.1187,
+      "step": 88065
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.1836294594721298e-05,
+      "loss": 0.114,
+      "step": 88070
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.1835825793446158e-05,
+      "loss": 0.2401,
+      "step": 88075
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.183535699217102e-05,
+      "loss": 0.0934,
+      "step": 88080
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.1834888190895881e-05,
+      "loss": 0.0237,
+      "step": 88085
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.1834419389620741e-05,
+      "loss": 0.0265,
+      "step": 88090
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.1833950588345603e-05,
+      "loss": 0.0286,
+      "step": 88095
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.1833481787070463e-05,
+      "loss": 0.0595,
+      "step": 88100
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.1833012985795322e-05,
+      "loss": 0.1194,
+      "step": 88105
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.1832544184520182e-05,
+      "loss": 0.0379,
+      "step": 88110
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.1832075383245042e-05,
+      "loss": 0.0703,
+      "step": 88115
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.1831606581969904e-05,
+      "loss": 0.1492,
+      "step": 88120
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.1831137780694764e-05,
+      "loss": 0.2666,
+      "step": 88125
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.1830668979419624e-05,
+      "loss": 0.1059,
+      "step": 88130
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.1830200178144487e-05,
+      "loss": 0.0229,
+      "step": 88135
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.1829731376869347e-05,
+      "loss": 0.0534,
+      "step": 88140
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.1829262575594207e-05,
+      "loss": 0.0788,
+      "step": 88145
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.1828793774319067e-05,
+      "loss": 0.0502,
+      "step": 88150
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.1828324973043927e-05,
+      "loss": 0.0303,
+      "step": 88155
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.1827856171768789e-05,
+      "loss": 0.1054,
+      "step": 88160
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.1827387370493648e-05,
+      "loss": 0.0756,
+      "step": 88165
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.1826918569218508e-05,
+      "loss": 0.1272,
+      "step": 88170
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.1826449767943368e-05,
+      "loss": 0.2779,
+      "step": 88175
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.1825980966668232e-05,
+      "loss": 0.0624,
+      "step": 88180
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.1825512165393092e-05,
+      "loss": 0.038,
+      "step": 88185
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.1825043364117951e-05,
+      "loss": 0.015,
+      "step": 88190
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.1824574562842811e-05,
+      "loss": 0.0463,
+      "step": 88195
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.1824105761567673e-05,
+      "loss": 0.0571,
+      "step": 88200
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.1823636960292533e-05,
+      "loss": 0.1115,
+      "step": 88205
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.1823168159017393e-05,
+      "loss": 0.1003,
+      "step": 88210
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.1822699357742253e-05,
+      "loss": 0.1697,
+      "step": 88215
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.1822230556467113e-05,
+      "loss": 0.1415,
+      "step": 88220
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.1821761755191976e-05,
+      "loss": 0.4007,
+      "step": 88225
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.1821292953916836e-05,
+      "loss": 0.0847,
+      "step": 88230
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.1820824152641696e-05,
+      "loss": 0.0565,
+      "step": 88235
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.1820355351366558e-05,
+      "loss": 0.045,
+      "step": 88240
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.1819886550091418e-05,
+      "loss": 0.0273,
+      "step": 88245
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.1819417748816277e-05,
+      "loss": 0.0581,
+      "step": 88250
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.1818948947541137e-05,
+      "loss": 0.0596,
+      "step": 88255
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.1818480146265997e-05,
+      "loss": 0.104,
+      "step": 88260
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.1818011344990859e-05,
+      "loss": 0.066,
+      "step": 88265
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.181754254371572e-05,
+      "loss": 0.1752,
+      "step": 88270
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.181707374244058e-05,
+      "loss": 0.1842,
+      "step": 88275
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.1816604941165442e-05,
+      "loss": 0.0878,
+      "step": 88280
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.1816136139890302e-05,
+      "loss": 0.0034,
+      "step": 88285
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.1815667338615162e-05,
+      "loss": 0.0331,
+      "step": 88290
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.1815198537340022e-05,
+      "loss": 0.062,
+      "step": 88295
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.1814729736064882e-05,
+      "loss": 0.0553,
+      "step": 88300
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.1814260934789744e-05,
+      "loss": 0.0708,
+      "step": 88305
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.1813792133514603e-05,
+      "loss": 0.118,
+      "step": 88310
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.1813323332239463e-05,
+      "loss": 0.031,
+      "step": 88315
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.1812854530964327e-05,
+      "loss": 0.2011,
+      "step": 88320
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.1812385729689187e-05,
+      "loss": 0.2144,
+      "step": 88325
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.1811916928414047e-05,
+      "loss": 0.0985,
+      "step": 88330
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.1811448127138907e-05,
+      "loss": 0.0294,
+      "step": 88335
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.1810979325863766e-05,
+      "loss": 0.0454,
+      "step": 88340
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.1810510524588628e-05,
+      "loss": 0.0536,
+      "step": 88345
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.1810041723313488e-05,
+      "loss": 0.0362,
+      "step": 88350
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.1809572922038348e-05,
+      "loss": 0.0595,
+      "step": 88355
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.1809104120763208e-05,
+      "loss": 0.1038,
+      "step": 88360
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.1808635319488071e-05,
+      "loss": 0.1226,
+      "step": 88365
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.1808166518212931e-05,
+      "loss": 0.1687,
+      "step": 88370
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.1807697716937791e-05,
+      "loss": 0.2789,
+      "step": 88375
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.1807228915662651e-05,
+      "loss": 0.1013,
+      "step": 88380
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.1806760114387513e-05,
+      "loss": 0.0386,
+      "step": 88385
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.1806291313112373e-05,
+      "loss": 0.059,
+      "step": 88390
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.1805822511837232e-05,
+      "loss": 0.0712,
+      "step": 88395
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.1805353710562092e-05,
+      "loss": 0.0891,
+      "step": 88400
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.1804884909286954e-05,
+      "loss": 0.0921,
+      "step": 88405
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.1804416108011816e-05,
+      "loss": 0.083,
+      "step": 88410
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.1803947306736676e-05,
+      "loss": 0.053,
+      "step": 88415
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.1803478505461537e-05,
+      "loss": 0.1004,
+      "step": 88420
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.1803009704186397e-05,
+      "loss": 0.3463,
+      "step": 88425
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.1802540902911257e-05,
+      "loss": 0.0652,
+      "step": 88430
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.1802072101636117e-05,
+      "loss": 0.0415,
+      "step": 88435
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.1801603300360977e-05,
+      "loss": 0.0458,
+      "step": 88440
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.1801134499085839e-05,
+      "loss": 0.0418,
+      "step": 88445
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.1800665697810699e-05,
+      "loss": 0.0401,
+      "step": 88450
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.1800196896535558e-05,
+      "loss": 0.0574,
+      "step": 88455
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.1799728095260422e-05,
+      "loss": 0.0761,
+      "step": 88460
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.1799259293985282e-05,
+      "loss": 0.0803,
+      "step": 88465
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.1798790492710142e-05,
+      "loss": 0.1139,
+      "step": 88470
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.1798321691435002e-05,
+      "loss": 0.2541,
+      "step": 88475
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.1797852890159862e-05,
+      "loss": 0.0633,
+      "step": 88480
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.1797384088884723e-05,
+      "loss": 0.0298,
+      "step": 88485
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.1796915287609583e-05,
+      "loss": 0.0426,
+      "step": 88490
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.1796446486334443e-05,
+      "loss": 0.0683,
+      "step": 88495
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.1795977685059303e-05,
+      "loss": 0.0408,
+      "step": 88500
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.1795508883784166e-05,
+      "loss": 0.0765,
+      "step": 88505
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.1795040082509026e-05,
+      "loss": 0.0715,
+      "step": 88510
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.1794571281233886e-05,
+      "loss": 0.12,
+      "step": 88515
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.1794102479958746e-05,
+      "loss": 0.1333,
+      "step": 88520
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.1793633678683608e-05,
+      "loss": 0.2048,
+      "step": 88525
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.1793164877408468e-05,
+      "loss": 0.1168,
+      "step": 88530
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.1792696076133328e-05,
+      "loss": 0.0439,
+      "step": 88535
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.1792227274858188e-05,
+      "loss": 0.0645,
+      "step": 88540
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.1791758473583047e-05,
+      "loss": 0.0506,
+      "step": 88545
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.179128967230791e-05,
+      "loss": 0.025,
+      "step": 88550
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.179082087103277e-05,
+      "loss": 0.0817,
+      "step": 88555
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.179035206975763e-05,
+      "loss": 0.1231,
+      "step": 88560
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.1789883268482492e-05,
+      "loss": 0.1282,
+      "step": 88565
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.1789414467207352e-05,
+      "loss": 0.2506,
+      "step": 88570
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.1788945665932212e-05,
+      "loss": 0.3161,
+      "step": 88575
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.1788476864657072e-05,
+      "loss": 0.0732,
+      "step": 88580
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.1788008063381932e-05,
+      "loss": 0.0347,
+      "step": 88585
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.1787539262106794e-05,
+      "loss": 0.0442,
+      "step": 88590
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.1787070460831655e-05,
+      "loss": 0.0373,
+      "step": 88595
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.1786601659556515e-05,
+      "loss": 0.1332,
+      "step": 88600
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.1786132858281377e-05,
+      "loss": 0.0915,
+      "step": 88605
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.1785664057006237e-05,
+      "loss": 0.0812,
+      "step": 88610
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.1785195255731097e-05,
+      "loss": 0.2585,
+      "step": 88615
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.1784726454455957e-05,
+      "loss": 0.1337,
+      "step": 88620
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.1784257653180817e-05,
+      "loss": 0.243,
+      "step": 88625
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.1783788851905678e-05,
+      "loss": 0.0926,
+      "step": 88630
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.1783320050630538e-05,
+      "loss": 0.0232,
+      "step": 88635
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.1782851249355398e-05,
+      "loss": 0.0341,
+      "step": 88640
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.1782382448080261e-05,
+      "loss": 0.0606,
+      "step": 88645
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.1781913646805121e-05,
+      "loss": 0.0525,
+      "step": 88650
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.1781444845529981e-05,
+      "loss": 0.0898,
+      "step": 88655
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.1780976044254841e-05,
+      "loss": 0.1552,
+      "step": 88660
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.1780507242979701e-05,
+      "loss": 0.0678,
+      "step": 88665
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.1780038441704563e-05,
+      "loss": 0.1634,
+      "step": 88670
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.1779569640429423e-05,
+      "loss": 0.2653,
+      "step": 88675
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.1779100839154283e-05,
+      "loss": 0.0456,
+      "step": 88680
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.1778632037879143e-05,
+      "loss": 0.0381,
+      "step": 88685
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.1778163236604006e-05,
+      "loss": 0.0179,
+      "step": 88690
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.1777694435328866e-05,
+      "loss": 0.0408,
+      "step": 88695
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.1777225634053726e-05,
+      "loss": 0.062,
+      "step": 88700
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.1776756832778586e-05,
+      "loss": 0.0861,
+      "step": 88705
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.1776288031503447e-05,
+      "loss": 0.1181,
+      "step": 88710
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.1775819230228307e-05,
+      "loss": 0.1903,
+      "step": 88715
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.1775350428953167e-05,
+      "loss": 0.109,
+      "step": 88720
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.1774881627678027e-05,
+      "loss": 0.2255,
+      "step": 88725
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.1774412826402887e-05,
+      "loss": 0.0749,
+      "step": 88730
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.177394402512775e-05,
+      "loss": 0.0773,
+      "step": 88735
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.177347522385261e-05,
+      "loss": 0.0129,
+      "step": 88740
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.177300642257747e-05,
+      "loss": 0.063,
+      "step": 88745
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.1772537621302332e-05,
+      "loss": 0.0323,
+      "step": 88750
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.1772068820027192e-05,
+      "loss": 0.087,
+      "step": 88755
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.1771600018752052e-05,
+      "loss": 0.152,
+      "step": 88760
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.1771131217476912e-05,
+      "loss": 0.169,
+      "step": 88765
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.1770662416201773e-05,
+      "loss": 0.1258,
+      "step": 88770
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.1770193614926633e-05,
+      "loss": 0.2229,
+      "step": 88775
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.1769724813651493e-05,
+      "loss": 0.059,
+      "step": 88780
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.1769256012376355e-05,
+      "loss": 0.0488,
+      "step": 88785
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.1768787211101216e-05,
+      "loss": 0.0462,
+      "step": 88790
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.1768318409826076e-05,
+      "loss": 0.0595,
+      "step": 88795
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.1767849608550936e-05,
+      "loss": 0.0647,
+      "step": 88800
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.1767380807275796e-05,
+      "loss": 0.0766,
+      "step": 88805
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.1766912006000658e-05,
+      "loss": 0.0546,
+      "step": 88810
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.1766443204725518e-05,
+      "loss": 0.0309,
+      "step": 88815
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.1765974403450378e-05,
+      "loss": 0.0951,
+      "step": 88820
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.1765505602175238e-05,
+      "loss": 0.1486,
+      "step": 88825
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.1765036800900101e-05,
+      "loss": 0.0767,
+      "step": 88830
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.1764567999624961e-05,
+      "loss": 0.0409,
+      "step": 88835
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.176409919834982e-05,
+      "loss": 0.0294,
+      "step": 88840
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.176363039707468e-05,
+      "loss": 0.0727,
+      "step": 88845
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.1763161595799542e-05,
+      "loss": 0.0128,
+      "step": 88850
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.1762692794524402e-05,
+      "loss": 0.0553,
+      "step": 88855
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.1762223993249262e-05,
+      "loss": 0.0832,
+      "step": 88860
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.1761755191974122e-05,
+      "loss": 0.0824,
+      "step": 88865
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.1761286390698982e-05,
+      "loss": 0.1037,
+      "step": 88870
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.1760817589423845e-05,
+      "loss": 0.4061,
+      "step": 88875
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.1760348788148705e-05,
+      "loss": 0.1145,
+      "step": 88880
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.1759879986873565e-05,
+      "loss": 0.2368,
+      "step": 88885
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.1759411185598427e-05,
+      "loss": 0.0173,
+      "step": 88890
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.1758942384323287e-05,
+      "loss": 0.0621,
+      "step": 88895
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.1758473583048147e-05,
+      "loss": 0.0346,
+      "step": 88900
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.1758004781773007e-05,
+      "loss": 0.0817,
+      "step": 88905
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.1757535980497867e-05,
+      "loss": 0.1684,
+      "step": 88910
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.1757067179222728e-05,
+      "loss": 0.1637,
+      "step": 88915
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.175659837794759e-05,
+      "loss": 0.1615,
+      "step": 88920
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.175612957667245e-05,
+      "loss": 0.1003,
+      "step": 88925
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.1755660775397311e-05,
+      "loss": 0.1181,
+      "step": 88930
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.1755191974122171e-05,
+      "loss": 0.0264,
+      "step": 88935
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.1754723172847031e-05,
+      "loss": 0.0501,
+      "step": 88940
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.1754254371571891e-05,
+      "loss": 0.0896,
+      "step": 88945
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.1753785570296751e-05,
+      "loss": 0.0177,
+      "step": 88950
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.1753316769021613e-05,
+      "loss": 0.0574,
+      "step": 88955
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.1752847967746473e-05,
+      "loss": 0.0604,
+      "step": 88960
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.1752379166471333e-05,
+      "loss": 0.1172,
+      "step": 88965
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.1751910365196196e-05,
+      "loss": 0.075,
+      "step": 88970
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.1751441563921056e-05,
+      "loss": 0.1926,
+      "step": 88975
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.1750972762645916e-05,
+      "loss": 0.1254,
+      "step": 88980
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.1750503961370776e-05,
+      "loss": 0.0377,
+      "step": 88985
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.1750035160095636e-05,
+      "loss": 0.0124,
+      "step": 88990
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.1749566358820497e-05,
+      "loss": 0.0527,
+      "step": 88995
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.1749097557545357e-05,
+      "loss": 0.0684,
+      "step": 89000
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.1748628756270217e-05,
+      "loss": 0.036,
+      "step": 89005
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.1748159954995077e-05,
+      "loss": 0.0722,
+      "step": 89010
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.174769115371994e-05,
+      "loss": 0.0797,
+      "step": 89015
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.17472223524448e-05,
+      "loss": 0.1527,
+      "step": 89020
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.174675355116966e-05,
+      "loss": 0.2188,
+      "step": 89025
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.174628474989452e-05,
+      "loss": 0.089,
+      "step": 89030
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.1745815948619382e-05,
+      "loss": 0.0208,
+      "step": 89035
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.1745347147344242e-05,
+      "loss": 0.0321,
+      "step": 89040
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.1744878346069102e-05,
+      "loss": 0.0461,
+      "step": 89045
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.1744409544793962e-05,
+      "loss": 0.0726,
+      "step": 89050
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.1743940743518822e-05,
+      "loss": 0.1387,
+      "step": 89055
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.1743471942243685e-05,
+      "loss": 0.1153,
+      "step": 89060
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.1743003140968545e-05,
+      "loss": 0.1698,
+      "step": 89065
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.1742534339693405e-05,
+      "loss": 0.1889,
+      "step": 89070
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.1742065538418266e-05,
+      "loss": 0.3885,
+      "step": 89075
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.1741596737143126e-05,
+      "loss": 0.075,
+      "step": 89080
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.1741127935867986e-05,
+      "loss": 0.0027,
+      "step": 89085
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.1740659134592846e-05,
+      "loss": 0.0209,
+      "step": 89090
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.1740190333317706e-05,
+      "loss": 0.1213,
+      "step": 89095
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.1739721532042568e-05,
+      "loss": 0.0651,
+      "step": 89100
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.1739252730767428e-05,
+      "loss": 0.0386,
+      "step": 89105
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.173878392949229e-05,
+      "loss": 0.0728,
+      "step": 89110
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.1738315128217151e-05,
+      "loss": 0.1389,
+      "step": 89115
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.1737846326942011e-05,
+      "loss": 0.1604,
+      "step": 89120
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.1737377525666871e-05,
+      "loss": 0.292,
+      "step": 89125
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.173690872439173e-05,
+      "loss": 0.0724,
+      "step": 89130
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.173643992311659e-05,
+      "loss": 0.0395,
+      "step": 89135
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.1735971121841452e-05,
+      "loss": 0.026,
+      "step": 89140
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.1735502320566312e-05,
+      "loss": 0.0254,
+      "step": 89145
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.1735033519291172e-05,
+      "loss": 0.0568,
+      "step": 89150
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.1734564718016036e-05,
+      "loss": 0.051,
+      "step": 89155
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.1734095916740896e-05,
+      "loss": 0.069,
+      "step": 89160
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.1733627115465755e-05,
+      "loss": 0.0804,
+      "step": 89165
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.1733158314190615e-05,
+      "loss": 0.1007,
+      "step": 89170
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.1732689512915475e-05,
+      "loss": 0.3237,
+      "step": 89175
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.1732220711640337e-05,
+      "loss": 0.0378,
+      "step": 89180
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.1731751910365197e-05,
+      "loss": 0.0284,
+      "step": 89185
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.1731283109090057e-05,
+      "loss": 0.0374,
+      "step": 89190
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.1730814307814917e-05,
+      "loss": 0.0462,
+      "step": 89195
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.173034550653978e-05,
+      "loss": 0.0606,
+      "step": 89200
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.172987670526464e-05,
+      "loss": 0.0739,
+      "step": 89205
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.17294079039895e-05,
+      "loss": 0.0919,
+      "step": 89210
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.172893910271436e-05,
+      "loss": 0.1291,
+      "step": 89215
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.1728470301439221e-05,
+      "loss": 0.1922,
+      "step": 89220
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.1728001500164081e-05,
+      "loss": 0.3333,
+      "step": 89225
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.1727532698888941e-05,
+      "loss": 0.095,
+      "step": 89230
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.1727063897613801e-05,
+      "loss": 0.0126,
+      "step": 89235
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.1726595096338663e-05,
+      "loss": 0.0252,
+      "step": 89240
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.1726126295063525e-05,
+      "loss": 0.0179,
+      "step": 89245
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.1725657493788384e-05,
+      "loss": 0.0795,
+      "step": 89250
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.1725188692513244e-05,
+      "loss": 0.1216,
+      "step": 89255
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.1724719891238106e-05,
+      "loss": 0.0173,
+      "step": 89260
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.1724251089962966e-05,
+      "loss": 0.1934,
+      "step": 89265
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.1723782288687826e-05,
+      "loss": 0.2105,
+      "step": 89270
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.1723313487412686e-05,
+      "loss": 0.2147,
+      "step": 89275
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.1722844686137547e-05,
+      "loss": 0.1033,
+      "step": 89280
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.1722375884862407e-05,
+      "loss": 0.0282,
+      "step": 89285
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.1721907083587267e-05,
+      "loss": 0.0108,
+      "step": 89290
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.1721438282312129e-05,
+      "loss": 0.0072,
+      "step": 89295
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.172096948103699e-05,
+      "loss": 0.1576,
+      "step": 89300
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.172050067976185e-05,
+      "loss": 0.0357,
+      "step": 89305
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.172003187848671e-05,
+      "loss": 0.0539,
+      "step": 89310
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.171956307721157e-05,
+      "loss": 0.1258,
+      "step": 89315
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.1719094275936432e-05,
+      "loss": 0.0998,
+      "step": 89320
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.1718625474661292e-05,
+      "loss": 0.2966,
+      "step": 89325
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.1718156673386152e-05,
+      "loss": 0.103,
+      "step": 89330
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.1717687872111012e-05,
+      "loss": 0.0213,
+      "step": 89335
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.1717219070835875e-05,
+      "loss": 0.0429,
+      "step": 89340
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.1716750269560735e-05,
+      "loss": 0.0503,
+      "step": 89345
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.1716281468285595e-05,
+      "loss": 0.0322,
+      "step": 89350
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.1715812667010455e-05,
+      "loss": 0.0499,
+      "step": 89355
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.1715343865735317e-05,
+      "loss": 0.1094,
+      "step": 89360
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.1714875064460177e-05,
+      "loss": 0.1059,
+      "step": 89365
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.1714406263185036e-05,
+      "loss": 0.3074,
+      "step": 89370
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.1713937461909896e-05,
+      "loss": 0.2705,
+      "step": 89375
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.1713468660634756e-05,
+      "loss": 0.0501,
+      "step": 89380
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.171299985935962e-05,
+      "loss": 0.0824,
+      "step": 89385
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.171253105808448e-05,
+      "loss": 0.0481,
+      "step": 89390
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.171206225680934e-05,
+      "loss": 0.0364,
+      "step": 89395
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.1711593455534201e-05,
+      "loss": 0.058,
+      "step": 89400
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.1711124654259061e-05,
+      "loss": 0.1301,
+      "step": 89405
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.1710655852983921e-05,
+      "loss": 0.0709,
+      "step": 89410
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.1710187051708781e-05,
+      "loss": 0.1452,
+      "step": 89415
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.1709718250433641e-05,
+      "loss": 0.1915,
+      "step": 89420
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.1709249449158502e-05,
+      "loss": 0.4685,
+      "step": 89425
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.1708780647883362e-05,
+      "loss": 0.092,
+      "step": 89430
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.1708311846608224e-05,
+      "loss": 0.0195,
+      "step": 89435
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.1707843045333086e-05,
+      "loss": 0.0541,
+      "step": 89440
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.1707374244057946e-05,
+      "loss": 0.0734,
+      "step": 89445
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.1706905442782806e-05,
+      "loss": 0.0399,
+      "step": 89450
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.1706436641507665e-05,
+      "loss": 0.0425,
+      "step": 89455
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.1705967840232525e-05,
+      "loss": 0.087,
+      "step": 89460
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.1705499038957387e-05,
+      "loss": 0.0955,
+      "step": 89465
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.1705030237682247e-05,
+      "loss": 0.1549,
+      "step": 89470
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.1704561436407107e-05,
+      "loss": 0.2599,
+      "step": 89475
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.170409263513197e-05,
+      "loss": 0.0578,
+      "step": 89480
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.170362383385683e-05,
+      "loss": 0.0126,
+      "step": 89485
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.170315503258169e-05,
+      "loss": 0.0218,
+      "step": 89490
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.170268623130655e-05,
+      "loss": 0.0482,
+      "step": 89495
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.170221743003141e-05,
+      "loss": 0.0779,
+      "step": 89500
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.1701748628756272e-05,
+      "loss": 0.0651,
+      "step": 89505
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.1701279827481132e-05,
+      "loss": 0.0489,
+      "step": 89510
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.1700811026205991e-05,
+      "loss": 0.0632,
+      "step": 89515
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.1700342224930851e-05,
+      "loss": 0.1013,
+      "step": 89520
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.1699873423655715e-05,
+      "loss": 0.3354,
+      "step": 89525
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.1699404622380575e-05,
+      "loss": 0.0844,
+      "step": 89530
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.1698935821105435e-05,
+      "loss": 0.0415,
+      "step": 89535
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.1698467019830295e-05,
+      "loss": 0.0494,
+      "step": 89540
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.1697998218555156e-05,
+      "loss": 0.0566,
+      "step": 89545
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.1697529417280016e-05,
+      "loss": 0.0648,
+      "step": 89550
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.1697060616004876e-05,
+      "loss": 0.0941,
+      "step": 89555
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.1696591814729736e-05,
+      "loss": 0.1188,
+      "step": 89560
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.1696123013454596e-05,
+      "loss": 0.1267,
+      "step": 89565
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.169565421217946e-05,
+      "loss": 0.184,
+      "step": 89570
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.1695185410904319e-05,
+      "loss": 0.5252,
+      "step": 89575
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.1694716609629179e-05,
+      "loss": 0.0873,
+      "step": 89580
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.169424780835404e-05,
+      "loss": 0.0438,
+      "step": 89585
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.16937790070789e-05,
+      "loss": 0.063,
+      "step": 89590
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.169331020580376e-05,
+      "loss": 0.0517,
+      "step": 89595
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.169284140452862e-05,
+      "loss": 0.0161,
+      "step": 89600
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.169237260325348e-05,
+      "loss": 0.0823,
+      "step": 89605
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.1691903801978342e-05,
+      "loss": 0.0543,
+      "step": 89610
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.1691435000703202e-05,
+      "loss": 0.0828,
+      "step": 89615
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.1690966199428064e-05,
+      "loss": 0.2017,
+      "step": 89620
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.1690497398152925e-05,
+      "loss": 0.3275,
+      "step": 89625
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.1690028596877785e-05,
+      "loss": 0.0912,
+      "step": 89630
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.1689559795602645e-05,
+      "loss": 0.0215,
+      "step": 89635
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.1689090994327505e-05,
+      "loss": 0.0207,
+      "step": 89640
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.1688622193052365e-05,
+      "loss": 0.0594,
+      "step": 89645
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.1688153391777227e-05,
+      "loss": 0.0652,
+      "step": 89650
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.1687684590502087e-05,
+      "loss": 0.0914,
+      "step": 89655
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.1687215789226946e-05,
+      "loss": 0.088,
+      "step": 89660
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.168674698795181e-05,
+      "loss": 0.1658,
+      "step": 89665
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.168627818667667e-05,
+      "loss": 0.1631,
+      "step": 89670
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.168580938540153e-05,
+      "loss": 0.1616,
+      "step": 89675
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.168534058412639e-05,
+      "loss": 0.0536,
+      "step": 89680
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.168487178285125e-05,
+      "loss": 0.0344,
+      "step": 89685
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.1684402981576111e-05,
+      "loss": 0.0607,
+      "step": 89690
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.1683934180300971e-05,
+      "loss": 0.0244,
+      "step": 89695
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.1683465379025831e-05,
+      "loss": 0.096,
+      "step": 89700
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.1682996577750691e-05,
+      "loss": 0.0742,
+      "step": 89705
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.1682527776475554e-05,
+      "loss": 0.1275,
+      "step": 89710
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.1682058975200414e-05,
+      "loss": 0.0448,
+      "step": 89715
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.1681590173925274e-05,
+      "loss": 0.1787,
+      "step": 89720
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.1681121372650134e-05,
+      "loss": 0.3124,
+      "step": 89725
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.1680652571374996e-05,
+      "loss": 0.1043,
+      "step": 89730
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.1680183770099856e-05,
+      "loss": 0.0047,
+      "step": 89735
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.1679714968824716e-05,
+      "loss": 0.0341,
+      "step": 89740
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.1679246167549576e-05,
+      "loss": 0.046,
+      "step": 89745
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.1678777366274437e-05,
+      "loss": 0.0342,
+      "step": 89750
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.1678308564999297e-05,
+      "loss": 0.0527,
+      "step": 89755
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.1677839763724159e-05,
+      "loss": 0.164,
+      "step": 89760
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.1677370962449019e-05,
+      "loss": 0.061,
+      "step": 89765
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.167690216117388e-05,
+      "loss": 0.19,
+      "step": 89770
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.167643335989874e-05,
+      "loss": 0.219,
+      "step": 89775
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.16759645586236e-05,
+      "loss": 0.0776,
+      "step": 89780
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.167549575734846e-05,
+      "loss": 0.0106,
+      "step": 89785
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.1675026956073322e-05,
+      "loss": 0.021,
+      "step": 89790
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.1674558154798182e-05,
+      "loss": 0.033,
+      "step": 89795
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.1674089353523042e-05,
+      "loss": 0.0703,
+      "step": 89800
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.1673620552247903e-05,
+      "loss": 0.1027,
+      "step": 89805
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.1673151750972765e-05,
+      "loss": 0.0881,
+      "step": 89810
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.1672682949697625e-05,
+      "loss": 0.0987,
+      "step": 89815
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.1672214148422485e-05,
+      "loss": 0.2062,
+      "step": 89820
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.1671745347147345e-05,
+      "loss": 0.3137,
+      "step": 89825
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.1671276545872206e-05,
+      "loss": 0.113,
+      "step": 89830
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.1670807744597066e-05,
+      "loss": 0.0464,
+      "step": 89835
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.1670338943321926e-05,
+      "loss": 0.0121,
+      "step": 89840
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.1669870142046786e-05,
+      "loss": 0.0268,
+      "step": 89845
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.166940134077165e-05,
+      "loss": 0.0599,
+      "step": 89850
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.166893253949651e-05,
+      "loss": 0.0611,
+      "step": 89855
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.166846373822137e-05,
+      "loss": 0.0742,
+      "step": 89860
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.166799493694623e-05,
+      "loss": 0.1433,
+      "step": 89865
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.166752613567109e-05,
+      "loss": 0.1798,
+      "step": 89870
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.166705733439595e-05,
+      "loss": 0.2045,
+      "step": 89875
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.166658853312081e-05,
+      "loss": 0.0834,
+      "step": 89880
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.166611973184567e-05,
+      "loss": 0.0339,
+      "step": 89885
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.166565093057053e-05,
+      "loss": 0.0613,
+      "step": 89890
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.1665182129295394e-05,
+      "loss": 0.0092,
+      "step": 89895
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.1664713328020254e-05,
+      "loss": 0.0491,
+      "step": 89900
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.1664244526745114e-05,
+      "loss": 0.0116,
+      "step": 89905
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.1663775725469975e-05,
+      "loss": 0.0667,
+      "step": 89910
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.1663306924194835e-05,
+      "loss": 0.1023,
+      "step": 89915
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.1662838122919695e-05,
+      "loss": 0.1983,
+      "step": 89920
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.1662369321644555e-05,
+      "loss": 0.3117,
+      "step": 89925
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.1661900520369415e-05,
+      "loss": 0.085,
+      "step": 89930
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.1661431719094277e-05,
+      "loss": 0.0078,
+      "step": 89935
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.1660962917819137e-05,
+      "loss": 0.0456,
+      "step": 89940
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.1660494116543998e-05,
+      "loss": 0.0304,
+      "step": 89945
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.166002531526886e-05,
+      "loss": 0.0178,
+      "step": 89950
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.165955651399372e-05,
+      "loss": 0.0747,
+      "step": 89955
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.165908771271858e-05,
+      "loss": 0.0913,
+      "step": 89960
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.165861891144344e-05,
+      "loss": 0.0636,
+      "step": 89965
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.16581501101683e-05,
+      "loss": 0.1351,
+      "step": 89970
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.1657681308893161e-05,
+      "loss": 0.3487,
+      "step": 89975
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.1657212507618021e-05,
+      "loss": 0.0798,
+      "step": 89980
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.1656743706342881e-05,
+      "loss": 0.0109,
+      "step": 89985
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.1656274905067744e-05,
+      "loss": 0.0226,
+      "step": 89990
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.1655806103792604e-05,
+      "loss": 0.0188,
+      "step": 89995
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.1655337302517464e-05,
+      "loss": 0.0779,
+      "step": 90000
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.1654868501242324e-05,
+      "loss": 0.0679,
+      "step": 90005
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.1654399699967184e-05,
+      "loss": 0.081,
+      "step": 90010
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.1653930898692046e-05,
+      "loss": 0.0588,
+      "step": 90015
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.1653462097416906e-05,
+      "loss": 0.2129,
+      "step": 90020
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.1652993296141766e-05,
+      "loss": 0.1604,
+      "step": 90025
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.1652524494866626e-05,
+      "loss": 0.0807,
+      "step": 90030
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.1652055693591489e-05,
+      "loss": 0.0473,
+      "step": 90035
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.1651586892316349e-05,
+      "loss": 0.0271,
+      "step": 90040
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.1651118091041209e-05,
+      "loss": 0.0115,
+      "step": 90045
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.1650649289766069e-05,
+      "loss": 0.0769,
+      "step": 90050
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.165018048849093e-05,
+      "loss": 0.0499,
+      "step": 90055
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.164971168721579e-05,
+      "loss": 0.113,
+      "step": 90060
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.164924288594065e-05,
+      "loss": 0.1396,
+      "step": 90065
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.164877408466551e-05,
+      "loss": 0.1436,
+      "step": 90070
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.164830528339037e-05,
+      "loss": 0.3258,
+      "step": 90075
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.1647836482115232e-05,
+      "loss": 0.0535,
+      "step": 90080
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.1647367680840093e-05,
+      "loss": 0.0282,
+      "step": 90085
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.1646898879564953e-05,
+      "loss": 0.046,
+      "step": 90090
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.1646430078289815e-05,
+      "loss": 0.0267,
+      "step": 90095
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.1645961277014675e-05,
+      "loss": 0.0636,
+      "step": 90100
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.1645492475739535e-05,
+      "loss": 0.1151,
+      "step": 90105
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.1645023674464395e-05,
+      "loss": 0.0741,
+      "step": 90110
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.1644554873189255e-05,
+      "loss": 0.0809,
+      "step": 90115
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.1644086071914116e-05,
+      "loss": 0.1993,
+      "step": 90120
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.1643617270638976e-05,
+      "loss": 0.2685,
+      "step": 90125
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.1643148469363838e-05,
+      "loss": 0.0708,
+      "step": 90130
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.16426796680887e-05,
+      "loss": 0.0637,
+      "step": 90135
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.164221086681356e-05,
+      "loss": 0.0553,
+      "step": 90140
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.164174206553842e-05,
+      "loss": 0.061,
+      "step": 90145
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.164127326426328e-05,
+      "loss": 0.0785,
+      "step": 90150
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.164080446298814e-05,
+      "loss": 0.0905,
+      "step": 90155
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.1640335661713e-05,
+      "loss": 0.0879,
+      "step": 90160
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.163986686043786e-05,
+      "loss": 0.1455,
+      "step": 90165
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.163939805916272e-05,
+      "loss": 0.0863,
+      "step": 90170
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.1638929257887584e-05,
+      "loss": 0.3679,
+      "step": 90175
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.1638460456612444e-05,
+      "loss": 0.0826,
+      "step": 90180
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.1637991655337304e-05,
+      "loss": 0.0402,
+      "step": 90185
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.1637522854062164e-05,
+      "loss": 0.024,
+      "step": 90190
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.1637054052787025e-05,
+      "loss": 0.0203,
+      "step": 90195
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.1636585251511885e-05,
+      "loss": 0.0285,
+      "step": 90200
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.1636116450236745e-05,
+      "loss": 0.123,
+      "step": 90205
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.1635647648961605e-05,
+      "loss": 0.0205,
+      "step": 90210
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.1635178847686465e-05,
+      "loss": 0.1254,
+      "step": 90215
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.1634710046411329e-05,
+      "loss": 0.2179,
+      "step": 90220
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.1634241245136188e-05,
+      "loss": 0.3697,
+      "step": 90225
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.1633772443861048e-05,
+      "loss": 0.1076,
+      "step": 90230
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.163330364258591e-05,
+      "loss": 0.0421,
+      "step": 90235
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.163283484131077e-05,
+      "loss": 0.0345,
+      "step": 90240
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.163236604003563e-05,
+      "loss": 0.0681,
+      "step": 90245
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.163189723876049e-05,
+      "loss": 0.0593,
+      "step": 90250
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.163142843748535e-05,
+      "loss": 0.0574,
+      "step": 90255
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.1630959636210211e-05,
+      "loss": 0.0452,
+      "step": 90260
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.1630490834935071e-05,
+      "loss": 0.1625,
+      "step": 90265
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.1630022033659933e-05,
+      "loss": 0.2212,
+      "step": 90270
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.1629553232384795e-05,
+      "loss": 0.2688,
+      "step": 90275
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.1629084431109654e-05,
+      "loss": 0.0968,
+      "step": 90280
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.1628615629834514e-05,
+      "loss": 0.0236,
+      "step": 90285
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.1628146828559374e-05,
+      "loss": 0.0285,
+      "step": 90290
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.1627678027284234e-05,
+      "loss": 0.1006,
+      "step": 90295
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.1627209226009096e-05,
+      "loss": 0.0232,
+      "step": 90300
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.1626740424733956e-05,
+      "loss": 0.1043,
+      "step": 90305
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.1626271623458816e-05,
+      "loss": 0.1092,
+      "step": 90310
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.1625802822183679e-05,
+      "loss": 0.0979,
+      "step": 90315
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.1625334020908539e-05,
+      "loss": 0.2579,
+      "step": 90320
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.1624865219633399e-05,
+      "loss": 0.4228,
+      "step": 90325
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.1624396418358259e-05,
+      "loss": 0.1107,
+      "step": 90330
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.1623927617083119e-05,
+      "loss": 0.038,
+      "step": 90335
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.162345881580798e-05,
+      "loss": 0.0136,
+      "step": 90340
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.162299001453284e-05,
+      "loss": 0.0173,
+      "step": 90345
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.16225212132577e-05,
+      "loss": 0.0287,
+      "step": 90350
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.162205241198256e-05,
+      "loss": 0.0393,
+      "step": 90355
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.1621583610707424e-05,
+      "loss": 0.0761,
+      "step": 90360
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.1621114809432284e-05,
+      "loss": 0.1793,
+      "step": 90365
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.1620646008157143e-05,
+      "loss": 0.2109,
+      "step": 90370
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.1620177206882003e-05,
+      "loss": 0.2366,
+      "step": 90375
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.1619708405606865e-05,
+      "loss": 0.0787,
+      "step": 90380
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.1619239604331725e-05,
+      "loss": 0.0256,
+      "step": 90385
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.1618770803056585e-05,
+      "loss": 0.0247,
+      "step": 90390
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.1618302001781445e-05,
+      "loss": 0.0377,
+      "step": 90395
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.1617833200506305e-05,
+      "loss": 0.0484,
+      "step": 90400
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.1617364399231166e-05,
+      "loss": 0.082,
+      "step": 90405
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.1616895597956028e-05,
+      "loss": 0.056,
+      "step": 90410
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.1616426796680888e-05,
+      "loss": 0.1347,
+      "step": 90415
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.161595799540575e-05,
+      "loss": 0.1939,
+      "step": 90420
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.161548919413061e-05,
+      "loss": 0.2519,
+      "step": 90425
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.161502039285547e-05,
+      "loss": 0.0694,
+      "step": 90430
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.161455159158033e-05,
+      "loss": 0.0292,
+      "step": 90435
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.161408279030519e-05,
+      "loss": 0.0794,
+      "step": 90440
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.1613613989030051e-05,
+      "loss": 0.0355,
+      "step": 90445
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.1613145187754911e-05,
+      "loss": 0.0542,
+      "step": 90450
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.1612676386479772e-05,
+      "loss": 0.0475,
+      "step": 90455
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.1612207585204634e-05,
+      "loss": 0.1139,
+      "step": 90460
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.1611738783929494e-05,
+      "loss": 0.0833,
+      "step": 90465
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.1611269982654354e-05,
+      "loss": 0.1518,
+      "step": 90470
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.1610801181379214e-05,
+      "loss": 0.2468,
+      "step": 90475
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.1610332380104074e-05,
+      "loss": 0.082,
+      "step": 90480
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.1609863578828935e-05,
+      "loss": 0.0228,
+      "step": 90485
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.1609394777553795e-05,
+      "loss": 0.0282,
+      "step": 90490
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.1608925976278655e-05,
+      "loss": 0.0354,
+      "step": 90495
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.1608457175003519e-05,
+      "loss": 0.0689,
+      "step": 90500
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.1607988373728379e-05,
+      "loss": 0.0947,
+      "step": 90505
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.1607519572453239e-05,
+      "loss": 0.1315,
+      "step": 90510
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.1607050771178098e-05,
+      "loss": 0.088,
+      "step": 90515
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.1606581969902958e-05,
+      "loss": 0.2375,
+      "step": 90520
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.160611316862782e-05,
+      "loss": 0.2542,
+      "step": 90525
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.160564436735268e-05,
+      "loss": 0.022,
+      "step": 90530
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.160517556607754e-05,
+      "loss": 0.0536,
+      "step": 90535
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.16047067648024e-05,
+      "loss": 0.0599,
+      "step": 90540
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.1604237963527263e-05,
+      "loss": 0.0576,
+      "step": 90545
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.1603769162252123e-05,
+      "loss": 0.135,
+      "step": 90550
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.1603300360976983e-05,
+      "loss": 0.028,
+      "step": 90555
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.1602831559701843e-05,
+      "loss": 0.1132,
+      "step": 90560
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.1602362758426705e-05,
+      "loss": 0.1297,
+      "step": 90565
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.1601893957151565e-05,
+      "loss": 0.1117,
+      "step": 90570
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.1601425155876424e-05,
+      "loss": 0.3883,
+      "step": 90575
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.1600956354601284e-05,
+      "loss": 0.0677,
+      "step": 90580
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.1600487553326146e-05,
+      "loss": 0.0066,
+      "step": 90585
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.1600018752051006e-05,
+      "loss": 0.0376,
+      "step": 90590
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.1599549950775868e-05,
+      "loss": 0.0679,
+      "step": 90595
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.1599081149500728e-05,
+      "loss": 0.0643,
+      "step": 90600
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.1598612348225589e-05,
+      "loss": 0.0511,
+      "step": 90605
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.1598143546950449e-05,
+      "loss": 0.0937,
+      "step": 90610
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.1597674745675309e-05,
+      "loss": 0.1283,
+      "step": 90615
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.1597205944400169e-05,
+      "loss": 0.0704,
+      "step": 90620
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.159673714312503e-05,
+      "loss": 0.2658,
+      "step": 90625
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.159626834184989e-05,
+      "loss": 0.0897,
+      "step": 90630
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.159579954057475e-05,
+      "loss": 0.0046,
+      "step": 90635
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.1595330739299612e-05,
+      "loss": 0.0517,
+      "step": 90640
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.1594861938024474e-05,
+      "loss": 0.0436,
+      "step": 90645
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.1594393136749334e-05,
+      "loss": 0.0218,
+      "step": 90650
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.1593924335474194e-05,
+      "loss": 0.0381,
+      "step": 90655
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.1593455534199053e-05,
+      "loss": 0.0627,
+      "step": 90660
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.1592986732923915e-05,
+      "loss": 0.1014,
+      "step": 90665
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.1592517931648775e-05,
+      "loss": 0.0805,
+      "step": 90670
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.1592049130373635e-05,
+      "loss": 0.2807,
+      "step": 90675
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.1591580329098495e-05,
+      "loss": 0.0884,
+      "step": 90680
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.1591111527823358e-05,
+      "loss": 0.0178,
+      "step": 90685
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.1590642726548218e-05,
+      "loss": 0.0513,
+      "step": 90690
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.1590173925273078e-05,
+      "loss": 0.0441,
+      "step": 90695
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.1589705123997938e-05,
+      "loss": 0.0771,
+      "step": 90700
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.15892363227228e-05,
+      "loss": 0.0506,
+      "step": 90705
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.158876752144766e-05,
+      "loss": 0.1414,
+      "step": 90710
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.158829872017252e-05,
+      "loss": 0.0922,
+      "step": 90715
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.158782991889738e-05,
+      "loss": 0.1986,
+      "step": 90720
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.158736111762224e-05,
+      "loss": 0.2862,
+      "step": 90725
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.1586892316347101e-05,
+      "loss": 0.1357,
+      "step": 90730
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.1586423515071963e-05,
+      "loss": 0.0374,
+      "step": 90735
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.1585954713796823e-05,
+      "loss": 0.0397,
+      "step": 90740
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.1585485912521684e-05,
+      "loss": 0.0721,
+      "step": 90745
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.1585017111246544e-05,
+      "loss": 0.0833,
+      "step": 90750
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.1584548309971404e-05,
+      "loss": 0.0783,
+      "step": 90755
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.1584079508696264e-05,
+      "loss": 0.1488,
+      "step": 90760
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1583610707421124e-05,
+      "loss": 0.1002,
+      "step": 90765
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1583141906145986e-05,
+      "loss": 0.2139,
+      "step": 90770
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1582673104870846e-05,
+      "loss": 0.3166,
+      "step": 90775
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1582204303595707e-05,
+      "loss": 0.1548,
+      "step": 90780
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1581735502320569e-05,
+      "loss": 0.017,
+      "step": 90785
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1581266701045429e-05,
+      "loss": 0.0412,
+      "step": 90790
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1580797899770289e-05,
+      "loss": 0.0272,
+      "step": 90795
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1580329098495149e-05,
+      "loss": 0.0953,
+      "step": 90800
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1579860297220008e-05,
+      "loss": 0.2096,
+      "step": 90805
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.157939149594487e-05,
+      "loss": 0.0589,
+      "step": 90810
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.157892269466973e-05,
+      "loss": 0.1322,
+      "step": 90815
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.157845389339459e-05,
+      "loss": 0.2027,
+      "step": 90820
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1577985092119453e-05,
+      "loss": 0.2512,
+      "step": 90825
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1577516290844313e-05,
+      "loss": 0.0679,
+      "step": 90830
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1577047489569173e-05,
+      "loss": 0.0326,
+      "step": 90835
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1576578688294033e-05,
+      "loss": 0.041,
+      "step": 90840
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1576109887018893e-05,
+      "loss": 0.0604,
+      "step": 90845
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1575641085743755e-05,
+      "loss": 0.0379,
+      "step": 90850
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1575172284468615e-05,
+      "loss": 0.0647,
+      "step": 90855
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1574703483193475e-05,
+      "loss": 0.0726,
+      "step": 90860
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1574234681918334e-05,
+      "loss": 0.0912,
+      "step": 90865
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1573765880643198e-05,
+      "loss": 0.0761,
+      "step": 90870
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1573297079368058e-05,
+      "loss": 0.1042,
+      "step": 90875
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1572828278092918e-05,
+      "loss": 0.0479,
+      "step": 90880
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1572359476817778e-05,
+      "loss": 0.0345,
+      "step": 90885
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.157189067554264e-05,
+      "loss": 0.0204,
+      "step": 90890
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.15714218742675e-05,
+      "loss": 0.0444,
+      "step": 90895
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1570953072992359e-05,
+      "loss": 0.0423,
+      "step": 90900
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1570484271717219e-05,
+      "loss": 0.0727,
+      "step": 90905
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1570015470442079e-05,
+      "loss": 0.1087,
+      "step": 90910
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.156954666916694e-05,
+      "loss": 0.0887,
+      "step": 90915
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1569077867891802e-05,
+      "loss": 0.1516,
+      "step": 90920
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1568609066616662e-05,
+      "loss": 0.3522,
+      "step": 90925
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1568140265341524e-05,
+      "loss": 0.0869,
+      "step": 90930
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1567671464066384e-05,
+      "loss": 0.0037,
+      "step": 90935
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1567202662791244e-05,
+      "loss": 0.0325,
+      "step": 90940
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1566733861516104e-05,
+      "loss": 0.0397,
+      "step": 90945
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1566265060240964e-05,
+      "loss": 0.0766,
+      "step": 90950
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1565796258965825e-05,
+      "loss": 0.0681,
+      "step": 90955
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1565327457690685e-05,
+      "loss": 0.0819,
+      "step": 90960
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1564858656415547e-05,
+      "loss": 0.1397,
+      "step": 90965
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1564389855140408e-05,
+      "loss": 0.1983,
+      "step": 90970
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1563921053865268e-05,
+      "loss": 0.2272,
+      "step": 90975
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1563452252590128e-05,
+      "loss": 0.075,
+      "step": 90980
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1562983451314988e-05,
+      "loss": 0.084,
+      "step": 90985
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1562514650039848e-05,
+      "loss": 0.0441,
+      "step": 90990
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.156204584876471e-05,
+      "loss": 0.1087,
+      "step": 90995
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.156157704748957e-05,
+      "loss": 0.0661,
+      "step": 91000
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.156110824621443e-05,
+      "loss": 0.1206,
+      "step": 91005
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1560639444939293e-05,
+      "loss": 0.0591,
+      "step": 91010
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1560170643664153e-05,
+      "loss": 0.1529,
+      "step": 91015
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1559701842389013e-05,
+      "loss": 0.1996,
+      "step": 91020
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1559233041113873e-05,
+      "loss": 0.3524,
+      "step": 91025
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1558764239838733e-05,
+      "loss": 0.0991,
+      "step": 91030
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1558295438563594e-05,
+      "loss": 0.0129,
+      "step": 91035
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1557826637288454e-05,
+      "loss": 0.0498,
+      "step": 91040
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1557357836013314e-05,
+      "loss": 0.0546,
+      "step": 91045
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1556889034738174e-05,
+      "loss": 0.0598,
+      "step": 91050
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1556420233463036e-05,
+      "loss": 0.0731,
+      "step": 91055
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1555951432187897e-05,
+      "loss": 0.1104,
+      "step": 91060
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1555482630912757e-05,
+      "loss": 0.1065,
+      "step": 91065
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1555013829637617e-05,
+      "loss": 0.243,
+      "step": 91070
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1554545028362479e-05,
+      "loss": 0.2123,
+      "step": 91075
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1554076227087339e-05,
+      "loss": 0.0422,
+      "step": 91080
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1553607425812199e-05,
+      "loss": 0.0264,
+      "step": 91085
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1553138624537059e-05,
+      "loss": 0.0343,
+      "step": 91090
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.155266982326192e-05,
+      "loss": 0.0504,
+      "step": 91095
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.155220102198678e-05,
+      "loss": 0.0373,
+      "step": 91100
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1551732220711642e-05,
+      "loss": 0.0501,
+      "step": 91105
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1551263419436502e-05,
+      "loss": 0.0663,
+      "step": 91110
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1550794618161363e-05,
+      "loss": 0.0887,
+      "step": 91115
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1550325816886223e-05,
+      "loss": 0.1535,
+      "step": 91120
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1549857015611083e-05,
+      "loss": 0.2512,
+      "step": 91125
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1549388214335943e-05,
+      "loss": 0.0946,
+      "step": 91130
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1548919413060805e-05,
+      "loss": 0.0098,
+      "step": 91135
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1548450611785665e-05,
+      "loss": 0.0319,
+      "step": 91140
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1547981810510525e-05,
+      "loss": 0.0227,
+      "step": 91145
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1547513009235386e-05,
+      "loss": 0.0815,
+      "step": 91150
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1547044207960248e-05,
+      "loss": 0.0746,
+      "step": 91155
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1546575406685108e-05,
+      "loss": 0.1372,
+      "step": 91160
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1546106605409968e-05,
+      "loss": 0.0507,
+      "step": 91165
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1545637804134828e-05,
+      "loss": 0.2027,
+      "step": 91170
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.154516900285969e-05,
+      "loss": 0.1872,
+      "step": 91175
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.154470020158455e-05,
+      "loss": 0.0551,
+      "step": 91180
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.154423140030941e-05,
+      "loss": 0.046,
+      "step": 91185
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1543762599034269e-05,
+      "loss": 0.028,
+      "step": 91190
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1543293797759132e-05,
+      "loss": 0.0266,
+      "step": 91195
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1542824996483992e-05,
+      "loss": 0.067,
+      "step": 91200
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1542356195208852e-05,
+      "loss": 0.0767,
+      "step": 91205
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1541887393933712e-05,
+      "loss": 0.0836,
+      "step": 91210
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1541418592658574e-05,
+      "loss": 0.1233,
+      "step": 91215
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1540949791383434e-05,
+      "loss": 0.1101,
+      "step": 91220
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1540480990108294e-05,
+      "loss": 0.1987,
+      "step": 91225
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1540012188833154e-05,
+      "loss": 0.057,
+      "step": 91230
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1539543387558014e-05,
+      "loss": 0.0362,
+      "step": 91235
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1539074586282875e-05,
+      "loss": 0.0245,
+      "step": 91240
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1538605785007737e-05,
+      "loss": 0.0242,
+      "step": 91245
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1538136983732597e-05,
+      "loss": 0.1013,
+      "step": 91250
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1537668182457458e-05,
+      "loss": 0.0648,
+      "step": 91255
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1537199381182318e-05,
+      "loss": 0.0399,
+      "step": 91260
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1536730579907178e-05,
+      "loss": 0.0487,
+      "step": 91265
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1536261778632038e-05,
+      "loss": 0.2504,
+      "step": 91270
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1535792977356898e-05,
+      "loss": 0.151,
+      "step": 91275
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.153532417608176e-05,
+      "loss": 0.0723,
+      "step": 91280
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.153485537480662e-05,
+      "loss": 0.0751,
+      "step": 91285
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1534386573531481e-05,
+      "loss": 0.0415,
+      "step": 91290
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1533917772256343e-05,
+      "loss": 0.0405,
+      "step": 91295
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1533448970981203e-05,
+      "loss": 0.0208,
+      "step": 91300
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1532980169706063e-05,
+      "loss": 0.025,
+      "step": 91305
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1532511368430923e-05,
+      "loss": 0.0711,
+      "step": 91310
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1532042567155783e-05,
+      "loss": 0.0693,
+      "step": 91315
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1531573765880644e-05,
+      "loss": 0.2113,
+      "step": 91320
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1531104964605504e-05,
+      "loss": 0.1946,
+      "step": 91325
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1530636163330364e-05,
+      "loss": 0.0494,
+      "step": 91330
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1530167362055228e-05,
+      "loss": 0.0235,
+      "step": 91335
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1529698560780087e-05,
+      "loss": 0.0209,
+      "step": 91340
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1529229759504947e-05,
+      "loss": 0.0524,
+      "step": 91345
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1528760958229807e-05,
+      "loss": 0.0715,
+      "step": 91350
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1528292156954667e-05,
+      "loss": 0.0529,
+      "step": 91355
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1527823355679529e-05,
+      "loss": 0.0976,
+      "step": 91360
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1527354554404389e-05,
+      "loss": 0.2976,
+      "step": 91365
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1526885753129249e-05,
+      "loss": 0.1699,
+      "step": 91370
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1526416951854109e-05,
+      "loss": 0.1805,
+      "step": 91375
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1525948150578969e-05,
+      "loss": 0.0969,
+      "step": 91380
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1525479349303832e-05,
+      "loss": 0.0386,
+      "step": 91385
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1525010548028692e-05,
+      "loss": 0.0124,
+      "step": 91390
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1524541746753552e-05,
+      "loss": 0.081,
+      "step": 91395
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1524072945478413e-05,
+      "loss": 0.0536,
+      "step": 91400
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1523604144203273e-05,
+      "loss": 0.0579,
+      "step": 91405
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1523135342928133e-05,
+      "loss": 0.0607,
+      "step": 91410
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1522666541652993e-05,
+      "loss": 0.1101,
+      "step": 91415
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1522197740377853e-05,
+      "loss": 0.2055,
+      "step": 91420
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1521728939102715e-05,
+      "loss": 0.284,
+      "step": 91425
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1521260137827576e-05,
+      "loss": 0.1254,
+      "step": 91430
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1520791336552436e-05,
+      "loss": 0.0306,
+      "step": 91435
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1520322535277298e-05,
+      "loss": 0.0168,
+      "step": 91440
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1519853734002158e-05,
+      "loss": 0.059,
+      "step": 91445
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1519384932727018e-05,
+      "loss": 0.0459,
+      "step": 91450
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1518916131451878e-05,
+      "loss": 0.069,
+      "step": 91455
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1518447330176738e-05,
+      "loss": 0.0843,
+      "step": 91460
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.15179785289016e-05,
+      "loss": 0.5684,
+      "step": 91465
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.151750972762646e-05,
+      "loss": 0.1952,
+      "step": 91470
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1517040926351321e-05,
+      "loss": 0.3945,
+      "step": 91475
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1516572125076183e-05,
+      "loss": 0.04,
+      "step": 91480
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1516103323801042e-05,
+      "loss": 0.0278,
+      "step": 91485
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1515634522525902e-05,
+      "loss": 0.0286,
+      "step": 91490
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1515165721250762e-05,
+      "loss": 0.0333,
+      "step": 91495
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1514696919975622e-05,
+      "loss": 0.084,
+      "step": 91500
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1514228118700484e-05,
+      "loss": 0.0698,
+      "step": 91505
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1513759317425344e-05,
+      "loss": 0.0549,
+      "step": 91510
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1513290516150204e-05,
+      "loss": 0.0892,
+      "step": 91515
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1512821714875067e-05,
+      "loss": 0.1317,
+      "step": 91520
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1512352913599927e-05,
+      "loss": 0.2714,
+      "step": 91525
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1511884112324787e-05,
+      "loss": 0.0958,
+      "step": 91530
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1511415311049647e-05,
+      "loss": 0.0136,
+      "step": 91535
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1510946509774507e-05,
+      "loss": 0.0703,
+      "step": 91540
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1510477708499368e-05,
+      "loss": 0.0202,
+      "step": 91545
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1510008907224228e-05,
+      "loss": 0.0117,
+      "step": 91550
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1509540105949088e-05,
+      "loss": 0.0474,
+      "step": 91555
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1509071304673948e-05,
+      "loss": 0.0763,
+      "step": 91560
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.150860250339881e-05,
+      "loss": 0.1485,
+      "step": 91565
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1508133702123672e-05,
+      "loss": 0.09,
+      "step": 91570
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1507664900848531e-05,
+      "loss": 0.2797,
+      "step": 91575
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1507196099573391e-05,
+      "loss": 0.1036,
+      "step": 91580
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1506727298298253e-05,
+      "loss": 0.0242,
+      "step": 91585
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1506258497023113e-05,
+      "loss": 0.0397,
+      "step": 91590
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1505789695747973e-05,
+      "loss": 0.0304,
+      "step": 91595
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1505320894472833e-05,
+      "loss": 0.0617,
+      "step": 91600
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1504852093197694e-05,
+      "loss": 0.0714,
+      "step": 91605
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1504383291922554e-05,
+      "loss": 0.0834,
+      "step": 91610
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1503914490647416e-05,
+      "loss": 0.1014,
+      "step": 91615
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.1503445689372278e-05,
+      "loss": 0.1623,
+      "step": 91620
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.1502976888097138e-05,
+      "loss": 0.2311,
+      "step": 91625
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.1502508086821998e-05,
+      "loss": 0.0357,
+      "step": 91630
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.1502039285546857e-05,
+      "loss": 0.0253,
+      "step": 91635
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.1501570484271717e-05,
+      "loss": 0.0131,
+      "step": 91640
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.1501101682996579e-05,
+      "loss": 0.0304,
+      "step": 91645
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.1500632881721439e-05,
+      "loss": 0.0276,
+      "step": 91650
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.1500164080446299e-05,
+      "loss": 0.0578,
+      "step": 91655
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.1499695279171162e-05,
+      "loss": 0.0585,
+      "step": 91660
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.1499226477896022e-05,
+      "loss": 0.116,
+      "step": 91665
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.1498757676620882e-05,
+      "loss": 0.1421,
+      "step": 91670
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.1498288875345742e-05,
+      "loss": 0.2871,
+      "step": 91675
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.1497820074070602e-05,
+      "loss": 0.0859,
+      "step": 91680
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.1497351272795464e-05,
+      "loss": 0.0371,
+      "step": 91685
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.1496882471520323e-05,
+      "loss": 0.0224,
+      "step": 91690
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.1496413670245183e-05,
+      "loss": 0.0406,
+      "step": 91695
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.1495944868970043e-05,
+      "loss": 0.0528,
+      "step": 91700
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.1495476067694903e-05,
+      "loss": 0.081,
+      "step": 91705
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.1495007266419767e-05,
+      "loss": 0.0952,
+      "step": 91710
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.1494538465144627e-05,
+      "loss": 0.195,
+      "step": 91715
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.1494069663869486e-05,
+      "loss": 0.1552,
+      "step": 91720
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.1493600862594348e-05,
+      "loss": 0.3942,
+      "step": 91725
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.1493132061319208e-05,
+      "loss": 0.1197,
+      "step": 91730
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.1492663260044068e-05,
+      "loss": 0.0155,
+      "step": 91735
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.1492194458768928e-05,
+      "loss": 0.0613,
+      "step": 91740
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.1491725657493788e-05,
+      "loss": 0.0329,
+      "step": 91745
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.149125685621865e-05,
+      "loss": 0.0667,
+      "step": 91750
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.1490788054943511e-05,
+      "loss": 0.0855,
+      "step": 91755
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.1490319253668371e-05,
+      "loss": 0.0843,
+      "step": 91760
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.1489850452393233e-05,
+      "loss": 0.0893,
+      "step": 91765
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.1489381651118093e-05,
+      "loss": 0.1834,
+      "step": 91770
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.1488912849842953e-05,
+      "loss": 0.212,
+      "step": 91775
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.1488444048567812e-05,
+      "loss": 0.0486,
+      "step": 91780
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.1487975247292672e-05,
+      "loss": 0.0235,
+      "step": 91785
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.1487506446017534e-05,
+      "loss": 0.0296,
+      "step": 91790
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.1487037644742394e-05,
+      "loss": 0.0367,
+      "step": 91795
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.1486568843467256e-05,
+      "loss": 0.0374,
+      "step": 91800
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.1486100042192117e-05,
+      "loss": 0.1294,
+      "step": 91805
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.1485631240916977e-05,
+      "loss": 0.0709,
+      "step": 91810
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.1485162439641837e-05,
+      "loss": 0.1642,
+      "step": 91815
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.1484693638366697e-05,
+      "loss": 0.1697,
+      "step": 91820
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.1484224837091557e-05,
+      "loss": 0.2103,
+      "step": 91825
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.1483756035816419e-05,
+      "loss": 0.0815,
+      "step": 91830
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.1483287234541278e-05,
+      "loss": 0.0419,
+      "step": 91835
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.1482818433266138e-05,
+      "loss": 0.0401,
+      "step": 91840
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.1482349631991002e-05,
+      "loss": 0.1041,
+      "step": 91845
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.1481880830715862e-05,
+      "loss": 0.0594,
+      "step": 91850
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.1481412029440722e-05,
+      "loss": 0.049,
+      "step": 91855
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.1480943228165582e-05,
+      "loss": 0.055,
+      "step": 91860
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.1480474426890441e-05,
+      "loss": 0.1907,
+      "step": 91865
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.1480005625615303e-05,
+      "loss": 0.1978,
+      "step": 91870
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.1479536824340163e-05,
+      "loss": 0.2849,
+      "step": 91875
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.1479068023065023e-05,
+      "loss": 0.0433,
+      "step": 91880
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.1478599221789883e-05,
+      "loss": 0.0516,
+      "step": 91885
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.1478130420514743e-05,
+      "loss": 0.0232,
+      "step": 91890
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.1477661619239606e-05,
+      "loss": 0.0767,
+      "step": 91895
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.1477192817964466e-05,
+      "loss": 0.0341,
+      "step": 91900
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.1476724016689326e-05,
+      "loss": 0.0907,
+      "step": 91905
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.1476255215414188e-05,
+      "loss": 0.1072,
+      "step": 91910
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.1475786414139048e-05,
+      "loss": 0.1205,
+      "step": 91915
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.1475317612863908e-05,
+      "loss": 0.1197,
+      "step": 91920
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.1474848811588767e-05,
+      "loss": 0.1426,
+      "step": 91925
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.1474380010313629e-05,
+      "loss": 0.1108,
+      "step": 91930
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.1473911209038489e-05,
+      "loss": 0.0342,
+      "step": 91935
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.147344240776335e-05,
+      "loss": 0.029,
+      "step": 91940
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.147297360648821e-05,
+      "loss": 0.0302,
+      "step": 91945
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.1472504805213072e-05,
+      "loss": 0.1783,
+      "step": 91950
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.1472036003937932e-05,
+      "loss": 0.0747,
+      "step": 91955
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.1471567202662792e-05,
+      "loss": 0.0703,
+      "step": 91960
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.1471098401387652e-05,
+      "loss": 0.0686,
+      "step": 91965
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.1470629600112514e-05,
+      "loss": 0.2284,
+      "step": 91970
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.1470160798837374e-05,
+      "loss": 0.2897,
+      "step": 91975
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.1469691997562234e-05,
+      "loss": 0.0665,
+      "step": 91980
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.1469223196287095e-05,
+      "loss": 0.0452,
+      "step": 91985
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.1468754395011957e-05,
+      "loss": 0.0176,
+      "step": 91990
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.1468285593736817e-05,
+      "loss": 0.0459,
+      "step": 91995
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.1467816792461677e-05,
+      "loss": 0.0803,
+      "step": 92000
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.1467347991186537e-05,
+      "loss": 0.1029,
+      "step": 92005
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.1466879189911398e-05,
+      "loss": 0.118,
+      "step": 92010
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.1466410388636258e-05,
+      "loss": 0.2179,
+      "step": 92015
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.1465941587361118e-05,
+      "loss": 0.1018,
+      "step": 92020
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.1465472786085978e-05,
+      "loss": 0.1854,
+      "step": 92025
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.1465003984810838e-05,
+      "loss": 0.0795,
+      "step": 92030
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.1464535183535701e-05,
+      "loss": 0.0179,
+      "step": 92035
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.1464066382260561e-05,
+      "loss": 0.0337,
+      "step": 92040
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.1463597580985421e-05,
+      "loss": 0.0106,
+      "step": 92045
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.1463128779710283e-05,
+      "loss": 0.0376,
+      "step": 92050
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.1462659978435143e-05,
+      "loss": 0.0449,
+      "step": 92055
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.1462191177160003e-05,
+      "loss": 0.0672,
+      "step": 92060
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.1461722375884863e-05,
+      "loss": 0.088,
+      "step": 92065
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.1461253574609722e-05,
+      "loss": 0.1922,
+      "step": 92070
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.1460784773334584e-05,
+      "loss": 0.2762,
+      "step": 92075
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.1460315972059446e-05,
+      "loss": 0.0693,
+      "step": 92080
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.1459847170784306e-05,
+      "loss": 0.027,
+      "step": 92085
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.1459378369509167e-05,
+      "loss": 0.0088,
+      "step": 92090
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.1458909568234027e-05,
+      "loss": 0.0344,
+      "step": 92095
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.1458440766958887e-05,
+      "loss": 0.0979,
+      "step": 92100
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.1457971965683747e-05,
+      "loss": 0.0436,
+      "step": 92105
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.1457503164408607e-05,
+      "loss": 0.0458,
+      "step": 92110
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.1457034363133469e-05,
+      "loss": 0.0857,
+      "step": 92115
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.1456565561858329e-05,
+      "loss": 0.1536,
+      "step": 92120
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.145609676058319e-05,
+      "loss": 0.3376,
+      "step": 92125
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.1455627959308052e-05,
+      "loss": 0.074,
+      "step": 92130
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.1455159158032912e-05,
+      "loss": 0.0353,
+      "step": 92135
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.1454690356757772e-05,
+      "loss": 0.0516,
+      "step": 92140
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.1454221555482632e-05,
+      "loss": 0.0492,
+      "step": 92145
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.1453752754207492e-05,
+      "loss": 0.0188,
+      "step": 92150
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.1453283952932353e-05,
+      "loss": 0.0925,
+      "step": 92155
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.1452815151657213e-05,
+      "loss": 0.05,
+      "step": 92160
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.1452346350382073e-05,
+      "loss": 0.1509,
+      "step": 92165
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.1451877549106936e-05,
+      "loss": 0.1062,
+      "step": 92170
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.1451408747831796e-05,
+      "loss": 0.2546,
+      "step": 92175
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.1450939946556656e-05,
+      "loss": 0.0603,
+      "step": 92180
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.1450471145281516e-05,
+      "loss": 0.0328,
+      "step": 92185
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.1450002344006376e-05,
+      "loss": 0.0494,
+      "step": 92190
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.1449533542731238e-05,
+      "loss": 0.0678,
+      "step": 92195
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.1449064741456098e-05,
+      "loss": 0.0348,
+      "step": 92200
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.1448595940180958e-05,
+      "loss": 0.0759,
+      "step": 92205
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.1448127138905818e-05,
+      "loss": 0.0946,
+      "step": 92210
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.1447658337630677e-05,
+      "loss": 0.0649,
+      "step": 92215
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.144718953635554e-05,
+      "loss": 0.1014,
+      "step": 92220
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.14467207350804e-05,
+      "loss": 0.2581,
+      "step": 92225
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.144625193380526e-05,
+      "loss": 0.1092,
+      "step": 92230
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.1445783132530122e-05,
+      "loss": 0.0297,
+      "step": 92235
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.1445314331254982e-05,
+      "loss": 0.0514,
+      "step": 92240
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.1444845529979842e-05,
+      "loss": 0.0675,
+      "step": 92245
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.1444376728704702e-05,
+      "loss": 0.1109,
+      "step": 92250
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.1443907927429562e-05,
+      "loss": 0.0423,
+      "step": 92255
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.1443439126154424e-05,
+      "loss": 0.0663,
+      "step": 92260
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.1442970324879285e-05,
+      "loss": 0.1463,
+      "step": 92265
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.1442501523604145e-05,
+      "loss": 0.1144,
+      "step": 92270
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.1442032722329007e-05,
+      "loss": 0.2856,
+      "step": 92275
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.1441563921053867e-05,
+      "loss": 0.0755,
+      "step": 92280
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.1441095119778727e-05,
+      "loss": 0.0132,
+      "step": 92285
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.1440626318503587e-05,
+      "loss": 0.0207,
+      "step": 92290
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.1440157517228447e-05,
+      "loss": 0.0565,
+      "step": 92295
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.1439688715953308e-05,
+      "loss": 0.0273,
+      "step": 92300
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.1439219914678168e-05,
+      "loss": 0.0751,
+      "step": 92305
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.143875111340303e-05,
+      "loss": 0.0303,
+      "step": 92310
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.1438282312127891e-05,
+      "loss": 0.1336,
+      "step": 92315
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.1437813510852751e-05,
+      "loss": 0.0892,
+      "step": 92320
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.1437344709577611e-05,
+      "loss": 0.262,
+      "step": 92325
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.1436875908302471e-05,
+      "loss": 0.0757,
+      "step": 92330
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.1436407107027331e-05,
+      "loss": 0.0104,
+      "step": 92335
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.1435938305752193e-05,
+      "loss": 0.0194,
+      "step": 92340
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.1435469504477053e-05,
+      "loss": 0.0316,
+      "step": 92345
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.1435000703201913e-05,
+      "loss": 0.0225,
+      "step": 92350
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.1434531901926773e-05,
+      "loss": 0.1253,
+      "step": 92355
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.1434063100651636e-05,
+      "loss": 0.1119,
+      "step": 92360
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.1433594299376496e-05,
+      "loss": 0.1393,
+      "step": 92365
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.1433125498101356e-05,
+      "loss": 0.1641,
+      "step": 92370
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.1432656696826216e-05,
+      "loss": 0.4148,
+      "step": 92375
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.1432187895551077e-05,
+      "loss": 0.0839,
+      "step": 92380
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.1431719094275937e-05,
+      "loss": 0.0213,
+      "step": 92385
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.1431250293000797e-05,
+      "loss": 0.0586,
+      "step": 92390
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.1430781491725657e-05,
+      "loss": 0.0278,
+      "step": 92395
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.1430312690450519e-05,
+      "loss": 0.0343,
+      "step": 92400
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.142984388917538e-05,
+      "loss": 0.0383,
+      "step": 92405
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.142937508790024e-05,
+      "loss": 0.0766,
+      "step": 92410
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.14289062866251e-05,
+      "loss": 0.0804,
+      "step": 92415
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.1428437485349962e-05,
+      "loss": 0.0692,
+      "step": 92420
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.1427968684074822e-05,
+      "loss": 0.1649,
+      "step": 92425
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.1427499882799682e-05,
+      "loss": 0.0648,
+      "step": 92430
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.1427031081524542e-05,
+      "loss": 0.0405,
+      "step": 92435
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.1426562280249403e-05,
+      "loss": 0.0459,
+      "step": 92440
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.1426093478974263e-05,
+      "loss": 0.0534,
+      "step": 92445
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.1425624677699125e-05,
+      "loss": 0.0571,
+      "step": 92450
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.1425155876423985e-05,
+      "loss": 0.1281,
+      "step": 92455
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.1424687075148846e-05,
+      "loss": 0.1047,
+      "step": 92460
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.1424218273873706e-05,
+      "loss": 0.1761,
+      "step": 92465
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.1423749472598566e-05,
+      "loss": 0.2301,
+      "step": 92470
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.1423280671323426e-05,
+      "loss": 0.2127,
+      "step": 92475
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.1422811870048288e-05,
+      "loss": 0.0692,
+      "step": 92480
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.1422343068773148e-05,
+      "loss": 0.0307,
+      "step": 92485
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.1421874267498008e-05,
+      "loss": 0.0709,
+      "step": 92490
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.142140546622287e-05,
+      "loss": 0.0676,
+      "step": 92495
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.1420936664947731e-05,
+      "loss": 0.0358,
+      "step": 92500
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.1420467863672591e-05,
+      "loss": 0.109,
+      "step": 92505
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.1419999062397451e-05,
+      "loss": 0.1057,
+      "step": 92510
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.141953026112231e-05,
+      "loss": 0.1385,
+      "step": 92515
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.1419061459847172e-05,
+      "loss": 0.1163,
+      "step": 92520
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.1418592658572032e-05,
+      "loss": 0.2456,
+      "step": 92525
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.1418123857296892e-05,
+      "loss": 0.0923,
+      "step": 92530
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.1417655056021752e-05,
+      "loss": 0.0199,
+      "step": 92535
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.1417186254746612e-05,
+      "loss": 0.0505,
+      "step": 92540
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.1416717453471475e-05,
+      "loss": 0.0625,
+      "step": 92545
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.1416248652196335e-05,
+      "loss": 0.0228,
+      "step": 92550
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.1415779850921195e-05,
+      "loss": 0.0506,
+      "step": 92555
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.1415311049646057e-05,
+      "loss": 0.0614,
+      "step": 92560
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.1414842248370917e-05,
+      "loss": 0.1146,
+      "step": 92565
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.1414373447095777e-05,
+      "loss": 0.1423,
+      "step": 92570
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.1413904645820637e-05,
+      "loss": 0.1711,
+      "step": 92575
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.1413435844545497e-05,
+      "loss": 0.0486,
+      "step": 92580
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.1412967043270358e-05,
+      "loss": 0.0206,
+      "step": 92585
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.141249824199522e-05,
+      "loss": 0.0247,
+      "step": 92590
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.141202944072008e-05,
+      "loss": 0.0561,
+      "step": 92595
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.1411560639444942e-05,
+      "loss": 0.0752,
+      "step": 92600
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.1411091838169801e-05,
+      "loss": 0.0944,
+      "step": 92605
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.1410623036894661e-05,
+      "loss": 0.0962,
+      "step": 92610
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.1410154235619521e-05,
+      "loss": 0.1155,
+      "step": 92615
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.1409685434344381e-05,
+      "loss": 0.1644,
+      "step": 92620
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.1409216633069243e-05,
+      "loss": 0.3262,
+      "step": 92625
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.1408747831794103e-05,
+      "loss": 0.0713,
+      "step": 92630
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.1408279030518964e-05,
+      "loss": 0.029,
+      "step": 92635
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.1407810229243826e-05,
+      "loss": 0.0507,
+      "step": 92640
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.1407341427968686e-05,
+      "loss": 0.0661,
+      "step": 92645
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.1406872626693546e-05,
+      "loss": 0.0631,
+      "step": 92650
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.1406403825418406e-05,
+      "loss": 0.0711,
+      "step": 92655
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.1405935024143266e-05,
+      "loss": 0.1305,
+      "step": 92660
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.1405466222868127e-05,
+      "loss": 0.1408,
+      "step": 92665
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.1404997421592987e-05,
+      "loss": 0.2081,
+      "step": 92670
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.1404528620317847e-05,
+      "loss": 0.2834,
+      "step": 92675
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.1404059819042707e-05,
+      "loss": 0.0993,
+      "step": 92680
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.140359101776757e-05,
+      "loss": 0.0104,
+      "step": 92685
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.140312221649243e-05,
+      "loss": 0.0497,
+      "step": 92690
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.140265341521729e-05,
+      "loss": 0.0498,
+      "step": 92695
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.140218461394215e-05,
+      "loss": 0.0869,
+      "step": 92700
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.1401715812667012e-05,
+      "loss": 0.0429,
+      "step": 92705
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.1401247011391872e-05,
+      "loss": 0.0689,
+      "step": 92710
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.1400778210116732e-05,
+      "loss": 0.156,
+      "step": 92715
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.1400309408841592e-05,
+      "loss": 0.1198,
+      "step": 92720
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.1399840607566452e-05,
+      "loss": 0.3197,
+      "step": 92725
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.1399371806291315e-05,
+      "loss": 0.0663,
+      "step": 92730
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.1398903005016175e-05,
+      "loss": 0.0414,
+      "step": 92735
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.1398434203741035e-05,
+      "loss": 0.039,
+      "step": 92740
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.1397965402465897e-05,
+      "loss": 0.0337,
+      "step": 92745
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.1397496601190756e-05,
+      "loss": 0.0979,
+      "step": 92750
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.1397027799915616e-05,
+      "loss": 0.0924,
+      "step": 92755
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.1396558998640476e-05,
+      "loss": 0.1765,
+      "step": 92760
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.1396090197365336e-05,
+      "loss": 0.0846,
+      "step": 92765
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.1395621396090198e-05,
+      "loss": 0.1251,
+      "step": 92770
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.139515259481506e-05,
+      "loss": 0.2777,
+      "step": 92775
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.139468379353992e-05,
+      "loss": 0.0719,
+      "step": 92780
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.1394214992264781e-05,
+      "loss": 0.0334,
+      "step": 92785
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.1393746190989641e-05,
+      "loss": 0.0253,
+      "step": 92790
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.1393277389714501e-05,
+      "loss": 0.0627,
+      "step": 92795
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.1392808588439361e-05,
+      "loss": 0.0869,
+      "step": 92800
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.139233978716422e-05,
+      "loss": 0.0711,
+      "step": 92805
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.1391870985889082e-05,
+      "loss": 0.1426,
+      "step": 92810
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.1391402184613942e-05,
+      "loss": 0.0801,
+      "step": 92815
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.1390933383338804e-05,
+      "loss": 0.1832,
+      "step": 92820
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.1390464582063666e-05,
+      "loss": 0.1939,
+      "step": 92825
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.1389995780788526e-05,
+      "loss": 0.081,
+      "step": 92830
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.1389526979513386e-05,
+      "loss": 0.0177,
+      "step": 92835
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.1389058178238245e-05,
+      "loss": 0.0086,
+      "step": 92840
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.1388589376963105e-05,
+      "loss": 0.0447,
+      "step": 92845
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.1388120575687967e-05,
+      "loss": 0.1108,
+      "step": 92850
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.1387651774412827e-05,
+      "loss": 0.0754,
+      "step": 92855
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.1387182973137687e-05,
+      "loss": 0.1089,
+      "step": 92860
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.1386714171862547e-05,
+      "loss": 0.0973,
+      "step": 92865
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.138624537058741e-05,
+      "loss": 0.2828,
+      "step": 92870
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.138577656931227e-05,
+      "loss": 0.3095,
+      "step": 92875
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.138530776803713e-05,
+      "loss": 0.092,
+      "step": 92880
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.138483896676199e-05,
+      "loss": 0.0204,
+      "step": 92885
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.1384370165486852e-05,
+      "loss": 0.0619,
+      "step": 92890
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.1383901364211711e-05,
+      "loss": 0.057,
+      "step": 92895
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.1383432562936571e-05,
+      "loss": 0.121,
+      "step": 92900
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.1382963761661431e-05,
+      "loss": 0.0572,
+      "step": 92905
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.1382494960386293e-05,
+      "loss": 0.0798,
+      "step": 92910
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.1382026159111155e-05,
+      "loss": 0.0695,
+      "step": 92915
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.1381557357836015e-05,
+      "loss": 0.1602,
+      "step": 92920
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.1381088556560874e-05,
+      "loss": 0.2295,
+      "step": 92925
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.1380619755285736e-05,
+      "loss": 0.0473,
+      "step": 92930
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.1380150954010596e-05,
+      "loss": 0.1084,
+      "step": 92935
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.1379682152735456e-05,
+      "loss": 0.0252,
+      "step": 92940
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.1379213351460316e-05,
+      "loss": 0.0559,
+      "step": 92945
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.1378744550185178e-05,
+      "loss": 0.0836,
+      "step": 92950
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.1378275748910037e-05,
+      "loss": 0.0414,
+      "step": 92955
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.1377806947634899e-05,
+      "loss": 0.081,
+      "step": 92960
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.1377338146359759e-05,
+      "loss": 0.0432,
+      "step": 92965
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.137686934508462e-05,
+      "loss": 0.1134,
+      "step": 92970
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.137640054380948e-05,
+      "loss": 0.1596,
+      "step": 92975
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.137593174253434e-05,
+      "loss": 0.0731,
+      "step": 92980
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.13754629412592e-05,
+      "loss": 0.0099,
+      "step": 92985
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.1374994139984062e-05,
+      "loss": 0.0077,
+      "step": 92990
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.1374525338708922e-05,
+      "loss": 0.031,
+      "step": 92995
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.1374056537433782e-05,
+      "loss": 0.0629,
+      "step": 93000
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.1373587736158642e-05,
+      "loss": 0.0635,
+      "step": 93005
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.1373118934883505e-05,
+      "loss": 0.0805,
+      "step": 93010
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.1372650133608365e-05,
+      "loss": 0.2164,
+      "step": 93015
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.1372181332333225e-05,
+      "loss": 0.2023,
+      "step": 93020
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.1371712531058085e-05,
+      "loss": 0.2579,
+      "step": 93025
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.1371243729782947e-05,
+      "loss": 0.0653,
+      "step": 93030
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.1370774928507807e-05,
+      "loss": 0.0023,
+      "step": 93035
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.1370306127232667e-05,
+      "loss": 0.0256,
+      "step": 93040
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.1369837325957526e-05,
+      "loss": 0.0836,
+      "step": 93045
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.1369368524682386e-05,
+      "loss": 0.0599,
+      "step": 93050
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.136889972340725e-05,
+      "loss": 0.0774,
+      "step": 93055
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.136843092213211e-05,
+      "loss": 0.1233,
+      "step": 93060
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.136796212085697e-05,
+      "loss": 0.0889,
+      "step": 93065
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.1367493319581831e-05,
+      "loss": 0.0753,
+      "step": 93070
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.1367024518306691e-05,
+      "loss": 0.2387,
+      "step": 93075
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.1366555717031551e-05,
+      "loss": 0.0767,
+      "step": 93080
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.1366086915756411e-05,
+      "loss": 0.0199,
+      "step": 93085
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.1365618114481271e-05,
+      "loss": 0.0746,
+      "step": 93090
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.1365149313206133e-05,
+      "loss": 0.0233,
+      "step": 93095
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.1364680511930994e-05,
+      "loss": 0.0937,
+      "step": 93100
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.1364211710655854e-05,
+      "loss": 0.2074,
+      "step": 93105
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.1363742909380716e-05,
+      "loss": 0.0425,
+      "step": 93110
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.1363274108105576e-05,
+      "loss": 0.0882,
+      "step": 93115
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.1362805306830436e-05,
+      "loss": 0.0885,
+      "step": 93120
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.1362336505555296e-05,
+      "loss": 0.2915,
+      "step": 93125
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.1361867704280155e-05,
+      "loss": 0.1091,
+      "step": 93130
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.1361398903005017e-05,
+      "loss": 0.0368,
+      "step": 93135
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.1360930101729877e-05,
+      "loss": 0.0263,
+      "step": 93140
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.1360461300454739e-05,
+      "loss": 0.0319,
+      "step": 93145
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.13599924991796e-05,
+      "loss": 0.0369,
+      "step": 93150
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.135952369790446e-05,
+      "loss": 0.0442,
+      "step": 93155
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.135905489662932e-05,
+      "loss": 0.0669,
+      "step": 93160
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.135858609535418e-05,
+      "loss": 0.0986,
+      "step": 93165
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.135811729407904e-05,
+      "loss": 0.0718,
+      "step": 93170
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.1357648492803902e-05,
+      "loss": 0.2575,
+      "step": 93175
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.1357179691528762e-05,
+      "loss": 0.0897,
+      "step": 93180
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.1356710890253622e-05,
+      "loss": 0.0338,
+      "step": 93185
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.1356242088978481e-05,
+      "loss": 0.0242,
+      "step": 93190
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.1355773287703345e-05,
+      "loss": 0.0657,
+      "step": 93195
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.1355304486428205e-05,
+      "loss": 0.0481,
+      "step": 93200
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.1354835685153065e-05,
+      "loss": 0.0743,
+      "step": 93205
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.1354366883877925e-05,
+      "loss": 0.0767,
+      "step": 93210
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.1353898082602786e-05,
+      "loss": 0.1051,
+      "step": 93215
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.1353429281327646e-05,
+      "loss": 0.1902,
+      "step": 93220
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.1352960480052506e-05,
+      "loss": 0.2896,
+      "step": 93225
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.1352491678777366e-05,
+      "loss": 0.0998,
+      "step": 93230
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.1352022877502226e-05,
+      "loss": 0.064,
+      "step": 93235
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.135155407622709e-05,
+      "loss": 0.0245,
+      "step": 93240
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.135108527495195e-05,
+      "loss": 0.0712,
+      "step": 93245
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.1350616473676809e-05,
+      "loss": 0.0593,
+      "step": 93250
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.135014767240167e-05,
+      "loss": 0.0652,
+      "step": 93255
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.134967887112653e-05,
+      "loss": 0.1075,
+      "step": 93260
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.134921006985139e-05,
+      "loss": 0.1495,
+      "step": 93265
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.134874126857625e-05,
+      "loss": 0.1475,
+      "step": 93270
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.134827246730111e-05,
+      "loss": 0.3304,
+      "step": 93275
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.1347803666025972e-05,
+      "loss": 0.0534,
+      "step": 93280
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.1347334864750834e-05,
+      "loss": 0.0177,
+      "step": 93285
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.1346866063475694e-05,
+      "loss": 0.0147,
+      "step": 93290
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.1346397262200555e-05,
+      "loss": 0.0469,
+      "step": 93295
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.1345928460925415e-05,
+      "loss": 0.0652,
+      "step": 93300
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.1345459659650275e-05,
+      "loss": 0.0227,
+      "step": 93305
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.1344990858375135e-05,
+      "loss": 0.0823,
+      "step": 93310
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.1344522057099995e-05,
+      "loss": 0.0464,
+      "step": 93315
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.1344053255824857e-05,
+      "loss": 0.1616,
+      "step": 93320
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.1343584454549717e-05,
+      "loss": 0.2665,
+      "step": 93325
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.1343115653274577e-05,
+      "loss": 0.174,
+      "step": 93330
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.134264685199944e-05,
+      "loss": 0.0116,
+      "step": 93335
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.13421780507243e-05,
+      "loss": 0.0323,
+      "step": 93340
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.134170924944916e-05,
+      "loss": 0.0394,
+      "step": 93345
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.134124044817402e-05,
+      "loss": 0.029,
+      "step": 93350
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.1340771646898881e-05,
+      "loss": 0.0564,
+      "step": 93355
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.1340302845623741e-05,
+      "loss": 0.1097,
+      "step": 93360
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.1339834044348601e-05,
+      "loss": 0.1954,
+      "step": 93365
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.1339365243073461e-05,
+      "loss": 0.1325,
+      "step": 93370
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.1338896441798321e-05,
+      "loss": 0.2073,
+      "step": 93375
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.1338427640523184e-05,
+      "loss": 0.0924,
+      "step": 93380
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.1337958839248044e-05,
+      "loss": 0.0208,
+      "step": 93385
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.1337490037972904e-05,
+      "loss": 0.0388,
+      "step": 93390
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.1337021236697766e-05,
+      "loss": 0.0524,
+      "step": 93395
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.1336552435422626e-05,
+      "loss": 0.0209,
+      "step": 93400
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.1336083634147486e-05,
+      "loss": 0.0644,
+      "step": 93405
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.1335614832872346e-05,
+      "loss": 0.1235,
+      "step": 93410
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.1335146031597206e-05,
+      "loss": 0.101,
+      "step": 93415
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.1334677230322067e-05,
+      "loss": 0.1767,
+      "step": 93420
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.1334208429046929e-05,
+      "loss": 0.314,
+      "step": 93425
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.1333739627771789e-05,
+      "loss": 0.1279,
+      "step": 93430
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.133327082649665e-05,
+      "loss": 0.0459,
+      "step": 93435
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.133280202522151e-05,
+      "loss": 0.0312,
+      "step": 93440
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.133233322394637e-05,
+      "loss": 0.071,
+      "step": 93445
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.133186442267123e-05,
+      "loss": 0.0254,
+      "step": 93450
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.133139562139609e-05,
+      "loss": 0.051,
+      "step": 93455
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.1330926820120952e-05,
+      "loss": 0.0682,
+      "step": 93460
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.1330458018845812e-05,
+      "loss": 0.0462,
+      "step": 93465
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.1329989217570673e-05,
+      "loss": 0.1674,
+      "step": 93470
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.1329520416295535e-05,
+      "loss": 0.2021,
+      "step": 93475
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.1329051615020395e-05,
+      "loss": 0.0761,
+      "step": 93480
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.1328582813745255e-05,
+      "loss": 0.026,
+      "step": 93485
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.1328114012470115e-05,
+      "loss": 0.0264,
+      "step": 93490
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.1327645211194975e-05,
+      "loss": 0.0651,
+      "step": 93495
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.1327176409919836e-05,
+      "loss": 0.0882,
+      "step": 93500
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.1326707608644696e-05,
+      "loss": 0.0469,
+      "step": 93505
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.1326238807369556e-05,
+      "loss": 0.0392,
+      "step": 93510
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.1325770006094416e-05,
+      "loss": 0.0775,
+      "step": 93515
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.132530120481928e-05,
+      "loss": 0.1573,
+      "step": 93520
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.132483240354414e-05,
+      "loss": 0.2538,
+      "step": 93525
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.1324363602269e-05,
+      "loss": 0.1064,
+      "step": 93530
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.132389480099386e-05,
+      "loss": 0.0241,
+      "step": 93535
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.1323425999718721e-05,
+      "loss": 0.0369,
+      "step": 93540
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.132295719844358e-05,
+      "loss": 0.0454,
+      "step": 93545
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.132248839716844e-05,
+      "loss": 0.0286,
+      "step": 93550
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.13220195958933e-05,
+      "loss": 0.0409,
+      "step": 93555
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.132155079461816e-05,
+      "loss": 0.147,
+      "step": 93560
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1321081993343024e-05,
+      "loss": 0.1259,
+      "step": 93565
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1320613192067884e-05,
+      "loss": 0.1534,
+      "step": 93570
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1320144390792744e-05,
+      "loss": 0.2241,
+      "step": 93575
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1319675589517605e-05,
+      "loss": 0.0898,
+      "step": 93580
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1319206788242465e-05,
+      "loss": 0.038,
+      "step": 93585
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1318737986967325e-05,
+      "loss": 0.0325,
+      "step": 93590
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1318269185692185e-05,
+      "loss": 0.0524,
+      "step": 93595
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1317800384417045e-05,
+      "loss": 0.0863,
+      "step": 93600
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1317331583141907e-05,
+      "loss": 0.0932,
+      "step": 93605
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1316862781866768e-05,
+      "loss": 0.065,
+      "step": 93610
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1316393980591628e-05,
+      "loss": 0.1651,
+      "step": 93615
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.131592517931649e-05,
+      "loss": 0.1685,
+      "step": 93620
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.131545637804135e-05,
+      "loss": 0.3206,
+      "step": 93625
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.131498757676621e-05,
+      "loss": 0.1088,
+      "step": 93630
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.131451877549107e-05,
+      "loss": 0.0235,
+      "step": 93635
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.131404997421593e-05,
+      "loss": 0.055,
+      "step": 93640
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1313581172940791e-05,
+      "loss": 0.023,
+      "step": 93645
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1313112371665651e-05,
+      "loss": 0.0641,
+      "step": 93650
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1312643570390511e-05,
+      "loss": 0.0778,
+      "step": 93655
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1312174769115375e-05,
+      "loss": 0.0932,
+      "step": 93660
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1311705967840234e-05,
+      "loss": 0.0475,
+      "step": 93665
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1311237166565094e-05,
+      "loss": 0.2037,
+      "step": 93670
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1310768365289954e-05,
+      "loss": 0.2547,
+      "step": 93675
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1310299564014814e-05,
+      "loss": 0.1241,
+      "step": 93680
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1309830762739676e-05,
+      "loss": 0.014,
+      "step": 93685
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1309361961464536e-05,
+      "loss": 0.0262,
+      "step": 93690
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1308893160189396e-05,
+      "loss": 0.0372,
+      "step": 93695
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1308424358914256e-05,
+      "loss": 0.0808,
+      "step": 93700
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1307955557639119e-05,
+      "loss": 0.0937,
+      "step": 93705
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1307486756363979e-05,
+      "loss": 0.1413,
+      "step": 93710
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1307017955088839e-05,
+      "loss": 0.1658,
+      "step": 93715
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1306549153813699e-05,
+      "loss": 0.1432,
+      "step": 93720
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.130608035253856e-05,
+      "loss": 0.2572,
+      "step": 93725
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.130561155126342e-05,
+      "loss": 0.0823,
+      "step": 93730
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.130514274998828e-05,
+      "loss": 0.0146,
+      "step": 93735
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.130467394871314e-05,
+      "loss": 0.0098,
+      "step": 93740
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1304205147438002e-05,
+      "loss": 0.0271,
+      "step": 93745
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1303736346162863e-05,
+      "loss": 0.0164,
+      "step": 93750
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1303267544887723e-05,
+      "loss": 0.0325,
+      "step": 93755
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1302798743612583e-05,
+      "loss": 0.1518,
+      "step": 93760
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1302329942337445e-05,
+      "loss": 0.0947,
+      "step": 93765
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1301861141062305e-05,
+      "loss": 0.1963,
+      "step": 93770
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1301392339787165e-05,
+      "loss": 0.3213,
+      "step": 93775
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1300923538512025e-05,
+      "loss": 0.1066,
+      "step": 93780
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1300454737236886e-05,
+      "loss": 0.0301,
+      "step": 93785
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1299985935961746e-05,
+      "loss": 0.0105,
+      "step": 93790
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1299517134686608e-05,
+      "loss": 0.0451,
+      "step": 93795
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1299048333411468e-05,
+      "loss": 0.0651,
+      "step": 93800
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.129857953213633e-05,
+      "loss": 0.0617,
+      "step": 93805
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.129811073086119e-05,
+      "loss": 0.0659,
+      "step": 93810
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.129764192958605e-05,
+      "loss": 0.121,
+      "step": 93815
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.129717312831091e-05,
+      "loss": 0.1004,
+      "step": 93820
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1296704327035771e-05,
+      "loss": 0.3267,
+      "step": 93825
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1296235525760631e-05,
+      "loss": 0.1201,
+      "step": 93830
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.129576672448549e-05,
+      "loss": 0.0243,
+      "step": 93835
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.129529792321035e-05,
+      "loss": 0.0153,
+      "step": 93840
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1294829121935214e-05,
+      "loss": 0.0669,
+      "step": 93845
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1294360320660074e-05,
+      "loss": 0.0889,
+      "step": 93850
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1293891519384934e-05,
+      "loss": 0.047,
+      "step": 93855
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1293422718109794e-05,
+      "loss": 0.0768,
+      "step": 93860
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1292953916834656e-05,
+      "loss": 0.1047,
+      "step": 93865
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1292485115559515e-05,
+      "loss": 0.1511,
+      "step": 93870
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1292016314284375e-05,
+      "loss": 0.1909,
+      "step": 93875
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1291547513009235e-05,
+      "loss": 0.0859,
+      "step": 93880
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1291078711734095e-05,
+      "loss": 0.0099,
+      "step": 93885
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1290609910458959e-05,
+      "loss": 0.0474,
+      "step": 93890
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1290141109183818e-05,
+      "loss": 0.0195,
+      "step": 93895
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1289672307908678e-05,
+      "loss": 0.0113,
+      "step": 93900
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.128920350663354e-05,
+      "loss": 0.0707,
+      "step": 93905
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.12887347053584e-05,
+      "loss": 0.0706,
+      "step": 93910
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.128826590408326e-05,
+      "loss": 0.0743,
+      "step": 93915
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.128779710280812e-05,
+      "loss": 0.0829,
+      "step": 93920
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.128732830153298e-05,
+      "loss": 0.1584,
+      "step": 93925
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1286859500257841e-05,
+      "loss": 0.1059,
+      "step": 93930
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1286390698982703e-05,
+      "loss": 0.029,
+      "step": 93935
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1285921897707563e-05,
+      "loss": 0.0266,
+      "step": 93940
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1285453096432425e-05,
+      "loss": 0.0672,
+      "step": 93945
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1284984295157285e-05,
+      "loss": 0.0235,
+      "step": 93950
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1284515493882144e-05,
+      "loss": 0.0488,
+      "step": 93955
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1284046692607004e-05,
+      "loss": 0.1255,
+      "step": 93960
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1283577891331864e-05,
+      "loss": 0.0715,
+      "step": 93965
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1283109090056726e-05,
+      "loss": 0.1523,
+      "step": 93970
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1282640288781586e-05,
+      "loss": 0.3252,
+      "step": 93975
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1282171487506446e-05,
+      "loss": 0.1029,
+      "step": 93980
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.128170268623131e-05,
+      "loss": 0.1262,
+      "step": 93985
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1281233884956169e-05,
+      "loss": 0.0488,
+      "step": 93990
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1280765083681029e-05,
+      "loss": 0.0067,
+      "step": 93995
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1280296282405889e-05,
+      "loss": 0.023,
+      "step": 94000
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1279827481130749e-05,
+      "loss": 0.0867,
+      "step": 94005
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.127935867985561e-05,
+      "loss": 0.0932,
+      "step": 94010
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.127888987858047e-05,
+      "loss": 0.0649,
+      "step": 94015
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.127842107730533e-05,
+      "loss": 0.1785,
+      "step": 94020
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.127795227603019e-05,
+      "loss": 0.2187,
+      "step": 94025
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1277483474755054e-05,
+      "loss": 0.0831,
+      "step": 94030
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1277014673479914e-05,
+      "loss": 0.0091,
+      "step": 94035
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1276545872204774e-05,
+      "loss": 0.0288,
+      "step": 94040
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1276077070929633e-05,
+      "loss": 0.0356,
+      "step": 94045
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1275608269654495e-05,
+      "loss": 0.0559,
+      "step": 94050
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1275139468379355e-05,
+      "loss": 0.0913,
+      "step": 94055
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1274670667104215e-05,
+      "loss": 0.0628,
+      "step": 94060
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1274201865829075e-05,
+      "loss": 0.1708,
+      "step": 94065
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1273733064553935e-05,
+      "loss": 0.1671,
+      "step": 94070
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1273264263278798e-05,
+      "loss": 0.2832,
+      "step": 94075
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1272795462003658e-05,
+      "loss": 0.087,
+      "step": 94080
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1272326660728518e-05,
+      "loss": 0.0399,
+      "step": 94085
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.127185785945338e-05,
+      "loss": 0.053,
+      "step": 94090
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.127138905817824e-05,
+      "loss": 0.0415,
+      "step": 94095
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.12709202569031e-05,
+      "loss": 0.1127,
+      "step": 94100
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.127045145562796e-05,
+      "loss": 0.0561,
+      "step": 94105
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.126998265435282e-05,
+      "loss": 0.0568,
+      "step": 94110
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1269513853077681e-05,
+      "loss": 0.1383,
+      "step": 94115
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1269045051802543e-05,
+      "loss": 0.1745,
+      "step": 94120
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1268576250527403e-05,
+      "loss": 0.2155,
+      "step": 94125
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1268107449252264e-05,
+      "loss": 0.0674,
+      "step": 94130
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1267638647977124e-05,
+      "loss": 0.0514,
+      "step": 94135
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1267169846701984e-05,
+      "loss": 0.0352,
+      "step": 94140
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1266701045426844e-05,
+      "loss": 0.0417,
+      "step": 94145
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1266232244151704e-05,
+      "loss": 0.0777,
+      "step": 94150
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1265763442876566e-05,
+      "loss": 0.0277,
+      "step": 94155
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1265294641601425e-05,
+      "loss": 0.0597,
+      "step": 94160
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1264825840326285e-05,
+      "loss": 0.0858,
+      "step": 94165
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1264357039051149e-05,
+      "loss": 0.1687,
+      "step": 94170
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1263888237776009e-05,
+      "loss": 0.3233,
+      "step": 94175
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1263419436500869e-05,
+      "loss": 0.1246,
+      "step": 94180
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1262950635225729e-05,
+      "loss": 0.0243,
+      "step": 94185
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.1262481833950588e-05,
+      "loss": 0.054,
+      "step": 94190
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.126201303267545e-05,
+      "loss": 0.0775,
+      "step": 94195
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.126154423140031e-05,
+      "loss": 0.0309,
+      "step": 94200
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.126107543012517e-05,
+      "loss": 0.1172,
+      "step": 94205
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.126060662885003e-05,
+      "loss": 0.079,
+      "step": 94210
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.1260137827574893e-05,
+      "loss": 0.0634,
+      "step": 94215
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.1259669026299753e-05,
+      "loss": 0.1968,
+      "step": 94220
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.1259200225024613e-05,
+      "loss": 0.3465,
+      "step": 94225
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.1258731423749473e-05,
+      "loss": 0.0498,
+      "step": 94230
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.1258262622474335e-05,
+      "loss": 0.0276,
+      "step": 94235
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.1257793821199195e-05,
+      "loss": 0.0654,
+      "step": 94240
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.1257325019924055e-05,
+      "loss": 0.0597,
+      "step": 94245
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.1256856218648914e-05,
+      "loss": 0.0425,
+      "step": 94250
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.1256387417373776e-05,
+      "loss": 0.1243,
+      "step": 94255
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.1255918616098638e-05,
+      "loss": 0.0957,
+      "step": 94260
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.1255449814823498e-05,
+      "loss": 0.1223,
+      "step": 94265
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.1254981013548358e-05,
+      "loss": 0.1075,
+      "step": 94270
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.125451221227322e-05,
+      "loss": 0.2353,
+      "step": 94275
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.1254043410998079e-05,
+      "loss": 0.0526,
+      "step": 94280
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.1253574609722939e-05,
+      "loss": 0.0169,
+      "step": 94285
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.1253105808447799e-05,
+      "loss": 0.0166,
+      "step": 94290
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.125263700717266e-05,
+      "loss": 0.0217,
+      "step": 94295
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.125216820589752e-05,
+      "loss": 0.044,
+      "step": 94300
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.125169940462238e-05,
+      "loss": 0.0456,
+      "step": 94305
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.1251230603347242e-05,
+      "loss": 0.0509,
+      "step": 94310
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.1250761802072104e-05,
+      "loss": 0.1222,
+      "step": 94315
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.1250293000796964e-05,
+      "loss": 0.276,
+      "step": 94320
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.1249824199521824e-05,
+      "loss": 0.2786,
+      "step": 94325
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.1249355398246684e-05,
+      "loss": 0.0699,
+      "step": 94330
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.1248886596971545e-05,
+      "loss": 0.0244,
+      "step": 94335
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.1248417795696405e-05,
+      "loss": 0.0295,
+      "step": 94340
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.1247948994421265e-05,
+      "loss": 0.0276,
+      "step": 94345
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.1247480193146125e-05,
+      "loss": 0.1131,
+      "step": 94350
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.1247011391870988e-05,
+      "loss": 0.0372,
+      "step": 94355
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.1246542590595848e-05,
+      "loss": 0.0917,
+      "step": 94360
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.1246073789320708e-05,
+      "loss": 0.0983,
+      "step": 94365
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.1245604988045568e-05,
+      "loss": 0.1681,
+      "step": 94370
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.124513618677043e-05,
+      "loss": 0.2011,
+      "step": 94375
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.124466738549529e-05,
+      "loss": 0.0976,
+      "step": 94380
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.124419858422015e-05,
+      "loss": 0.0407,
+      "step": 94385
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.124372978294501e-05,
+      "loss": 0.0502,
+      "step": 94390
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.124326098166987e-05,
+      "loss": 0.0199,
+      "step": 94395
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.1242792180394733e-05,
+      "loss": 0.0882,
+      "step": 94400
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.1242323379119593e-05,
+      "loss": 0.0577,
+      "step": 94405
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.1241854577844453e-05,
+      "loss": 0.0946,
+      "step": 94410
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.1241385776569314e-05,
+      "loss": 0.0738,
+      "step": 94415
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.1240916975294174e-05,
+      "loss": 0.085,
+      "step": 94420
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.1240448174019034e-05,
+      "loss": 0.235,
+      "step": 94425
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.1239979372743894e-05,
+      "loss": 0.1043,
+      "step": 94430
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.1239510571468754e-05,
+      "loss": 0.0541,
+      "step": 94435
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.1239041770193616e-05,
+      "loss": 0.0314,
+      "step": 94440
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.1238572968918476e-05,
+      "loss": 0.0914,
+      "step": 94445
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.1238104167643337e-05,
+      "loss": 0.0338,
+      "step": 94450
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.1237635366368199e-05,
+      "loss": 0.0784,
+      "step": 94455
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.1237166565093059e-05,
+      "loss": 0.1103,
+      "step": 94460
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.1236697763817919e-05,
+      "loss": 0.1155,
+      "step": 94465
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.1236228962542779e-05,
+      "loss": 0.1427,
+      "step": 94470
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.1235760161267639e-05,
+      "loss": 0.179,
+      "step": 94475
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.12352913599925e-05,
+      "loss": 0.0425,
+      "step": 94480
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.123482255871736e-05,
+      "loss": 0.0324,
+      "step": 94485
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.123435375744222e-05,
+      "loss": 0.0479,
+      "step": 94490
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.1233884956167083e-05,
+      "loss": 0.021,
+      "step": 94495
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.1233416154891943e-05,
+      "loss": 0.0872,
+      "step": 94500
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.1232947353616803e-05,
+      "loss": 0.0825,
+      "step": 94505
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.1232478552341663e-05,
+      "loss": 0.0618,
+      "step": 94510
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.1232009751066523e-05,
+      "loss": 0.0927,
+      "step": 94515
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.1231540949791385e-05,
+      "loss": 0.1971,
+      "step": 94520
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.1231072148516245e-05,
+      "loss": 0.224,
+      "step": 94525
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.1230603347241105e-05,
+      "loss": 0.0707,
+      "step": 94530
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.1230134545965965e-05,
+      "loss": 0.0205,
+      "step": 94535
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.1229665744690828e-05,
+      "loss": 0.0229,
+      "step": 94540
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.1229196943415688e-05,
+      "loss": 0.0709,
+      "step": 94545
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.1228728142140548e-05,
+      "loss": 0.1253,
+      "step": 94550
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.1228259340865408e-05,
+      "loss": 0.0335,
+      "step": 94555
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.122779053959027e-05,
+      "loss": 0.0418,
+      "step": 94560
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.122732173831513e-05,
+      "loss": 0.1633,
+      "step": 94565
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.122685293703999e-05,
+      "loss": 0.0939,
+      "step": 94570
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.1226384135764849e-05,
+      "loss": 0.2821,
+      "step": 94575
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.1225915334489709e-05,
+      "loss": 0.0763,
+      "step": 94580
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.1225446533214572e-05,
+      "loss": 0.0498,
+      "step": 94585
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.1224977731939432e-05,
+      "loss": 0.0285,
+      "step": 94590
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.1224508930664292e-05,
+      "loss": 0.0782,
+      "step": 94595
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.1224040129389154e-05,
+      "loss": 0.0302,
+      "step": 94600
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.1223571328114014e-05,
+      "loss": 0.0348,
+      "step": 94605
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.1223102526838874e-05,
+      "loss": 0.0701,
+      "step": 94610
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.1222633725563734e-05,
+      "loss": 0.1278,
+      "step": 94615
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.1222164924288594e-05,
+      "loss": 0.0813,
+      "step": 94620
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.1221696123013455e-05,
+      "loss": 0.37,
+      "step": 94625
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.1221227321738315e-05,
+      "loss": 0.0652,
+      "step": 94630
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.1220758520463177e-05,
+      "loss": 0.0244,
+      "step": 94635
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.1220289719188038e-05,
+      "loss": 0.0597,
+      "step": 94640
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.1219820917912898e-05,
+      "loss": 0.0138,
+      "step": 94645
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.1219352116637758e-05,
+      "loss": 0.0577,
+      "step": 94650
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.1218883315362618e-05,
+      "loss": 0.0518,
+      "step": 94655
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.1218414514087478e-05,
+      "loss": 0.1242,
+      "step": 94660
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.121794571281234e-05,
+      "loss": 0.1233,
+      "step": 94665
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.12174769115372e-05,
+      "loss": 0.0707,
+      "step": 94670
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.121700811026206e-05,
+      "loss": 0.2971,
+      "step": 94675
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.1216539308986923e-05,
+      "loss": 0.0634,
+      "step": 94680
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.1216070507711783e-05,
+      "loss": 0.0151,
+      "step": 94685
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.1215601706436643e-05,
+      "loss": 0.0477,
+      "step": 94690
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.1215132905161503e-05,
+      "loss": 0.0449,
+      "step": 94695
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.1214664103886363e-05,
+      "loss": 0.0705,
+      "step": 94700
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.1214195302611224e-05,
+      "loss": 0.0546,
+      "step": 94705
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.1213726501336084e-05,
+      "loss": 0.0878,
+      "step": 94710
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.1213257700060944e-05,
+      "loss": 0.0852,
+      "step": 94715
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.1212788898785804e-05,
+      "loss": 0.0628,
+      "step": 94720
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.1212320097510667e-05,
+      "loss": 0.2429,
+      "step": 94725
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.1211851296235527e-05,
+      "loss": 0.0501,
+      "step": 94730
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.1211382494960387e-05,
+      "loss": 0.0319,
+      "step": 94735
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.1210913693685247e-05,
+      "loss": 0.0394,
+      "step": 94740
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.1210444892410109e-05,
+      "loss": 0.0397,
+      "step": 94745
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.1209976091134969e-05,
+      "loss": 0.0548,
+      "step": 94750
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.1209507289859829e-05,
+      "loss": 0.15,
+      "step": 94755
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.1209038488584689e-05,
+      "loss": 0.0634,
+      "step": 94760
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.120856968730955e-05,
+      "loss": 0.0811,
+      "step": 94765
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.120810088603441e-05,
+      "loss": 0.1546,
+      "step": 94770
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.1207632084759272e-05,
+      "loss": 0.245,
+      "step": 94775
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.1207163283484133e-05,
+      "loss": 0.0774,
+      "step": 94780
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.1206694482208993e-05,
+      "loss": 0.0428,
+      "step": 94785
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.1206225680933853e-05,
+      "loss": 0.045,
+      "step": 94790
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.1205756879658713e-05,
+      "loss": 0.0884,
+      "step": 94795
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.1205288078383573e-05,
+      "loss": 0.0341,
+      "step": 94800
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.1204819277108435e-05,
+      "loss": 0.0694,
+      "step": 94805
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.1204350475833295e-05,
+      "loss": 0.0746,
+      "step": 94810
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.1203881674558155e-05,
+      "loss": 0.1387,
+      "step": 94815
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.1203412873283018e-05,
+      "loss": 0.1527,
+      "step": 94820
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.1202944072007878e-05,
+      "loss": 0.2188,
+      "step": 94825
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.1202475270732738e-05,
+      "loss": 0.0657,
+      "step": 94830
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.1202006469457598e-05,
+      "loss": 0.0172,
+      "step": 94835
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.1201537668182458e-05,
+      "loss": 0.034,
+      "step": 94840
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.120106886690732e-05,
+      "loss": 0.0754,
+      "step": 94845
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.120060006563218e-05,
+      "loss": 0.0494,
+      "step": 94850
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.120013126435704e-05,
+      "loss": 0.0505,
+      "step": 94855
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.11996624630819e-05,
+      "loss": 0.0729,
+      "step": 94860
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.1199193661806763e-05,
+      "loss": 0.11,
+      "step": 94865
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.1198724860531622e-05,
+      "loss": 0.1641,
+      "step": 94870
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.1198256059256482e-05,
+      "loss": 0.1853,
+      "step": 94875
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.1197787257981342e-05,
+      "loss": 0.0978,
+      "step": 94880
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.1197318456706204e-05,
+      "loss": 0.012,
+      "step": 94885
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.1196849655431064e-05,
+      "loss": 0.0485,
+      "step": 94890
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.1196380854155924e-05,
+      "loss": 0.0729,
+      "step": 94895
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.1195912052880784e-05,
+      "loss": 0.0622,
+      "step": 94900
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.1195443251605644e-05,
+      "loss": 0.0556,
+      "step": 94905
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.1194974450330507e-05,
+      "loss": 0.0774,
+      "step": 94910
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.1194505649055367e-05,
+      "loss": 0.0702,
+      "step": 94915
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.1194036847780227e-05,
+      "loss": 0.2151,
+      "step": 94920
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.1193568046505088e-05,
+      "loss": 0.1308,
+      "step": 94925
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.1193099245229948e-05,
+      "loss": 0.0543,
+      "step": 94930
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.1192630443954808e-05,
+      "loss": 0.0192,
+      "step": 94935
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.1192161642679668e-05,
+      "loss": 0.066,
+      "step": 94940
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.1191692841404528e-05,
+      "loss": 0.0214,
+      "step": 94945
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.119122404012939e-05,
+      "loss": 0.0609,
+      "step": 94950
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.119075523885425e-05,
+      "loss": 0.0662,
+      "step": 94955
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.1190286437579111e-05,
+      "loss": 0.0625,
+      "step": 94960
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.1189817636303973e-05,
+      "loss": 0.0978,
+      "step": 94965
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.1189348835028833e-05,
+      "loss": 0.1464,
+      "step": 94970
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.1188880033753693e-05,
+      "loss": 0.2409,
+      "step": 94975
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.1188411232478553e-05,
+      "loss": 0.0647,
+      "step": 94980
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.1187942431203413e-05,
+      "loss": 0.0181,
+      "step": 94985
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.1187473629928274e-05,
+      "loss": 0.0826,
+      "step": 94990
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.1187004828653134e-05,
+      "loss": 0.0638,
+      "step": 94995
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.1186536027377994e-05,
+      "loss": 0.0776,
+      "step": 95000
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.1186067226102858e-05,
+      "loss": 0.0641,
+      "step": 95005
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.1185598424827718e-05,
+      "loss": 0.0684,
+      "step": 95010
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.1185129623552577e-05,
+      "loss": 0.1146,
+      "step": 95015
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.1184660822277437e-05,
+      "loss": 0.1459,
+      "step": 95020
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.1184192021002297e-05,
+      "loss": 0.3169,
+      "step": 95025
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.1183723219727159e-05,
+      "loss": 0.0857,
+      "step": 95030
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.1183254418452019e-05,
+      "loss": 0.0188,
+      "step": 95035
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.1182785617176879e-05,
+      "loss": 0.0165,
+      "step": 95040
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.1182316815901739e-05,
+      "loss": 0.0268,
+      "step": 95045
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.1181848014626602e-05,
+      "loss": 0.0907,
+      "step": 95050
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.1181379213351462e-05,
+      "loss": 0.0898,
+      "step": 95055
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.1180910412076322e-05,
+      "loss": 0.0895,
+      "step": 95060
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.1180441610801182e-05,
+      "loss": 0.1027,
+      "step": 95065
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.1179972809526044e-05,
+      "loss": 0.1274,
+      "step": 95070
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.1179504008250903e-05,
+      "loss": 0.277,
+      "step": 95075
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.1179035206975763e-05,
+      "loss": 0.0948,
+      "step": 95080
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.1178566405700623e-05,
+      "loss": 0.0314,
+      "step": 95085
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.1178097604425483e-05,
+      "loss": 0.0511,
+      "step": 95090
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.1177628803150345e-05,
+      "loss": 0.0339,
+      "step": 95095
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.1177160001875206e-05,
+      "loss": 0.0502,
+      "step": 95100
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.1176691200600066e-05,
+      "loss": 0.1113,
+      "step": 95105
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.1176222399324928e-05,
+      "loss": 0.139,
+      "step": 95110
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.1175753598049788e-05,
+      "loss": 0.1397,
+      "step": 95115
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.1175284796774648e-05,
+      "loss": 0.2609,
+      "step": 95120
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.1174815995499508e-05,
+      "loss": 0.1834,
+      "step": 95125
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.117434719422437e-05,
+      "loss": 0.0553,
+      "step": 95130
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.117387839294923e-05,
+      "loss": 0.043,
+      "step": 95135
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.117340959167409e-05,
+      "loss": 0.03,
+      "step": 95140
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.1172940790398951e-05,
+      "loss": 0.0375,
+      "step": 95145
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.1172471989123813e-05,
+      "loss": 0.0403,
+      "step": 95150
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.1172003187848673e-05,
+      "loss": 0.0278,
+      "step": 95155
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.1171534386573532e-05,
+      "loss": 0.0922,
+      "step": 95160
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.1171065585298392e-05,
+      "loss": 0.0873,
+      "step": 95165
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.1170596784023254e-05,
+      "loss": 0.1153,
+      "step": 95170
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.1170127982748114e-05,
+      "loss": 0.1616,
+      "step": 95175
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.1169659181472974e-05,
+      "loss": 0.1208,
+      "step": 95180
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.1169190380197834e-05,
+      "loss": 0.0132,
+      "step": 95185
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.1168721578922697e-05,
+      "loss": 0.0177,
+      "step": 95190
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.1168252777647557e-05,
+      "loss": 0.0466,
+      "step": 95195
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.1167783976372417e-05,
+      "loss": 0.0432,
+      "step": 95200
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.1167315175097277e-05,
+      "loss": 0.0164,
+      "step": 95205
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.1166846373822139e-05,
+      "loss": 0.0618,
+      "step": 95210
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.1166377572546999e-05,
+      "loss": 0.0556,
+      "step": 95215
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.1165908771271858e-05,
+      "loss": 0.1505,
+      "step": 95220
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.1165439969996718e-05,
+      "loss": 0.2956,
+      "step": 95225
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.1164971168721578e-05,
+      "loss": 0.095,
+      "step": 95230
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.1164502367446442e-05,
+      "loss": 0.045,
+      "step": 95235
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.1164033566171302e-05,
+      "loss": 0.0196,
+      "step": 95240
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.1163564764896162e-05,
+      "loss": 0.0183,
+      "step": 95245
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.1163095963621023e-05,
+      "loss": 0.0611,
+      "step": 95250
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.1162627162345883e-05,
+      "loss": 0.0417,
+      "step": 95255
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.1162158361070743e-05,
+      "loss": 0.1163,
+      "step": 95260
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.1161689559795603e-05,
+      "loss": 0.1493,
+      "step": 95265
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.1161220758520463e-05,
+      "loss": 0.0971,
+      "step": 95270
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.1160751957245325e-05,
+      "loss": 0.2363,
+      "step": 95275
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.1160283155970184e-05,
+      "loss": 0.0752,
+      "step": 95280
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.1159814354695046e-05,
+      "loss": 0.0119,
+      "step": 95285
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.1159345553419908e-05,
+      "loss": 0.0405,
+      "step": 95290
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.1158876752144768e-05,
+      "loss": 0.038,
+      "step": 95295
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.1158407950869628e-05,
+      "loss": 0.0414,
+      "step": 95300
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.1157939149594487e-05,
+      "loss": 0.0728,
+      "step": 95305
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.1157470348319347e-05,
+      "loss": 0.0689,
+      "step": 95310
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.1157001547044209e-05,
+      "loss": 0.0786,
+      "step": 95315
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.1156532745769069e-05,
+      "loss": 0.0901,
+      "step": 95320
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.1156063944493929e-05,
+      "loss": 0.3152,
+      "step": 95325
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.1155595143218792e-05,
+      "loss": 0.0653,
+      "step": 95330
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.1155126341943652e-05,
+      "loss": 0.0816,
+      "step": 95335
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.1154657540668512e-05,
+      "loss": 0.0327,
+      "step": 95340
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.1154188739393372e-05,
+      "loss": 0.0665,
+      "step": 95345
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.1153719938118232e-05,
+      "loss": 0.0672,
+      "step": 95350
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.1153251136843094e-05,
+      "loss": 0.0251,
+      "step": 95355
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.1152782335567954e-05,
+      "loss": 0.1739,
+      "step": 95360
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.1152313534292813e-05,
+      "loss": 0.0909,
+      "step": 95365
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.1151844733017673e-05,
+      "loss": 0.1172,
+      "step": 95370
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.1151375931742537e-05,
+      "loss": 0.334,
+      "step": 95375
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.1150907130467397e-05,
+      "loss": 0.0837,
+      "step": 95380
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.1150438329192257e-05,
+      "loss": 0.0251,
+      "step": 95385
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.1149969527917117e-05,
+      "loss": 0.0642,
+      "step": 95390
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.1149500726641978e-05,
+      "loss": 0.0333,
+      "step": 95395
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.1149031925366838e-05,
+      "loss": 0.0664,
+      "step": 95400
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.1148563124091698e-05,
+      "loss": 0.0883,
+      "step": 95405
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.1148094322816558e-05,
+      "loss": 0.0742,
+      "step": 95410
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.1147625521541418e-05,
+      "loss": 0.1999,
+      "step": 95415
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.114715672026628e-05,
+      "loss": 0.1842,
+      "step": 95420
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.1146687918991141e-05,
+      "loss": 0.2251,
+      "step": 95425
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.1146219117716001e-05,
+      "loss": 0.0577,
+      "step": 95430
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.1145750316440863e-05,
+      "loss": 0.0651,
+      "step": 95435
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.1145281515165723e-05,
+      "loss": 0.0185,
+      "step": 95440
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.1144812713890583e-05,
+      "loss": 0.0541,
+      "step": 95445
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.1144343912615443e-05,
+      "loss": 0.0319,
+      "step": 95450
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.1143875111340302e-05,
+      "loss": 0.0499,
+      "step": 95455
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.1143406310065164e-05,
+      "loss": 0.117,
+      "step": 95460
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.1142937508790024e-05,
+      "loss": 0.0796,
+      "step": 95465
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.1142468707514886e-05,
+      "loss": 0.1977,
+      "step": 95470
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.1141999906239747e-05,
+      "loss": 0.2544,
+      "step": 95475
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.1141531104964607e-05,
+      "loss": 0.0535,
+      "step": 95480
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.1141062303689467e-05,
+      "loss": 0.0592,
+      "step": 95485
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.1140593502414327e-05,
+      "loss": 0.0436,
+      "step": 95490
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.1140124701139187e-05,
+      "loss": 0.0454,
+      "step": 95495
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.1139655899864049e-05,
+      "loss": 0.0685,
+      "step": 95500
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.1139187098588909e-05,
+      "loss": 0.0472,
+      "step": 95505
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.1138718297313768e-05,
+      "loss": 0.0682,
+      "step": 95510
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.1138249496038632e-05,
+      "loss": 0.0703,
+      "step": 95515
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.1137780694763492e-05,
+      "loss": 0.1434,
+      "step": 95520
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.1137311893488352e-05,
+      "loss": 0.3051,
+      "step": 95525
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.1136843092213212e-05,
+      "loss": 0.0487,
+      "step": 95530
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.1136374290938072e-05,
+      "loss": 0.0317,
+      "step": 95535
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.1135905489662933e-05,
+      "loss": 0.0255,
+      "step": 95540
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.1135436688387793e-05,
+      "loss": 0.0182,
+      "step": 95545
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.1134967887112653e-05,
+      "loss": 0.0671,
+      "step": 95550
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.1134499085837513e-05,
+      "loss": 0.0563,
+      "step": 95555
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.1134030284562376e-05,
+      "loss": 0.1338,
+      "step": 95560
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.1133561483287236e-05,
+      "loss": 0.069,
+      "step": 95565
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.1133092682012096e-05,
+      "loss": 0.1806,
+      "step": 95570
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.1132623880736956e-05,
+      "loss": 0.2525,
+      "step": 95575
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.1132155079461818e-05,
+      "loss": 0.0892,
+      "step": 95580
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.1131686278186678e-05,
+      "loss": 0.0073,
+      "step": 95585
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.1131217476911538e-05,
+      "loss": 0.0148,
+      "step": 95590
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.1130748675636398e-05,
+      "loss": 0.0387,
+      "step": 95595
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.1130279874361259e-05,
+      "loss": 0.0584,
+      "step": 95600
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.1129811073086119e-05,
+      "loss": 0.0672,
+      "step": 95605
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.112934227181098e-05,
+      "loss": 0.0871,
+      "step": 95610
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.112887347053584e-05,
+      "loss": 0.0725,
+      "step": 95615
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.1128404669260702e-05,
+      "loss": 0.1412,
+      "step": 95620
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.1127935867985562e-05,
+      "loss": 0.2711,
+      "step": 95625
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.1127467066710422e-05,
+      "loss": 0.0837,
+      "step": 95630
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.1126998265435282e-05,
+      "loss": 0.0292,
+      "step": 95635
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.1126529464160144e-05,
+      "loss": 0.0107,
+      "step": 95640
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.1126060662885004e-05,
+      "loss": 0.0795,
+      "step": 95645
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.1125591861609864e-05,
+      "loss": 0.0506,
+      "step": 95650
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.1125123060334725e-05,
+      "loss": 0.0403,
+      "step": 95655
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.1124654259059587e-05,
+      "loss": 0.1757,
+      "step": 95660
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.1124185457784447e-05,
+      "loss": 0.0437,
+      "step": 95665
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.1123716656509307e-05,
+      "loss": 0.1074,
+      "step": 95670
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.1123247855234167e-05,
+      "loss": 0.1198,
+      "step": 95675
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.1122779053959028e-05,
+      "loss": 0.053,
+      "step": 95680
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.1122310252683888e-05,
+      "loss": 0.0107,
+      "step": 95685
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.1121841451408748e-05,
+      "loss": 0.0362,
+      "step": 95690
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.1121372650133608e-05,
+      "loss": 0.0381,
+      "step": 95695
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.1120903848858471e-05,
+      "loss": 0.0542,
+      "step": 95700
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.1120435047583331e-05,
+      "loss": 0.0464,
+      "step": 95705
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.1119966246308191e-05,
+      "loss": 0.0754,
+      "step": 95710
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.1119497445033051e-05,
+      "loss": 0.1106,
+      "step": 95715
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.1119028643757913e-05,
+      "loss": 0.1349,
+      "step": 95720
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.1118559842482773e-05,
+      "loss": 0.2704,
+      "step": 95725
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.1118091041207633e-05,
+      "loss": 0.0986,
+      "step": 95730
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.1117622239932493e-05,
+      "loss": 0.0458,
+      "step": 95735
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.1117153438657353e-05,
+      "loss": 0.0116,
+      "step": 95740
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.1116684637382214e-05,
+      "loss": 0.0146,
+      "step": 95745
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.1116215836107076e-05,
+      "loss": 0.058,
+      "step": 95750
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.1115747034831936e-05,
+      "loss": 0.0346,
+      "step": 95755
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.1115278233556797e-05,
+      "loss": 0.1073,
+      "step": 95760
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.1114809432281657e-05,
+      "loss": 0.1048,
+      "step": 95765
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.1114340631006517e-05,
+      "loss": 0.1791,
+      "step": 95770
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.1113871829731377e-05,
+      "loss": 0.2328,
+      "step": 95775
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.1113403028456237e-05,
+      "loss": 0.0619,
+      "step": 95780
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.1112934227181099e-05,
+      "loss": 0.046,
+      "step": 95785
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.1112465425905959e-05,
+      "loss": 0.0233,
+      "step": 95790
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.111199662463082e-05,
+      "loss": 0.037,
+      "step": 95795
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.1111527823355682e-05,
+      "loss": 0.0482,
+      "step": 95800
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.1111059022080542e-05,
+      "loss": 0.0661,
+      "step": 95805
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.1110590220805402e-05,
+      "loss": 0.0709,
+      "step": 95810
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.1110121419530262e-05,
+      "loss": 0.1305,
+      "step": 95815
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.1109652618255122e-05,
+      "loss": 0.1817,
+      "step": 95820
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.1109183816979983e-05,
+      "loss": 0.1981,
+      "step": 95825
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.1108715015704843e-05,
+      "loss": 0.0654,
+      "step": 95830
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.1108246214429703e-05,
+      "loss": 0.0232,
+      "step": 95835
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.1107777413154566e-05,
+      "loss": 0.0285,
+      "step": 95840
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.1107308611879426e-05,
+      "loss": 0.0306,
+      "step": 95845
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.1106839810604286e-05,
+      "loss": 0.0335,
+      "step": 95850
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.1106371009329146e-05,
+      "loss": 0.0693,
+      "step": 95855
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.1105902208054006e-05,
+      "loss": 0.0439,
+      "step": 95860
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.1105433406778868e-05,
+      "loss": 0.0739,
+      "step": 95865
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.1104964605503728e-05,
+      "loss": 0.175,
+      "step": 95870
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.1104495804228588e-05,
+      "loss": 0.1857,
+      "step": 95875
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.1104027002953448e-05,
+      "loss": 0.0797,
+      "step": 95880
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.1103558201678311e-05,
+      "loss": 0.0092,
+      "step": 95885
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.1103089400403171e-05,
+      "loss": 0.0372,
+      "step": 95890
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.110262059912803e-05,
+      "loss": 0.036,
+      "step": 95895
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.110215179785289e-05,
+      "loss": 0.0757,
+      "step": 95900
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.1101682996577752e-05,
+      "loss": 0.0829,
+      "step": 95905
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.1101214195302612e-05,
+      "loss": 0.105,
+      "step": 95910
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.1100745394027472e-05,
+      "loss": 0.0873,
+      "step": 95915
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.1100276592752332e-05,
+      "loss": 0.1886,
+      "step": 95920
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.1099807791477192e-05,
+      "loss": 0.2514,
+      "step": 95925
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.1099338990202054e-05,
+      "loss": 0.0861,
+      "step": 95930
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.1098870188926915e-05,
+      "loss": 0.0121,
+      "step": 95935
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.1098401387651775e-05,
+      "loss": 0.047,
+      "step": 95940
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.1097932586376637e-05,
+      "loss": 0.0204,
+      "step": 95945
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.1097463785101497e-05,
+      "loss": 0.0529,
+      "step": 95950
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.1096994983826357e-05,
+      "loss": 0.0537,
+      "step": 95955
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.1096526182551217e-05,
+      "loss": 0.0243,
+      "step": 95960
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.1096057381276077e-05,
+      "loss": 0.1391,
+      "step": 95965
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.1095588580000938e-05,
+      "loss": 0.1364,
+      "step": 95970
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.1095119778725798e-05,
+      "loss": 0.1301,
+      "step": 95975
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.109465097745066e-05,
+      "loss": 0.0858,
+      "step": 95980
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.1094182176175521e-05,
+      "loss": 0.0424,
+      "step": 95985
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.1093713374900381e-05,
+      "loss": 0.0076,
+      "step": 95990
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.1093244573625241e-05,
+      "loss": 0.0142,
+      "step": 95995
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.1092775772350101e-05,
+      "loss": 0.0197,
+      "step": 96000
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.1092306971074961e-05,
+      "loss": 0.0696,
+      "step": 96005
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.1091838169799823e-05,
+      "loss": 0.0602,
+      "step": 96010
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.1091369368524683e-05,
+      "loss": 0.0852,
+      "step": 96015
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.1090900567249543e-05,
+      "loss": 0.1463,
+      "step": 96020
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.1090431765974406e-05,
+      "loss": 0.1661,
+      "step": 96025
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.1089962964699266e-05,
+      "loss": 0.1079,
+      "step": 96030
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.1089494163424126e-05,
+      "loss": 0.2677,
+      "step": 96035
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.1089025362148986e-05,
+      "loss": 0.061,
+      "step": 96040
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.1088556560873846e-05,
+      "loss": 0.0189,
+      "step": 96045
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.1088087759598707e-05,
+      "loss": 0.027,
+      "step": 96050
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.1087618958323567e-05,
+      "loss": 0.1223,
+      "step": 96055
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.1087150157048427e-05,
+      "loss": 0.1034,
+      "step": 96060
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.1086681355773287e-05,
+      "loss": 0.2181,
+      "step": 96065
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.1086212554498149e-05,
+      "loss": 0.2042,
+      "step": 96070
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.108574375322301e-05,
+      "loss": 0.255,
+      "step": 96075
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.108527495194787e-05,
+      "loss": 0.0895,
+      "step": 96080
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.108480615067273e-05,
+      "loss": 0.0154,
+      "step": 96085
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.1084337349397592e-05,
+      "loss": 0.0624,
+      "step": 96090
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.1083868548122452e-05,
+      "loss": 0.015,
+      "step": 96095
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.1083399746847312e-05,
+      "loss": 0.1369,
+      "step": 96100
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.1082930945572172e-05,
+      "loss": 0.1338,
+      "step": 96105
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.1082462144297033e-05,
+      "loss": 0.0547,
+      "step": 96110
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.1081993343021893e-05,
+      "loss": 0.0968,
+      "step": 96115
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.1081524541746755e-05,
+      "loss": 0.1239,
+      "step": 96120
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.1081055740471615e-05,
+      "loss": 0.185,
+      "step": 96125
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.1080586939196476e-05,
+      "loss": 0.1027,
+      "step": 96130
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.1080118137921336e-05,
+      "loss": 0.0296,
+      "step": 96135
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.1079649336646196e-05,
+      "loss": 0.0374,
+      "step": 96140
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.1079180535371056e-05,
+      "loss": 0.0256,
+      "step": 96145
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.1078711734095918e-05,
+      "loss": 0.1339,
+      "step": 96150
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.1078242932820778e-05,
+      "loss": 0.0594,
+      "step": 96155
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.1077774131545638e-05,
+      "loss": 0.1316,
+      "step": 96160
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.10773053302705e-05,
+      "loss": 0.0907,
+      "step": 96165
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.1076836528995361e-05,
+      "loss": 0.2559,
+      "step": 96170
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.1076367727720221e-05,
+      "loss": 0.1893,
+      "step": 96175
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.1075898926445081e-05,
+      "loss": 0.0744,
+      "step": 96180
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.107543012516994e-05,
+      "loss": 0.0304,
+      "step": 96185
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.1074961323894802e-05,
+      "loss": 0.0477,
+      "step": 96190
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.1074492522619662e-05,
+      "loss": 0.0658,
+      "step": 96195
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.1074023721344522e-05,
+      "loss": 0.0421,
+      "step": 96200
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.1073554920069382e-05,
+      "loss": 0.0284,
+      "step": 96205
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.1073086118794246e-05,
+      "loss": 0.0383,
+      "step": 96210
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.1072617317519106e-05,
+      "loss": 0.0802,
+      "step": 96215
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.1072148516243965e-05,
+      "loss": 0.145,
+      "step": 96220
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.1071679714968825e-05,
+      "loss": 0.227,
+      "step": 96225
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.1071210913693687e-05,
+      "loss": 0.0797,
+      "step": 96230
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.1070742112418547e-05,
+      "loss": 0.029,
+      "step": 96235
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.1070273311143407e-05,
+      "loss": 0.0172,
+      "step": 96240
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.1069804509868267e-05,
+      "loss": 0.0487,
+      "step": 96245
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.1069335708593127e-05,
+      "loss": 0.0531,
+      "step": 96250
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.1068866907317988e-05,
+      "loss": 0.0498,
+      "step": 96255
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.106839810604285e-05,
+      "loss": 0.1319,
+      "step": 96260
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.106792930476771e-05,
+      "loss": 0.0511,
+      "step": 96265
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.1067460503492572e-05,
+      "loss": 0.1645,
+      "step": 96270
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.1066991702217432e-05,
+      "loss": 0.2393,
+      "step": 96275
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.1066522900942291e-05,
+      "loss": 0.0761,
+      "step": 96280
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.1066054099667151e-05,
+      "loss": 0.0396,
+      "step": 96285
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.1065585298392011e-05,
+      "loss": 0.0337,
+      "step": 96290
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.1065116497116873e-05,
+      "loss": 0.0499,
+      "step": 96295
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.1064647695841733e-05,
+      "loss": 0.0542,
+      "step": 96300
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.1064178894566594e-05,
+      "loss": 0.1207,
+      "step": 96305
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.1063710093291456e-05,
+      "loss": 0.1424,
+      "step": 96310
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.1063241292016316e-05,
+      "loss": 0.1092,
+      "step": 96315
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.1062772490741176e-05,
+      "loss": 0.207,
+      "step": 96320
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.1062303689466036e-05,
+      "loss": 0.24,
+      "step": 96325
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.1061834888190896e-05,
+      "loss": 0.0939,
+      "step": 96330
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.1061366086915757e-05,
+      "loss": 0.0077,
+      "step": 96335
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.1060897285640617e-05,
+      "loss": 0.0183,
+      "step": 96340
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.1060428484365477e-05,
+      "loss": 0.0182,
+      "step": 96345
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.105995968309034e-05,
+      "loss": 0.0428,
+      "step": 96350
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.10594908818152e-05,
+      "loss": 0.0747,
+      "step": 96355
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.105902208054006e-05,
+      "loss": 0.0367,
+      "step": 96360
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.105855327926492e-05,
+      "loss": 0.133,
+      "step": 96365
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.105808447798978e-05,
+      "loss": 0.1218,
+      "step": 96370
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.1057615676714642e-05,
+      "loss": 0.2208,
+      "step": 96375
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.1057146875439502e-05,
+      "loss": 0.0698,
+      "step": 96380
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.1056678074164362e-05,
+      "loss": 0.0295,
+      "step": 96385
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.1056209272889222e-05,
+      "loss": 0.035,
+      "step": 96390
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.1055740471614082e-05,
+      "loss": 0.0509,
+      "step": 96395
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.1055271670338945e-05,
+      "loss": 0.0959,
+      "step": 96400
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.1054802869063805e-05,
+      "loss": 0.0893,
+      "step": 96405
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.1054334067788665e-05,
+      "loss": 0.0758,
+      "step": 96410
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.1053865266513527e-05,
+      "loss": 0.1056,
+      "step": 96415
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.1053396465238387e-05,
+      "loss": 0.1086,
+      "step": 96420
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.1052927663963246e-05,
+      "loss": 0.3389,
+      "step": 96425
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.1052458862688106e-05,
+      "loss": 0.0704,
+      "step": 96430
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.1051990061412966e-05,
+      "loss": 0.0478,
+      "step": 96435
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.1051521260137828e-05,
+      "loss": 0.0457,
+      "step": 96440
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.105105245886269e-05,
+      "loss": 0.0443,
+      "step": 96445
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.105058365758755e-05,
+      "loss": 0.0494,
+      "step": 96450
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.1050114856312411e-05,
+      "loss": 0.0535,
+      "step": 96455
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.1049646055037271e-05,
+      "loss": 0.1155,
+      "step": 96460
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.1049177253762131e-05,
+      "loss": 0.0679,
+      "step": 96465
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.1048708452486991e-05,
+      "loss": 0.1323,
+      "step": 96470
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.1048239651211851e-05,
+      "loss": 0.2943,
+      "step": 96475
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.1047770849936713e-05,
+      "loss": 0.1359,
+      "step": 96480
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.1047302048661572e-05,
+      "loss": 0.0188,
+      "step": 96485
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.1046833247386434e-05,
+      "loss": 0.0058,
+      "step": 96490
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.1046364446111296e-05,
+      "loss": 0.1123,
+      "step": 96495
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.1045895644836156e-05,
+      "loss": 0.092,
+      "step": 96500
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.1045426843561016e-05,
+      "loss": 0.0793,
+      "step": 96505
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.1044958042285875e-05,
+      "loss": 0.0414,
+      "step": 96510
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.1044489241010735e-05,
+      "loss": 0.0444,
+      "step": 96515
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.1044020439735597e-05,
+      "loss": 0.1698,
+      "step": 96520
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.1043551638460457e-05,
+      "loss": 0.2276,
+      "step": 96525
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.1043082837185317e-05,
+      "loss": 0.0498,
+      "step": 96530
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.104261403591018e-05,
+      "loss": 0.0123,
+      "step": 96535
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.104214523463504e-05,
+      "loss": 0.0793,
+      "step": 96540
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.10416764333599e-05,
+      "loss": 0.0751,
+      "step": 96545
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.104120763208476e-05,
+      "loss": 0.044,
+      "step": 96550
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.1040738830809622e-05,
+      "loss": 0.0604,
+      "step": 96555
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.1040270029534482e-05,
+      "loss": 0.0385,
+      "step": 96560
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.1039801228259342e-05,
+      "loss": 0.0994,
+      "step": 96565
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.1039332426984201e-05,
+      "loss": 0.1932,
+      "step": 96570
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.1038863625709061e-05,
+      "loss": 0.2327,
+      "step": 96575
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.1038394824433923e-05,
+      "loss": 0.0545,
+      "step": 96580
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.1037926023158785e-05,
+      "loss": 0.2896,
+      "step": 96585
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.1037457221883645e-05,
+      "loss": 0.0264,
+      "step": 96590
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.1036988420608506e-05,
+      "loss": 0.0415,
+      "step": 96595
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.1036519619333366e-05,
+      "loss": 0.0908,
+      "step": 96600
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.1036050818058226e-05,
+      "loss": 0.0537,
+      "step": 96605
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.1035582016783086e-05,
+      "loss": 0.0708,
+      "step": 96610
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.1035113215507946e-05,
+      "loss": 0.0977,
+      "step": 96615
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.1034644414232808e-05,
+      "loss": 0.1554,
+      "step": 96620
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.1034175612957668e-05,
+      "loss": 0.3359,
+      "step": 96625
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.1033706811682529e-05,
+      "loss": 0.095,
+      "step": 96630
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.103323801040739e-05,
+      "loss": 0.0484,
+      "step": 96635
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.103276920913225e-05,
+      "loss": 0.0477,
+      "step": 96640
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.103230040785711e-05,
+      "loss": 0.1291,
+      "step": 96645
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.103183160658197e-05,
+      "loss": 0.0511,
+      "step": 96650
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.103136280530683e-05,
+      "loss": 0.0412,
+      "step": 96655
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.1030894004031692e-05,
+      "loss": 0.0718,
+      "step": 96660
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.1030425202756552e-05,
+      "loss": 0.1408,
+      "step": 96665
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.1029956401481412e-05,
+      "loss": 0.1417,
+      "step": 96670
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.1029487600206275e-05,
+      "loss": 0.3219,
+      "step": 96675
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.1029018798931135e-05,
+      "loss": 0.0997,
+      "step": 96680
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.1028549997655995e-05,
+      "loss": 0.0185,
+      "step": 96685
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.1028081196380855e-05,
+      "loss": 0.0311,
+      "step": 96690
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.1027612395105715e-05,
+      "loss": 0.0496,
+      "step": 96695
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.1027143593830577e-05,
+      "loss": 0.0159,
+      "step": 96700
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.1026674792555437e-05,
+      "loss": 0.0784,
+      "step": 96705
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.1026205991280297e-05,
+      "loss": 0.1331,
+      "step": 96710
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.1025737190005156e-05,
+      "loss": 0.1623,
+      "step": 96715
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.1025268388730016e-05,
+      "loss": 0.1472,
+      "step": 96720
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.102479958745488e-05,
+      "loss": 0.2449,
+      "step": 96725
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.102433078617974e-05,
+      "loss": 0.0831,
+      "step": 96730
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.10238619849046e-05,
+      "loss": 0.0175,
+      "step": 96735
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.1023393183629461e-05,
+      "loss": 0.0469,
+      "step": 96740
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.1022924382354321e-05,
+      "loss": 0.0333,
+      "step": 96745
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.1022455581079181e-05,
+      "loss": 0.0576,
+      "step": 96750
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.1021986779804041e-05,
+      "loss": 0.0975,
+      "step": 96755
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.1021517978528901e-05,
+      "loss": 0.1719,
+      "step": 96760
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.1021049177253763e-05,
+      "loss": 0.1369,
+      "step": 96765
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.1020580375978624e-05,
+      "loss": 0.1846,
+      "step": 96770
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.1020111574703484e-05,
+      "loss": 0.1975,
+      "step": 96775
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.1019642773428346e-05,
+      "loss": 0.0786,
+      "step": 96780
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.1019173972153206e-05,
+      "loss": 0.0112,
+      "step": 96785
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.1018705170878066e-05,
+      "loss": 0.0262,
+      "step": 96790
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.1018236369602926e-05,
+      "loss": 0.0202,
+      "step": 96795
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.1017767568327786e-05,
+      "loss": 0.0724,
+      "step": 96800
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.1017298767052647e-05,
+      "loss": 0.0506,
+      "step": 96805
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.1016829965777507e-05,
+      "loss": 0.0837,
+      "step": 96810
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.1016361164502369e-05,
+      "loss": 0.1212,
+      "step": 96815
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.101589236322723e-05,
+      "loss": 0.1663,
+      "step": 96820
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.101542356195209e-05,
+      "loss": 0.298,
+      "step": 96825
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.101495476067695e-05,
+      "loss": 0.0688,
+      "step": 96830
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.101448595940181e-05,
+      "loss": 0.0154,
+      "step": 96835
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.101401715812667e-05,
+      "loss": 0.0637,
+      "step": 96840
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.1013548356851532e-05,
+      "loss": 0.0434,
+      "step": 96845
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.1013079555576392e-05,
+      "loss": 0.0713,
+      "step": 96850
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.1012610754301252e-05,
+      "loss": 0.0701,
+      "step": 96855
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.1012141953026115e-05,
+      "loss": 0.0347,
+      "step": 96860
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.1011673151750975e-05,
+      "loss": 0.0871,
+      "step": 96865
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.1011204350475835e-05,
+      "loss": 0.1951,
+      "step": 96870
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.1010735549200695e-05,
+      "loss": 0.2353,
+      "step": 96875
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.1010266747925555e-05,
+      "loss": 0.0605,
+      "step": 96880
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.1009797946650416e-05,
+      "loss": 0.0668,
+      "step": 96885
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.1009329145375276e-05,
+      "loss": 0.0718,
+      "step": 96890
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.1008860344100136e-05,
+      "loss": 0.0293,
+      "step": 96895
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.1008391542824996e-05,
+      "loss": 0.0657,
+      "step": 96900
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.1007922741549858e-05,
+      "loss": 0.0703,
+      "step": 96905
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.100745394027472e-05,
+      "loss": 0.1935,
+      "step": 96910
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.100698513899958e-05,
+      "loss": 0.0927,
+      "step": 96915
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.100651633772444e-05,
+      "loss": 0.1122,
+      "step": 96920
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.10060475364493e-05,
+      "loss": 0.2593,
+      "step": 96925
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.100557873517416e-05,
+      "loss": 0.0908,
+      "step": 96930
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.100510993389902e-05,
+      "loss": 0.0327,
+      "step": 96935
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.100464113262388e-05,
+      "loss": 0.0196,
+      "step": 96940
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.1004172331348742e-05,
+      "loss": 0.0489,
+      "step": 96945
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.1003703530073602e-05,
+      "loss": 0.0452,
+      "step": 96950
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.1003234728798464e-05,
+      "loss": 0.0649,
+      "step": 96955
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.1002765927523324e-05,
+      "loss": 0.0449,
+      "step": 96960
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.1002297126248185e-05,
+      "loss": 0.1372,
+      "step": 96965
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.1001828324973045e-05,
+      "loss": 0.0901,
+      "step": 96970
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.1001359523697905e-05,
+      "loss": 0.3421,
+      "step": 96975
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.1000890722422765e-05,
+      "loss": 0.1052,
+      "step": 96980
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.1000421921147627e-05,
+      "loss": 0.014,
+      "step": 96985
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.0999953119872487e-05,
+      "loss": 0.065,
+      "step": 96990
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.0999484318597347e-05,
+      "loss": 0.0455,
+      "step": 96995
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.0999015517322208e-05,
+      "loss": 0.1408,
+      "step": 97000
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.099854671604707e-05,
+      "loss": 0.0374,
+      "step": 97005
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.099807791477193e-05,
+      "loss": 0.1551,
+      "step": 97010
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.099760911349679e-05,
+      "loss": 0.1214,
+      "step": 97015
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.099714031222165e-05,
+      "loss": 0.1428,
+      "step": 97020
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.0996671510946511e-05,
+      "loss": 0.4012,
+      "step": 97025
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.0996202709671371e-05,
+      "loss": 0.087,
+      "step": 97030
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.0995733908396231e-05,
+      "loss": 0.0144,
+      "step": 97035
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.0995265107121091e-05,
+      "loss": 0.0311,
+      "step": 97040
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.0994796305845951e-05,
+      "loss": 0.0335,
+      "step": 97045
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.0994327504570814e-05,
+      "loss": 0.0715,
+      "step": 97050
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.0993858703295674e-05,
+      "loss": 0.0725,
+      "step": 97055
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.0993389902020534e-05,
+      "loss": 0.1107,
+      "step": 97060
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.0992921100745396e-05,
+      "loss": 0.1399,
+      "step": 97065
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.0992452299470256e-05,
+      "loss": 0.2294,
+      "step": 97070
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.0991983498195116e-05,
+      "loss": 0.2816,
+      "step": 97075
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.0991514696919976e-05,
+      "loss": 0.0701,
+      "step": 97080
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.0991045895644836e-05,
+      "loss": 0.0097,
+      "step": 97085
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.0990577094369697e-05,
+      "loss": 0.0163,
+      "step": 97090
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.0990108293094559e-05,
+      "loss": 0.0571,
+      "step": 97095
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.0989639491819419e-05,
+      "loss": 0.0786,
+      "step": 97100
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.098917069054428e-05,
+      "loss": 0.0738,
+      "step": 97105
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.098870188926914e-05,
+      "loss": 0.0865,
+      "step": 97110
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.0988233087994e-05,
+      "loss": 0.088,
+      "step": 97115
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.098776428671886e-05,
+      "loss": 0.1743,
+      "step": 97120
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.098729548544372e-05,
+      "loss": 0.2513,
+      "step": 97125
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.0986826684168582e-05,
+      "loss": 0.069,
+      "step": 97130
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.0986357882893442e-05,
+      "loss": 0.0248,
+      "step": 97135
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.0985889081618303e-05,
+      "loss": 0.0325,
+      "step": 97140
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.0985420280343165e-05,
+      "loss": 0.0395,
+      "step": 97145
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.0984951479068025e-05,
+      "loss": 0.0381,
+      "step": 97150
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.0984482677792885e-05,
+      "loss": 0.0915,
+      "step": 97155
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.0984013876517745e-05,
+      "loss": 0.0811,
+      "step": 97160
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.0983545075242605e-05,
+      "loss": 0.1434,
+      "step": 97165
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.0983076273967466e-05,
+      "loss": 0.2167,
+      "step": 97170
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.0982607472692326e-05,
+      "loss": 0.3037,
+      "step": 97175
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.0982138671417186e-05,
+      "loss": 0.0661,
+      "step": 97180
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.098166987014205e-05,
+      "loss": 0.043,
+      "step": 97185
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.098120106886691e-05,
+      "loss": 0.051,
+      "step": 97190
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.098073226759177e-05,
+      "loss": 0.012,
+      "step": 97195
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.098026346631663e-05,
+      "loss": 0.0995,
+      "step": 97200
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.097979466504149e-05,
+      "loss": 0.0614,
+      "step": 97205
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.0979325863766351e-05,
+      "loss": 0.0419,
+      "step": 97210
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.097885706249121e-05,
+      "loss": 0.1193,
+      "step": 97215
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.097838826121607e-05,
+      "loss": 0.0944,
+      "step": 97220
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.097791945994093e-05,
+      "loss": 0.3821,
+      "step": 97225
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.097745065866579e-05,
+      "loss": 0.0986,
+      "step": 97230
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.0976981857390654e-05,
+      "loss": 0.0231,
+      "step": 97235
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.0976513056115514e-05,
+      "loss": 0.0322,
+      "step": 97240
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.0976044254840374e-05,
+      "loss": 0.0597,
+      "step": 97245
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.0975575453565235e-05,
+      "loss": 0.0502,
+      "step": 97250
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.0975106652290095e-05,
+      "loss": 0.0478,
+      "step": 97255
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.0974637851014955e-05,
+      "loss": 0.1236,
+      "step": 97260
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.0974169049739815e-05,
+      "loss": 0.1108,
+      "step": 97265
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.0973700248464675e-05,
+      "loss": 0.1377,
+      "step": 97270
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.0973231447189537e-05,
+      "loss": 0.2356,
+      "step": 97275
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.0972762645914398e-05,
+      "loss": 0.0937,
+      "step": 97280
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.0972293844639258e-05,
+      "loss": 0.0699,
+      "step": 97285
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.097182504336412e-05,
+      "loss": 0.0228,
+      "step": 97290
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.097135624208898e-05,
+      "loss": 0.0322,
+      "step": 97295
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.097088744081384e-05,
+      "loss": 0.0416,
+      "step": 97300
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.09704186395387e-05,
+      "loss": 0.0185,
+      "step": 97305
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.096994983826356e-05,
+      "loss": 0.081,
+      "step": 97310
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.0969481036988421e-05,
+      "loss": 0.0595,
+      "step": 97315
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.0969012235713281e-05,
+      "loss": 0.2023,
+      "step": 97320
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.0968543434438143e-05,
+      "loss": 0.2479,
+      "step": 97325
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.0968074633163005e-05,
+      "loss": 0.0563,
+      "step": 97330
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.0967605831887864e-05,
+      "loss": 0.0366,
+      "step": 97335
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.0967137030612724e-05,
+      "loss": 0.0332,
+      "step": 97340
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.0966668229337584e-05,
+      "loss": 0.1006,
+      "step": 97345
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.0966199428062444e-05,
+      "loss": 0.0936,
+      "step": 97350
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.0965730626787306e-05,
+      "loss": 0.0536,
+      "step": 97355
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.0965261825512166e-05,
+      "loss": 0.1128,
+      "step": 97360
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.0964793024237026e-05,
+      "loss": 0.0861,
+      "step": 97365
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.0964324222961886e-05,
+      "loss": 0.1503,
+      "step": 97370
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.0963855421686749e-05,
+      "loss": 0.2551,
+      "step": 97375
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.0963386620411609e-05,
+      "loss": 0.0587,
+      "step": 97380
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.0962917819136469e-05,
+      "loss": 0.007,
+      "step": 97385
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.0962449017861329e-05,
+      "loss": 0.0194,
+      "step": 97390
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.096198021658619e-05,
+      "loss": 0.0409,
+      "step": 97395
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.096151141531105e-05,
+      "loss": 0.0656,
+      "step": 97400
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.096104261403591e-05,
+      "loss": 0.0702,
+      "step": 97405
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.096057381276077e-05,
+      "loss": 0.0919,
+      "step": 97410
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0960105011485632e-05,
+      "loss": 0.0922,
+      "step": 97415
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0959636210210494e-05,
+      "loss": 0.1553,
+      "step": 97420
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0959167408935353e-05,
+      "loss": 0.1833,
+      "step": 97425
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0958698607660213e-05,
+      "loss": 0.0837,
+      "step": 97430
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0958229806385075e-05,
+      "loss": 0.0326,
+      "step": 97435
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0957761005109935e-05,
+      "loss": 0.0537,
+      "step": 97440
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0957292203834795e-05,
+      "loss": 0.0605,
+      "step": 97445
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0956823402559655e-05,
+      "loss": 0.0339,
+      "step": 97450
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0956354601284516e-05,
+      "loss": 0.0426,
+      "step": 97455
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0955885800009376e-05,
+      "loss": 0.0434,
+      "step": 97460
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0955416998734238e-05,
+      "loss": 0.1599,
+      "step": 97465
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0954948197459098e-05,
+      "loss": 0.1171,
+      "step": 97470
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.095447939618396e-05,
+      "loss": 0.3212,
+      "step": 97475
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.095401059490882e-05,
+      "loss": 0.0729,
+      "step": 97480
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.095354179363368e-05,
+      "loss": 0.0176,
+      "step": 97485
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.095307299235854e-05,
+      "loss": 0.0497,
+      "step": 97490
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0952604191083401e-05,
+      "loss": 0.0452,
+      "step": 97495
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0952135389808261e-05,
+      "loss": 0.0537,
+      "step": 97500
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0951666588533121e-05,
+      "loss": 0.0738,
+      "step": 97505
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0951197787257983e-05,
+      "loss": 0.0585,
+      "step": 97510
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0950728985982844e-05,
+      "loss": 0.0748,
+      "step": 97515
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0950260184707704e-05,
+      "loss": 0.113,
+      "step": 97520
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0949791383432564e-05,
+      "loss": 0.1528,
+      "step": 97525
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0949322582157424e-05,
+      "loss": 0.1041,
+      "step": 97530
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0948853780882286e-05,
+      "loss": 0.012,
+      "step": 97535
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0948384979607145e-05,
+      "loss": 0.0394,
+      "step": 97540
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0947916178332005e-05,
+      "loss": 0.0728,
+      "step": 97545
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0947447377056865e-05,
+      "loss": 0.0417,
+      "step": 97550
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0946978575781725e-05,
+      "loss": 0.0681,
+      "step": 97555
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0946509774506589e-05,
+      "loss": 0.0422,
+      "step": 97560
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0946040973231449e-05,
+      "loss": 0.0566,
+      "step": 97565
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0945572171956308e-05,
+      "loss": 0.1136,
+      "step": 97570
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.094510337068117e-05,
+      "loss": 0.3118,
+      "step": 97575
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.094463456940603e-05,
+      "loss": 0.0719,
+      "step": 97580
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.094416576813089e-05,
+      "loss": 0.0333,
+      "step": 97585
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.094369696685575e-05,
+      "loss": 0.0595,
+      "step": 97590
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.094322816558061e-05,
+      "loss": 0.0278,
+      "step": 97595
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0942759364305471e-05,
+      "loss": 0.0473,
+      "step": 97600
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0942290563030333e-05,
+      "loss": 0.0964,
+      "step": 97605
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0941821761755193e-05,
+      "loss": 0.1222,
+      "step": 97610
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0941352960480055e-05,
+      "loss": 0.1608,
+      "step": 97615
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0940884159204915e-05,
+      "loss": 0.1823,
+      "step": 97620
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0940415357929775e-05,
+      "loss": 0.4506,
+      "step": 97625
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0939946556654634e-05,
+      "loss": 0.1061,
+      "step": 97630
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0939477755379494e-05,
+      "loss": 0.0073,
+      "step": 97635
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0939008954104356e-05,
+      "loss": 0.044,
+      "step": 97640
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0938540152829216e-05,
+      "loss": 0.0332,
+      "step": 97645
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0938071351554078e-05,
+      "loss": 0.0764,
+      "step": 97650
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.093760255027894e-05,
+      "loss": 0.0479,
+      "step": 97655
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0937133749003799e-05,
+      "loss": 0.0683,
+      "step": 97660
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0936664947728659e-05,
+      "loss": 0.0851,
+      "step": 97665
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0936196146453519e-05,
+      "loss": 0.1891,
+      "step": 97670
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0935727345178379e-05,
+      "loss": 0.149,
+      "step": 97675
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.093525854390324e-05,
+      "loss": 0.0344,
+      "step": 97680
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.09347897426281e-05,
+      "loss": 0.0037,
+      "step": 97685
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.093432094135296e-05,
+      "loss": 0.0173,
+      "step": 97690
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.093385214007782e-05,
+      "loss": 0.0239,
+      "step": 97695
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0933383338802684e-05,
+      "loss": 0.0625,
+      "step": 97700
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0932914537527544e-05,
+      "loss": 0.0484,
+      "step": 97705
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0932445736252404e-05,
+      "loss": 0.028,
+      "step": 97710
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0931976934977263e-05,
+      "loss": 0.0654,
+      "step": 97715
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0931508133702125e-05,
+      "loss": 0.1062,
+      "step": 97720
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0931039332426985e-05,
+      "loss": 0.2395,
+      "step": 97725
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0930570531151845e-05,
+      "loss": 0.0758,
+      "step": 97730
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0930101729876705e-05,
+      "loss": 0.0345,
+      "step": 97735
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0929632928601565e-05,
+      "loss": 0.0282,
+      "step": 97740
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0929164127326428e-05,
+      "loss": 0.03,
+      "step": 97745
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0928695326051288e-05,
+      "loss": 0.0638,
+      "step": 97750
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0928226524776148e-05,
+      "loss": 0.0724,
+      "step": 97755
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.092775772350101e-05,
+      "loss": 0.1126,
+      "step": 97760
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.092728892222587e-05,
+      "loss": 0.1215,
+      "step": 97765
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.092682012095073e-05,
+      "loss": 0.1396,
+      "step": 97770
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.092635131967559e-05,
+      "loss": 0.3429,
+      "step": 97775
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.092588251840045e-05,
+      "loss": 0.0768,
+      "step": 97780
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0925413717125311e-05,
+      "loss": 0.0162,
+      "step": 97785
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0924944915850173e-05,
+      "loss": 0.0282,
+      "step": 97790
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0924476114575033e-05,
+      "loss": 0.0279,
+      "step": 97795
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0924007313299894e-05,
+      "loss": 0.0783,
+      "step": 97800
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0923538512024754e-05,
+      "loss": 0.0671,
+      "step": 97805
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0923069710749614e-05,
+      "loss": 0.0647,
+      "step": 97810
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0922600909474474e-05,
+      "loss": 0.1167,
+      "step": 97815
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0922132108199334e-05,
+      "loss": 0.0914,
+      "step": 97820
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0921663306924196e-05,
+      "loss": 0.2441,
+      "step": 97825
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0921194505649056e-05,
+      "loss": 0.0606,
+      "step": 97830
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.0920725704373917e-05,
+      "loss": 0.0078,
+      "step": 97835
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.0920256903098779e-05,
+      "loss": 0.0232,
+      "step": 97840
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.0919788101823639e-05,
+      "loss": 0.0427,
+      "step": 97845
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.0919319300548499e-05,
+      "loss": 0.0964,
+      "step": 97850
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.0918850499273359e-05,
+      "loss": 0.0811,
+      "step": 97855
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.0918381697998219e-05,
+      "loss": 0.1582,
+      "step": 97860
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.091791289672308e-05,
+      "loss": 0.1579,
+      "step": 97865
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.091744409544794e-05,
+      "loss": 0.1226,
+      "step": 97870
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.09169752941728e-05,
+      "loss": 0.1606,
+      "step": 97875
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.091650649289766e-05,
+      "loss": 0.0293,
+      "step": 97880
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.0916037691622523e-05,
+      "loss": 0.0428,
+      "step": 97885
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.0915568890347383e-05,
+      "loss": 0.0083,
+      "step": 97890
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.0915100089072243e-05,
+      "loss": 0.0634,
+      "step": 97895
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.0914631287797103e-05,
+      "loss": 0.0564,
+      "step": 97900
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.0914162486521965e-05,
+      "loss": 0.0286,
+      "step": 97905
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.0913693685246825e-05,
+      "loss": 0.0439,
+      "step": 97910
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.0913224883971685e-05,
+      "loss": 0.1408,
+      "step": 97915
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.0912756082696544e-05,
+      "loss": 0.1428,
+      "step": 97920
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.0912287281421406e-05,
+      "loss": 0.2285,
+      "step": 97925
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.0911818480146268e-05,
+      "loss": 0.0525,
+      "step": 97930
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.0911349678871128e-05,
+      "loss": 0.034,
+      "step": 97935
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.0910880877595988e-05,
+      "loss": 0.0188,
+      "step": 97940
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.091041207632085e-05,
+      "loss": 0.0551,
+      "step": 97945
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.090994327504571e-05,
+      "loss": 0.1143,
+      "step": 97950
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.0909474473770569e-05,
+      "loss": 0.0686,
+      "step": 97955
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.0909005672495429e-05,
+      "loss": 0.0711,
+      "step": 97960
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.090853687122029e-05,
+      "loss": 0.0532,
+      "step": 97965
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.090806806994515e-05,
+      "loss": 0.1223,
+      "step": 97970
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.0907599268670012e-05,
+      "loss": 0.4614,
+      "step": 97975
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.0907130467394874e-05,
+      "loss": 0.0924,
+      "step": 97980
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.0906661666119734e-05,
+      "loss": 0.044,
+      "step": 97985
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.0906192864844594e-05,
+      "loss": 0.0362,
+      "step": 97990
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.0905724063569454e-05,
+      "loss": 0.0519,
+      "step": 97995
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.0905255262294314e-05,
+      "loss": 0.1229,
+      "step": 98000
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.0904786461019175e-05,
+      "loss": 0.0834,
+      "step": 98005
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.0904317659744035e-05,
+      "loss": 0.0484,
+      "step": 98010
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.0903848858468895e-05,
+      "loss": 0.0684,
+      "step": 98015
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.0903380057193755e-05,
+      "loss": 0.1274,
+      "step": 98020
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.0902911255918618e-05,
+      "loss": 0.2913,
+      "step": 98025
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.0902442454643478e-05,
+      "loss": 0.0674,
+      "step": 98030
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.0901973653368338e-05,
+      "loss": 0.0507,
+      "step": 98035
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.0901504852093198e-05,
+      "loss": 0.0237,
+      "step": 98040
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.090103605081806e-05,
+      "loss": 0.0243,
+      "step": 98045
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.090056724954292e-05,
+      "loss": 0.0373,
+      "step": 98050
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.090009844826778e-05,
+      "loss": 0.0581,
+      "step": 98055
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.089962964699264e-05,
+      "loss": 0.0991,
+      "step": 98060
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.08991608457175e-05,
+      "loss": 0.0774,
+      "step": 98065
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.0898692044442363e-05,
+      "loss": 0.1783,
+      "step": 98070
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.0898223243167223e-05,
+      "loss": 0.1901,
+      "step": 98075
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.0897754441892083e-05,
+      "loss": 0.065,
+      "step": 98080
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.0897285640616944e-05,
+      "loss": 0.0115,
+      "step": 98085
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.0896816839341804e-05,
+      "loss": 0.0487,
+      "step": 98090
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.0896348038066664e-05,
+      "loss": 0.0425,
+      "step": 98095
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.0895879236791524e-05,
+      "loss": 0.0566,
+      "step": 98100
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.0895410435516384e-05,
+      "loss": 0.0341,
+      "step": 98105
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.0894941634241246e-05,
+      "loss": 0.0749,
+      "step": 98110
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.0894472832966107e-05,
+      "loss": 0.0787,
+      "step": 98115
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.0894004031690967e-05,
+      "loss": 0.0786,
+      "step": 98120
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.0893535230415829e-05,
+      "loss": 0.1973,
+      "step": 98125
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.0893066429140689e-05,
+      "loss": 0.1024,
+      "step": 98130
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.0892597627865549e-05,
+      "loss": 0.0435,
+      "step": 98135
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.0892128826590409e-05,
+      "loss": 0.0598,
+      "step": 98140
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.0891660025315269e-05,
+      "loss": 0.0788,
+      "step": 98145
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.089119122404013e-05,
+      "loss": 0.0257,
+      "step": 98150
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.089072242276499e-05,
+      "loss": 0.0886,
+      "step": 98155
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.0890253621489852e-05,
+      "loss": 0.0717,
+      "step": 98160
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.0889784820214713e-05,
+      "loss": 0.1966,
+      "step": 98165
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.0889316018939573e-05,
+      "loss": 0.1449,
+      "step": 98170
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.0888847217664433e-05,
+      "loss": 0.1456,
+      "step": 98175
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.0888378416389293e-05,
+      "loss": 0.0836,
+      "step": 98180
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.0887909615114153e-05,
+      "loss": 0.0204,
+      "step": 98185
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.0887440813839015e-05,
+      "loss": 0.0212,
+      "step": 98190
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.0886972012563875e-05,
+      "loss": 0.0514,
+      "step": 98195
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.0886503211288735e-05,
+      "loss": 0.0795,
+      "step": 98200
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.0886034410013595e-05,
+      "loss": 0.0567,
+      "step": 98205
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.0885565608738458e-05,
+      "loss": 0.0893,
+      "step": 98210
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.0885096807463318e-05,
+      "loss": 0.1304,
+      "step": 98215
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.0884628006188178e-05,
+      "loss": 0.2232,
+      "step": 98220
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.0884159204913038e-05,
+      "loss": 0.3183,
+      "step": 98225
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.08836904036379e-05,
+      "loss": 0.0753,
+      "step": 98230
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.088322160236276e-05,
+      "loss": 0.025,
+      "step": 98235
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.088275280108762e-05,
+      "loss": 0.018,
+      "step": 98240
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.0882283999812479e-05,
+      "loss": 0.0605,
+      "step": 98245
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.0881815198537339e-05,
+      "loss": 0.0564,
+      "step": 98250
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.0881346397262202e-05,
+      "loss": 0.1028,
+      "step": 98255
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.0880877595987062e-05,
+      "loss": 0.0965,
+      "step": 98260
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.0880408794711922e-05,
+      "loss": 0.0909,
+      "step": 98265
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.0879939993436784e-05,
+      "loss": 0.1554,
+      "step": 98270
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.0879471192161644e-05,
+      "loss": 0.2125,
+      "step": 98275
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.0879002390886504e-05,
+      "loss": 0.0855,
+      "step": 98280
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.0878533589611364e-05,
+      "loss": 0.0064,
+      "step": 98285
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.0878064788336225e-05,
+      "loss": 0.0419,
+      "step": 98290
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.0877595987061085e-05,
+      "loss": 0.0686,
+      "step": 98295
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.0877127185785947e-05,
+      "loss": 0.0698,
+      "step": 98300
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.0876658384510807e-05,
+      "loss": 0.071,
+      "step": 98305
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.0876189583235668e-05,
+      "loss": 0.041,
+      "step": 98310
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.0875720781960528e-05,
+      "loss": 0.1351,
+      "step": 98315
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.0875251980685388e-05,
+      "loss": 0.0993,
+      "step": 98320
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.0874783179410248e-05,
+      "loss": 0.2124,
+      "step": 98325
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.087431437813511e-05,
+      "loss": 0.1036,
+      "step": 98330
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.087384557685997e-05,
+      "loss": 0.0395,
+      "step": 98335
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.087337677558483e-05,
+      "loss": 0.0865,
+      "step": 98340
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.087290797430969e-05,
+      "loss": 0.0526,
+      "step": 98345
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.0872439173034553e-05,
+      "loss": 0.0604,
+      "step": 98350
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.0871970371759413e-05,
+      "loss": 0.0464,
+      "step": 98355
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.0871501570484273e-05,
+      "loss": 0.1454,
+      "step": 98360
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.0871032769209133e-05,
+      "loss": 0.1192,
+      "step": 98365
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.0870563967933994e-05,
+      "loss": 0.1233,
+      "step": 98370
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.0870095166658854e-05,
+      "loss": 0.2417,
+      "step": 98375
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.0869626365383714e-05,
+      "loss": 0.0797,
+      "step": 98380
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.0869157564108574e-05,
+      "loss": 0.0076,
+      "step": 98385
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.0868688762833434e-05,
+      "loss": 0.0633,
+      "step": 98390
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.0868219961558297e-05,
+      "loss": 0.0569,
+      "step": 98395
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.0867751160283157e-05,
+      "loss": 0.0983,
+      "step": 98400
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.0867282359008017e-05,
+      "loss": 0.0639,
+      "step": 98405
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.0866813557732879e-05,
+      "loss": 0.0683,
+      "step": 98410
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.0866344756457739e-05,
+      "loss": 0.207,
+      "step": 98415
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.0865875955182599e-05,
+      "loss": 0.2157,
+      "step": 98420
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.0865407153907459e-05,
+      "loss": 0.1907,
+      "step": 98425
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.0864938352632319e-05,
+      "loss": 0.0587,
+      "step": 98430
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.086446955135718e-05,
+      "loss": 0.0251,
+      "step": 98435
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.0864000750082042e-05,
+      "loss": 0.0553,
+      "step": 98440
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.0863531948806902e-05,
+      "loss": 0.0419,
+      "step": 98445
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.0863063147531764e-05,
+      "loss": 0.0544,
+      "step": 98450
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.0862594346256623e-05,
+      "loss": 0.0795,
+      "step": 98455
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.0862125544981483e-05,
+      "loss": 0.1192,
+      "step": 98460
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.0861656743706343e-05,
+      "loss": 0.0826,
+      "step": 98465
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.0861187942431203e-05,
+      "loss": 0.2061,
+      "step": 98470
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.0860719141156065e-05,
+      "loss": 0.2154,
+      "step": 98475
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.0860250339880925e-05,
+      "loss": 0.0606,
+      "step": 98480
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.0859781538605786e-05,
+      "loss": 0.0192,
+      "step": 98485
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.0859312737330648e-05,
+      "loss": 0.0225,
+      "step": 98490
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.0858843936055508e-05,
+      "loss": 0.0393,
+      "step": 98495
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.0858375134780368e-05,
+      "loss": 0.0352,
+      "step": 98500
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.0857906333505228e-05,
+      "loss": 0.0757,
+      "step": 98505
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.0857437532230088e-05,
+      "loss": 0.0837,
+      "step": 98510
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.085696873095495e-05,
+      "loss": 0.1346,
+      "step": 98515
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.085649992967981e-05,
+      "loss": 0.1277,
+      "step": 98520
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.085603112840467e-05,
+      "loss": 0.1872,
+      "step": 98525
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.085556232712953e-05,
+      "loss": 0.0699,
+      "step": 98530
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.0855093525854393e-05,
+      "loss": 0.0043,
+      "step": 98535
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.0854624724579253e-05,
+      "loss": 0.0165,
+      "step": 98540
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.0854155923304112e-05,
+      "loss": 0.0389,
+      "step": 98545
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.0853687122028972e-05,
+      "loss": 0.0459,
+      "step": 98550
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.0853218320753834e-05,
+      "loss": 0.0789,
+      "step": 98555
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.0852749519478694e-05,
+      "loss": 0.0961,
+      "step": 98560
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.0852280718203554e-05,
+      "loss": 0.0978,
+      "step": 98565
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.0851811916928414e-05,
+      "loss": 0.1127,
+      "step": 98570
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.0851343115653274e-05,
+      "loss": 0.2313,
+      "step": 98575
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.0850874314378137e-05,
+      "loss": 0.0617,
+      "step": 98580
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.0850405513102997e-05,
+      "loss": 0.0073,
+      "step": 98585
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.0849936711827857e-05,
+      "loss": 0.0523,
+      "step": 98590
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.0849467910552719e-05,
+      "loss": 0.0491,
+      "step": 98595
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.0848999109277578e-05,
+      "loss": 0.0915,
+      "step": 98600
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.0848530308002438e-05,
+      "loss": 0.1004,
+      "step": 98605
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.0848061506727298e-05,
+      "loss": 0.0599,
+      "step": 98610
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.0847592705452158e-05,
+      "loss": 0.1543,
+      "step": 98615
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.084712390417702e-05,
+      "loss": 0.1006,
+      "step": 98620
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.0846655102901882e-05,
+      "loss": 0.3397,
+      "step": 98625
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.0846186301626741e-05,
+      "loss": 0.0878,
+      "step": 98630
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.0845717500351603e-05,
+      "loss": 0.0588,
+      "step": 98635
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.0845248699076463e-05,
+      "loss": 0.0212,
+      "step": 98640
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.0844779897801323e-05,
+      "loss": 0.0454,
+      "step": 98645
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.0844311096526183e-05,
+      "loss": 0.0405,
+      "step": 98650
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.0843842295251043e-05,
+      "loss": 0.0378,
+      "step": 98655
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.0843373493975904e-05,
+      "loss": 0.1456,
+      "step": 98660
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.0842904692700764e-05,
+      "loss": 0.1298,
+      "step": 98665
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.0842435891425624e-05,
+      "loss": 0.1485,
+      "step": 98670
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.0841967090150488e-05,
+      "loss": 0.302,
+      "step": 98675
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.0841498288875348e-05,
+      "loss": 0.0655,
+      "step": 98680
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.0841029487600208e-05,
+      "loss": 0.0255,
+      "step": 98685
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.0840560686325067e-05,
+      "loss": 0.0089,
+      "step": 98690
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.0840091885049927e-05,
+      "loss": 0.0543,
+      "step": 98695
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.0839623083774789e-05,
+      "loss": 0.0458,
+      "step": 98700
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.0839154282499649e-05,
+      "loss": 0.039,
+      "step": 98705
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.0838685481224509e-05,
+      "loss": 0.0698,
+      "step": 98710
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.0838216679949369e-05,
+      "loss": 0.1055,
+      "step": 98715
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.0837747878674232e-05,
+      "loss": 0.3573,
+      "step": 98720
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.0837279077399092e-05,
+      "loss": 0.3102,
+      "step": 98725
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.0836810276123952e-05,
+      "loss": 0.0689,
+      "step": 98730
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.0836341474848812e-05,
+      "loss": 0.0161,
+      "step": 98735
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.0835872673573674e-05,
+      "loss": 0.0524,
+      "step": 98740
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.0835403872298533e-05,
+      "loss": 0.0708,
+      "step": 98745
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.0834935071023393e-05,
+      "loss": 0.0438,
+      "step": 98750
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.0834466269748253e-05,
+      "loss": 0.1151,
+      "step": 98755
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.0833997468473115e-05,
+      "loss": 0.0805,
+      "step": 98760
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.0833528667197977e-05,
+      "loss": 0.1929,
+      "step": 98765
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.0833059865922837e-05,
+      "loss": 0.1317,
+      "step": 98770
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.0832591064647696e-05,
+      "loss": 0.2985,
+      "step": 98775
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.0832122263372558e-05,
+      "loss": 0.0765,
+      "step": 98780
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.0831653462097418e-05,
+      "loss": 0.0017,
+      "step": 98785
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.0831184660822278e-05,
+      "loss": 0.0453,
+      "step": 98790
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.0830715859547138e-05,
+      "loss": 0.037,
+      "step": 98795
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.0830247058272e-05,
+      "loss": 0.0671,
+      "step": 98800
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.082977825699686e-05,
+      "loss": 0.057,
+      "step": 98805
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.0829309455721721e-05,
+      "loss": 0.1384,
+      "step": 98810
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.0828840654446581e-05,
+      "loss": 0.1596,
+      "step": 98815
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.0828371853171443e-05,
+      "loss": 0.286,
+      "step": 98820
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.0827903051896303e-05,
+      "loss": 0.2032,
+      "step": 98825
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.0827434250621163e-05,
+      "loss": 0.0679,
+      "step": 98830
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.0826965449346022e-05,
+      "loss": 0.0172,
+      "step": 98835
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.0826496648070884e-05,
+      "loss": 0.035,
+      "step": 98840
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.0826027846795744e-05,
+      "loss": 0.049,
+      "step": 98845
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.0825559045520604e-05,
+      "loss": 0.0568,
+      "step": 98850
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.0825090244245464e-05,
+      "loss": 0.0966,
+      "step": 98855
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.0824621442970327e-05,
+      "loss": 0.0621,
+      "step": 98860
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.0824152641695187e-05,
+      "loss": 0.0731,
+      "step": 98865
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.0823683840420047e-05,
+      "loss": 0.1169,
+      "step": 98870
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.0823215039144907e-05,
+      "loss": 0.1449,
+      "step": 98875
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.0822746237869769e-05,
+      "loss": 0.0552,
+      "step": 98880
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.0822277436594629e-05,
+      "loss": 0.0256,
+      "step": 98885
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.0821808635319489e-05,
+      "loss": 0.0353,
+      "step": 98890
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.0821339834044348e-05,
+      "loss": 0.0843,
+      "step": 98895
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.0820871032769208e-05,
+      "loss": 0.0545,
+      "step": 98900
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.0820402231494072e-05,
+      "loss": 0.0478,
+      "step": 98905
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.0819933430218932e-05,
+      "loss": 0.033,
+      "step": 98910
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.0819464628943792e-05,
+      "loss": 0.0588,
+      "step": 98915
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.0818995827668653e-05,
+      "loss": 0.1776,
+      "step": 98920
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.0818527026393513e-05,
+      "loss": 0.3594,
+      "step": 98925
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.0818058225118373e-05,
+      "loss": 0.0635,
+      "step": 98930
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.0817589423843233e-05,
+      "loss": 0.0541,
+      "step": 98935
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.0817120622568093e-05,
+      "loss": 0.0135,
+      "step": 98940
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.0816651821292955e-05,
+      "loss": 0.044,
+      "step": 98945
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.0816183020017816e-05,
+      "loss": 0.039,
+      "step": 98950
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.0815714218742676e-05,
+      "loss": 0.0487,
+      "step": 98955
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.0815245417467538e-05,
+      "loss": 0.0814,
+      "step": 98960
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.0814776616192398e-05,
+      "loss": 0.1315,
+      "step": 98965
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.0814307814917258e-05,
+      "loss": 0.1204,
+      "step": 98970
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.0813839013642118e-05,
+      "loss": 0.2536,
+      "step": 98975
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.0813370212366977e-05,
+      "loss": 0.0815,
+      "step": 98980
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.0812901411091839e-05,
+      "loss": 0.0141,
+      "step": 98985
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.0812432609816699e-05,
+      "loss": 0.0449,
+      "step": 98990
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.0811963808541559e-05,
+      "loss": 0.0351,
+      "step": 98995
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.0811495007266422e-05,
+      "loss": 0.0248,
+      "step": 99000
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.0811026205991282e-05,
+      "loss": 0.0664,
+      "step": 99005
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.0810557404716142e-05,
+      "loss": 0.0684,
+      "step": 99010
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.0810088603441002e-05,
+      "loss": 0.0888,
+      "step": 99015
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.0809619802165862e-05,
+      "loss": 0.1889,
+      "step": 99020
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.0809151000890724e-05,
+      "loss": 0.1735,
+      "step": 99025
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.0808682199615584e-05,
+      "loss": 0.0884,
+      "step": 99030
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.0808213398340444e-05,
+      "loss": 0.0138,
+      "step": 99035
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.0807744597065303e-05,
+      "loss": 0.039,
+      "step": 99040
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.0807275795790167e-05,
+      "loss": 0.0216,
+      "step": 99045
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.0806806994515027e-05,
+      "loss": 0.0492,
+      "step": 99050
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.0806338193239887e-05,
+      "loss": 0.1089,
+      "step": 99055
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.0805869391964747e-05,
+      "loss": 0.0727,
+      "step": 99060
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.0805400590689608e-05,
+      "loss": 0.0978,
+      "step": 99065
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.0804931789414468e-05,
+      "loss": 0.137,
+      "step": 99070
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.0804462988139328e-05,
+      "loss": 0.2223,
+      "step": 99075
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.0803994186864188e-05,
+      "loss": 0.0674,
+      "step": 99080
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.0803525385589048e-05,
+      "loss": 0.0336,
+      "step": 99085
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.0803056584313911e-05,
+      "loss": 0.0316,
+      "step": 99090
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.0802587783038771e-05,
+      "loss": 0.0132,
+      "step": 99095
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.0802118981763631e-05,
+      "loss": 0.0319,
+      "step": 99100
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.0801650180488493e-05,
+      "loss": 0.0532,
+      "step": 99105
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.0801181379213353e-05,
+      "loss": 0.0389,
+      "step": 99110
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.0800712577938213e-05,
+      "loss": 0.0504,
+      "step": 99115
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.0800243776663073e-05,
+      "loss": 0.2022,
+      "step": 99120
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.0799774975387932e-05,
+      "loss": 0.2379,
+      "step": 99125
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.0799306174112794e-05,
+      "loss": 0.078,
+      "step": 99130
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.0798837372837656e-05,
+      "loss": 0.0242,
+      "step": 99135
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.0798368571562516e-05,
+      "loss": 0.0157,
+      "step": 99140
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.0797899770287377e-05,
+      "loss": 0.0853,
+      "step": 99145
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.0797430969012237e-05,
+      "loss": 0.0341,
+      "step": 99150
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.0796962167737097e-05,
+      "loss": 0.0882,
+      "step": 99155
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.0796493366461957e-05,
+      "loss": 0.103,
+      "step": 99160
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.0796024565186817e-05,
+      "loss": 0.0606,
+      "step": 99165
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.0795555763911679e-05,
+      "loss": 0.1645,
+      "step": 99170
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.0795086962636539e-05,
+      "loss": 0.1672,
+      "step": 99175
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.0794618161361399e-05,
+      "loss": 0.092,
+      "step": 99180
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.0794149360086262e-05,
+      "loss": 0.0337,
+      "step": 99185
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.0793680558811122e-05,
+      "loss": 0.0625,
+      "step": 99190
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.0793211757535982e-05,
+      "loss": 0.0462,
+      "step": 99195
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.0792742956260842e-05,
+      "loss": 0.0298,
+      "step": 99200
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.0792274154985702e-05,
+      "loss": 0.0773,
+      "step": 99205
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.0791805353710563e-05,
+      "loss": 0.1238,
+      "step": 99210
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.0791336552435423e-05,
+      "loss": 0.0865,
+      "step": 99215
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.0790867751160283e-05,
+      "loss": 0.1429,
+      "step": 99220
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.0790398949885143e-05,
+      "loss": 0.2869,
+      "step": 99225
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.0789930148610006e-05,
+      "loss": 0.067,
+      "step": 99230
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.0789461347334866e-05,
+      "loss": 0.0572,
+      "step": 99235
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.0788992546059726e-05,
+      "loss": 0.0255,
+      "step": 99240
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.0788523744784586e-05,
+      "loss": 0.0248,
+      "step": 99245
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.0788054943509448e-05,
+      "loss": 0.0834,
+      "step": 99250
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.0787586142234308e-05,
+      "loss": 0.0635,
+      "step": 99255
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.0787117340959168e-05,
+      "loss": 0.0725,
+      "step": 99260
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.0786648539684028e-05,
+      "loss": 0.1468,
+      "step": 99265
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.078617973840889e-05,
+      "loss": 0.1187,
+      "step": 99270
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.078571093713375e-05,
+      "loss": 0.2333,
+      "step": 99275
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.078524213585861e-05,
+      "loss": 0.1005,
+      "step": 99280
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.078477333458347e-05,
+      "loss": 0.0119,
+      "step": 99285
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.0784304533308332e-05,
+      "loss": 0.0043,
+      "step": 99290
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.0783835732033192e-05,
+      "loss": 0.038,
+      "step": 99295
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.0783366930758052e-05,
+      "loss": 0.071,
+      "step": 99300
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.0782898129482912e-05,
+      "loss": 0.0802,
+      "step": 99305
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.0782429328207774e-05,
+      "loss": 0.0974,
+      "step": 99310
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.0781960526932634e-05,
+      "loss": 0.1049,
+      "step": 99315
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.0781491725657494e-05,
+      "loss": 0.2026,
+      "step": 99320
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.0781022924382355e-05,
+      "loss": 0.3167,
+      "step": 99325
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.0780554123107217e-05,
+      "loss": 0.1143,
+      "step": 99330
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.0780085321832077e-05,
+      "loss": 0.0335,
+      "step": 99335
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.0779616520556937e-05,
+      "loss": 0.0593,
+      "step": 99340
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.0779147719281797e-05,
+      "loss": 0.0417,
+      "step": 99345
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.0778678918006658e-05,
+      "loss": 0.0406,
+      "step": 99350
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.0778210116731518e-05,
+      "loss": 0.0409,
+      "step": 99355
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.0777741315456378e-05,
+      "loss": 0.0867,
+      "step": 99360
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.0777272514181238e-05,
+      "loss": 0.1873,
+      "step": 99365
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.0776803712906101e-05,
+      "loss": 0.0758,
+      "step": 99370
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.0776334911630961e-05,
+      "loss": 0.2972,
+      "step": 99375
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.0775866110355821e-05,
+      "loss": 0.0862,
+      "step": 99380
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.0775397309080681e-05,
+      "loss": 0.0447,
+      "step": 99385
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.0774928507805543e-05,
+      "loss": 0.02,
+      "step": 99390
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.0774459706530403e-05,
+      "loss": 0.016,
+      "step": 99395
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.0773990905255263e-05,
+      "loss": 0.0336,
+      "step": 99400
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.0773522103980123e-05,
+      "loss": 0.0406,
+      "step": 99405
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.0773053302704983e-05,
+      "loss": 0.1362,
+      "step": 99410
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.0772584501429846e-05,
+      "loss": 0.1222,
+      "step": 99415
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.0772115700154706e-05,
+      "loss": 0.0985,
+      "step": 99420
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.0771646898879566e-05,
+      "loss": 0.1979,
+      "step": 99425
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.0771178097604427e-05,
+      "loss": 0.1238,
+      "step": 99430
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.0770709296329287e-05,
+      "loss": 0.0166,
+      "step": 99435
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.0770240495054147e-05,
+      "loss": 0.0284,
+      "step": 99440
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.0769771693779007e-05,
+      "loss": 0.0267,
+      "step": 99445
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.0769302892503867e-05,
+      "loss": 0.0786,
+      "step": 99450
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.0768834091228729e-05,
+      "loss": 0.0815,
+      "step": 99455
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.076836528995359e-05,
+      "loss": 0.1014,
+      "step": 99460
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.076789648867845e-05,
+      "loss": 0.075,
+      "step": 99465
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.0767427687403312e-05,
+      "loss": 0.1781,
+      "step": 99470
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.0766958886128172e-05,
+      "loss": 0.2461,
+      "step": 99475
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.0766490084853032e-05,
+      "loss": 0.0696,
+      "step": 99480
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.0766021283577892e-05,
+      "loss": 0.0542,
+      "step": 99485
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.0765552482302752e-05,
+      "loss": 0.0734,
+      "step": 99490
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.0765083681027613e-05,
+      "loss": 0.0164,
+      "step": 99495
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.0764614879752473e-05,
+      "loss": 0.0494,
+      "step": 99500
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.0764146078477333e-05,
+      "loss": 0.1086,
+      "step": 99505
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.0763677277202197e-05,
+      "loss": 0.0644,
+      "step": 99510
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.0763208475927056e-05,
+      "loss": 0.1442,
+      "step": 99515
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.0762739674651916e-05,
+      "loss": 0.178,
+      "step": 99520
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.0762270873376776e-05,
+      "loss": 0.2618,
+      "step": 99525
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.0761802072101636e-05,
+      "loss": 0.0942,
+      "step": 99530
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.0761333270826498e-05,
+      "loss": 0.0282,
+      "step": 99535
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.0760864469551358e-05,
+      "loss": 0.0709,
+      "step": 99540
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.0760395668276218e-05,
+      "loss": 0.0218,
+      "step": 99545
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.0759926867001078e-05,
+      "loss": 0.0258,
+      "step": 99550
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.0759458065725941e-05,
+      "loss": 0.0801,
+      "step": 99555
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.0758989264450801e-05,
+      "loss": 0.0565,
+      "step": 99560
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.0758520463175661e-05,
+      "loss": 0.1175,
+      "step": 99565
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.075805166190052e-05,
+      "loss": 0.0922,
+      "step": 99570
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.0757582860625382e-05,
+      "loss": 0.3132,
+      "step": 99575
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.0757114059350242e-05,
+      "loss": 0.0812,
+      "step": 99580
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.0756645258075102e-05,
+      "loss": 0.0387,
+      "step": 99585
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.0756176456799962e-05,
+      "loss": 0.013,
+      "step": 99590
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.0755707655524822e-05,
+      "loss": 0.0419,
+      "step": 99595
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.0755238854249685e-05,
+      "loss": 0.0777,
+      "step": 99600
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.0754770052974545e-05,
+      "loss": 0.0271,
+      "step": 99605
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.0754301251699405e-05,
+      "loss": 0.0288,
+      "step": 99610
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.0753832450424267e-05,
+      "loss": 0.1725,
+      "step": 99615
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.0753363649149127e-05,
+      "loss": 0.1753,
+      "step": 99620
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.0752894847873987e-05,
+      "loss": 0.272,
+      "step": 99625
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.0752426046598847e-05,
+      "loss": 0.0909,
+      "step": 99630
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.0751957245323707e-05,
+      "loss": 0.0129,
+      "step": 99635
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.0751488444048568e-05,
+      "loss": 0.0199,
+      "step": 99640
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.0751019642773428e-05,
+      "loss": 0.0234,
+      "step": 99645
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.075055084149829e-05,
+      "loss": 0.0879,
+      "step": 99650
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.0750082040223152e-05,
+      "loss": 0.0563,
+      "step": 99655
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.0749613238948011e-05,
+      "loss": 0.0947,
+      "step": 99660
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.0749144437672871e-05,
+      "loss": 0.1726,
+      "step": 99665
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.0748675636397731e-05,
+      "loss": 0.1608,
+      "step": 99670
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.0748206835122591e-05,
+      "loss": 0.2688,
+      "step": 99675
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.0747738033847453e-05,
+      "loss": 0.0452,
+      "step": 99680
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.0747269232572313e-05,
+      "loss": 0.011,
+      "step": 99685
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.0746800431297173e-05,
+      "loss": 0.0227,
+      "step": 99690
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.0746331630022036e-05,
+      "loss": 0.0385,
+      "step": 99695
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.0745862828746896e-05,
+      "loss": 0.089,
+      "step": 99700
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.0745394027471756e-05,
+      "loss": 0.08,
+      "step": 99705
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.0744925226196616e-05,
+      "loss": 0.0665,
+      "step": 99710
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.0744456424921478e-05,
+      "loss": 0.2098,
+      "step": 99715
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.0743987623646337e-05,
+      "loss": 0.1671,
+      "step": 99720
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.0743518822371197e-05,
+      "loss": 0.2883,
+      "step": 99725
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.0743050021096057e-05,
+      "loss": 0.0748,
+      "step": 99730
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.0742581219820917e-05,
+      "loss": 0.0249,
+      "step": 99735
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.074211241854578e-05,
+      "loss": 0.0332,
+      "step": 99740
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.074164361727064e-05,
+      "loss": 0.0898,
+      "step": 99745
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.07411748159955e-05,
+      "loss": 0.0587,
+      "step": 99750
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.0740706014720362e-05,
+      "loss": 0.0309,
+      "step": 99755
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.0740237213445222e-05,
+      "loss": 0.0881,
+      "step": 99760
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.0739768412170082e-05,
+      "loss": 0.1376,
+      "step": 99765
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.0739299610894942e-05,
+      "loss": 0.2707,
+      "step": 99770
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.0738830809619802e-05,
+      "loss": 0.2035,
+      "step": 99775
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.0738362008344663e-05,
+      "loss": 0.0786,
+      "step": 99780
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.0737893207069525e-05,
+      "loss": 0.0299,
+      "step": 99785
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.0737424405794385e-05,
+      "loss": 0.0207,
+      "step": 99790
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.0736955604519247e-05,
+      "loss": 0.0134,
+      "step": 99795
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.0736486803244107e-05,
+      "loss": 0.0427,
+      "step": 99800
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.0736018001968966e-05,
+      "loss": 0.0398,
+      "step": 99805
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.0735549200693826e-05,
+      "loss": 0.1027,
+      "step": 99810
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.0735080399418686e-05,
+      "loss": 0.0696,
+      "step": 99815
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.0734611598143548e-05,
+      "loss": 0.1245,
+      "step": 99820
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.0734142796868408e-05,
+      "loss": 0.2651,
+      "step": 99825
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.0733673995593268e-05,
+      "loss": 0.0804,
+      "step": 99830
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.0733205194318131e-05,
+      "loss": 0.008,
+      "step": 99835
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.0732736393042991e-05,
+      "loss": 0.0325,
+      "step": 99840
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.0732267591767851e-05,
+      "loss": 0.0206,
+      "step": 99845
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.0731798790492711e-05,
+      "loss": 0.0482,
+      "step": 99850
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.0731329989217571e-05,
+      "loss": 0.06,
+      "step": 99855
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.0730861187942433e-05,
+      "loss": 0.1655,
+      "step": 99860
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.0730392386667292e-05,
+      "loss": 0.0665,
+      "step": 99865
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.0729923585392152e-05,
+      "loss": 0.1627,
+      "step": 99870
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.0729454784117012e-05,
+      "loss": 0.1678,
+      "step": 99875
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.0728985982841876e-05,
+      "loss": 0.0892,
+      "step": 99880
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.0728517181566736e-05,
+      "loss": 0.0208,
+      "step": 99885
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.0728048380291596e-05,
+      "loss": 0.0626,
+      "step": 99890
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.0727579579016455e-05,
+      "loss": 0.0392,
+      "step": 99895
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.0727110777741317e-05,
+      "loss": 0.0458,
+      "step": 99900
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.0726641976466177e-05,
+      "loss": 0.0586,
+      "step": 99905
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.0726173175191037e-05,
+      "loss": 0.0368,
+      "step": 99910
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.0725704373915897e-05,
+      "loss": 0.1235,
+      "step": 99915
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.0725235572640757e-05,
+      "loss": 0.1358,
+      "step": 99920
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.072476677136562e-05,
+      "loss": 0.2023,
+      "step": 99925
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.072429797009048e-05,
+      "loss": 0.0468,
+      "step": 99930
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.072382916881534e-05,
+      "loss": 0.018,
+      "step": 99935
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.0723360367540202e-05,
+      "loss": 0.0191,
+      "step": 99940
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.0722891566265062e-05,
+      "loss": 0.038,
+      "step": 99945
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.0722422764989922e-05,
+      "loss": 0.0758,
+      "step": 99950
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.0721953963714781e-05,
+      "loss": 0.0867,
+      "step": 99955
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.0721485162439641e-05,
+      "loss": 0.0558,
+      "step": 99960
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.0721016361164503e-05,
+      "loss": 0.1859,
+      "step": 99965
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.0720547559889363e-05,
+      "loss": 0.1605,
+      "step": 99970
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.0720078758614225e-05,
+      "loss": 0.1372,
+      "step": 99975
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.0719609957339086e-05,
+      "loss": 0.1288,
+      "step": 99980
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.0719141156063946e-05,
+      "loss": 0.0231,
+      "step": 99985
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.0718672354788806e-05,
+      "loss": 0.0178,
+      "step": 99990
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.0718203553513666e-05,
+      "loss": 0.0467,
+      "step": 99995
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.0717734752238526e-05,
+      "loss": 0.0779,
+      "step": 100000
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.0717265950963388e-05,
+      "loss": 0.0453,
+      "step": 100005
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.0716797149688247e-05,
+      "loss": 0.0382,
+      "step": 100010
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.0716328348413107e-05,
+      "loss": 0.1943,
+      "step": 100015
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.071585954713797e-05,
+      "loss": 0.1339,
+      "step": 100020
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.071539074586283e-05,
+      "loss": 0.2929,
+      "step": 100025
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.071492194458769e-05,
+      "loss": 0.0649,
+      "step": 100030
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.071445314331255e-05,
+      "loss": 0.0202,
+      "step": 100035
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.071398434203741e-05,
+      "loss": 0.0123,
+      "step": 100040
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.0713515540762272e-05,
+      "loss": 0.0731,
+      "step": 100045
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.0713046739487132e-05,
+      "loss": 0.0526,
+      "step": 100050
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.0712577938211992e-05,
+      "loss": 0.0832,
+      "step": 100055
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.0712109136936852e-05,
+      "loss": 0.0807,
+      "step": 100060
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.0711640335661715e-05,
+      "loss": 0.0992,
+      "step": 100065
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.0711171534386575e-05,
+      "loss": 0.2506,
+      "step": 100070
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.0710702733111435e-05,
+      "loss": 0.2901,
+      "step": 100075
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.0710233931836295e-05,
+      "loss": 0.1019,
+      "step": 100080
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.0709765130561157e-05,
+      "loss": 0.0498,
+      "step": 100085
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.0709296329286017e-05,
+      "loss": 0.0206,
+      "step": 100090
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.0708827528010877e-05,
+      "loss": 0.0778,
+      "step": 100095
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.0708358726735736e-05,
+      "loss": 0.0532,
+      "step": 100100
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.0707889925460598e-05,
+      "loss": 0.1336,
+      "step": 100105
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.070742112418546e-05,
+      "loss": 0.0676,
+      "step": 100110
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.070695232291032e-05,
+      "loss": 0.1286,
+      "step": 100115
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.070648352163518e-05,
+      "loss": 0.2123,
+      "step": 100120
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.0706014720360041e-05,
+      "loss": 0.315,
+      "step": 100125
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.0705545919084901e-05,
+      "loss": 0.0815,
+      "step": 100130
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.0705077117809761e-05,
+      "loss": 0.0161,
+      "step": 100135
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.0704608316534621e-05,
+      "loss": 0.0348,
+      "step": 100140
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.0704139515259483e-05,
+      "loss": 0.0347,
+      "step": 100145
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.0703670713984343e-05,
+      "loss": 0.0313,
+      "step": 100150
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.0703201912709202e-05,
+      "loss": 0.0339,
+      "step": 100155
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.0702733111434064e-05,
+      "loss": 0.1409,
+      "step": 100160
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.0702264310158926e-05,
+      "loss": 0.144,
+      "step": 100165
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.0701795508883786e-05,
+      "loss": 0.1255,
+      "step": 100170
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.0701326707608646e-05,
+      "loss": 0.2538,
+      "step": 100175
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.0700857906333506e-05,
+      "loss": 0.0603,
+      "step": 100180
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.0700389105058367e-05,
+      "loss": 0.0213,
+      "step": 100185
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.0699920303783227e-05,
+      "loss": 0.0201,
+      "step": 100190
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.0699451502508087e-05,
+      "loss": 0.0163,
+      "step": 100195
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.0698982701232947e-05,
+      "loss": 0.0369,
+      "step": 100200
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.069851389995781e-05,
+      "loss": 0.053,
+      "step": 100205
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.069804509868267e-05,
+      "loss": 0.1032,
+      "step": 100210
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.069757629740753e-05,
+      "loss": 0.1269,
+      "step": 100215
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.069710749613239e-05,
+      "loss": 0.1276,
+      "step": 100220
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.0696638694857252e-05,
+      "loss": 0.2841,
+      "step": 100225
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.0696169893582112e-05,
+      "loss": 0.0599,
+      "step": 100230
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.0695701092306972e-05,
+      "loss": 0.0166,
+      "step": 100235
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.0695232291031832e-05,
+      "loss": 0.0124,
+      "step": 100240
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.0694763489756691e-05,
+      "loss": 0.0527,
+      "step": 100245
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.0694294688481555e-05,
+      "loss": 0.0974,
+      "step": 100250
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.0693825887206415e-05,
+      "loss": 0.0282,
+      "step": 100255
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.0693357085931275e-05,
+      "loss": 0.071,
+      "step": 100260
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.0692888284656136e-05,
+      "loss": 0.0878,
+      "step": 100265
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.0692419483380996e-05,
+      "loss": 0.1519,
+      "step": 100270
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.0691950682105856e-05,
+      "loss": 0.1982,
+      "step": 100275
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.0691481880830716e-05,
+      "loss": 0.0716,
+      "step": 100280
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.0691013079555576e-05,
+      "loss": 0.0221,
+      "step": 100285
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.0690544278280438e-05,
+      "loss": 0.0543,
+      "step": 100290
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.0690075477005298e-05,
+      "loss": 0.0433,
+      "step": 100295
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.068960667573016e-05,
+      "loss": 0.0965,
+      "step": 100300
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.068913787445502e-05,
+      "loss": 0.033,
+      "step": 100305
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.068866907317988e-05,
+      "loss": 0.0709,
+      "step": 100310
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.068820027190474e-05,
+      "loss": 0.1506,
+      "step": 100315
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.06877314706296e-05,
+      "loss": 0.081,
+      "step": 100320
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.068726266935446e-05,
+      "loss": 0.2956,
+      "step": 100325
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.0686793868079322e-05,
+      "loss": 0.0425,
+      "step": 100330
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.0686325066804182e-05,
+      "loss": 0.0386,
+      "step": 100335
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.0685856265529042e-05,
+      "loss": 0.0421,
+      "step": 100340
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.0685387464253905e-05,
+      "loss": 0.0551,
+      "step": 100345
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.0684918662978765e-05,
+      "loss": 0.0615,
+      "step": 100350
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.0684449861703625e-05,
+      "loss": 0.1055,
+      "step": 100355
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.0683981060428485e-05,
+      "loss": 0.0648,
+      "step": 100360
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.0683512259153345e-05,
+      "loss": 0.1031,
+      "step": 100365
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.0683043457878207e-05,
+      "loss": 0.2118,
+      "step": 100370
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.0682574656603067e-05,
+      "loss": 0.2273,
+      "step": 100375
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.0682105855327927e-05,
+      "loss": 0.0594,
+      "step": 100380
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.0681637054052787e-05,
+      "loss": 0.0534,
+      "step": 100385
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.068116825277765e-05,
+      "loss": 0.0265,
+      "step": 100390
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.068069945150251e-05,
+      "loss": 0.0198,
+      "step": 100395
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.068023065022737e-05,
+      "loss": 0.0553,
+      "step": 100400
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.067976184895223e-05,
+      "loss": 0.0506,
+      "step": 100405
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.0679293047677091e-05,
+      "loss": 0.0465,
+      "step": 100410
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.0678824246401951e-05,
+      "loss": 0.0542,
+      "step": 100415
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.0678355445126811e-05,
+      "loss": 0.0767,
+      "step": 100420
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.0677886643851671e-05,
+      "loss": 0.4844,
+      "step": 100425
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.0677417842576531e-05,
+      "loss": 0.0618,
+      "step": 100430
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.0676949041301394e-05,
+      "loss": 0.0378,
+      "step": 100435
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.0676480240026254e-05,
+      "loss": 0.0389,
+      "step": 100440
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.0676011438751114e-05,
+      "loss": 0.0847,
+      "step": 100445
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.0675542637475976e-05,
+      "loss": 0.1025,
+      "step": 100450
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.0675073836200836e-05,
+      "loss": 0.0869,
+      "step": 100455
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.0674605034925696e-05,
+      "loss": 0.137,
+      "step": 100460
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.0674136233650556e-05,
+      "loss": 0.1748,
+      "step": 100465
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.0673667432375416e-05,
+      "loss": 0.2259,
+      "step": 100470
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.0673198631100277e-05,
+      "loss": 0.3033,
+      "step": 100475
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.0672729829825137e-05,
+      "loss": 0.0709,
+      "step": 100480
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.0672261028549999e-05,
+      "loss": 0.0298,
+      "step": 100485
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.067179222727486e-05,
+      "loss": 0.0149,
+      "step": 100490
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.067132342599972e-05,
+      "loss": 0.0639,
+      "step": 100495
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.067085462472458e-05,
+      "loss": 0.0498,
+      "step": 100500
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.067038582344944e-05,
+      "loss": 0.0721,
+      "step": 100505
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.06699170221743e-05,
+      "loss": 0.1122,
+      "step": 100510
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.0669448220899162e-05,
+      "loss": 0.1297,
+      "step": 100515
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.0668979419624022e-05,
+      "loss": 0.2213,
+      "step": 100520
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.0668510618348882e-05,
+      "loss": 0.213,
+      "step": 100525
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.0668041817073745e-05,
+      "loss": 0.0283,
+      "step": 100530
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.0667573015798605e-05,
+      "loss": 0.0678,
+      "step": 100535
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.0667104214523465e-05,
+      "loss": 0.0633,
+      "step": 100540
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.0666635413248325e-05,
+      "loss": 0.0746,
+      "step": 100545
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.0666166611973185e-05,
+      "loss": 0.0363,
+      "step": 100550
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.0665697810698046e-05,
+      "loss": 0.1217,
+      "step": 100555
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.0665229009422906e-05,
+      "loss": 0.1239,
+      "step": 100560
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.0664760208147766e-05,
+      "loss": 0.0527,
+      "step": 100565
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.0664291406872626e-05,
+      "loss": 0.084,
+      "step": 100570
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.066382260559749e-05,
+      "loss": 0.2952,
+      "step": 100575
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.066335380432235e-05,
+      "loss": 0.076,
+      "step": 100580
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.066288500304721e-05,
+      "loss": 0.0167,
+      "step": 100585
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.066241620177207e-05,
+      "loss": 0.0179,
+      "step": 100590
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.0661947400496931e-05,
+      "loss": 0.006,
+      "step": 100595
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.066147859922179e-05,
+      "loss": 0.0712,
+      "step": 100600
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.066100979794665e-05,
+      "loss": 0.0385,
+      "step": 100605
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.066054099667151e-05,
+      "loss": 0.1344,
+      "step": 100610
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.0660072195396372e-05,
+      "loss": 0.1812,
+      "step": 100615
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.0659603394121232e-05,
+      "loss": 0.1248,
+      "step": 100620
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.0659134592846094e-05,
+      "loss": 0.2997,
+      "step": 100625
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.0658665791570954e-05,
+      "loss": 0.1126,
+      "step": 100630
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.0658196990295815e-05,
+      "loss": 0.0013,
+      "step": 100635
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.0657728189020675e-05,
+      "loss": 0.0172,
+      "step": 100640
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.0657259387745535e-05,
+      "loss": 0.0729,
+      "step": 100645
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.0656790586470395e-05,
+      "loss": 0.0848,
+      "step": 100650
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.0656321785195257e-05,
+      "loss": 0.0482,
+      "step": 100655
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.0655852983920117e-05,
+      "loss": 0.08,
+      "step": 100660
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.0655384182644977e-05,
+      "loss": 0.0696,
+      "step": 100665
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.0654915381369838e-05,
+      "loss": 0.1357,
+      "step": 100670
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.06544465800947e-05,
+      "loss": 0.2,
+      "step": 100675
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.065397777881956e-05,
+      "loss": 0.1173,
+      "step": 100680
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.065350897754442e-05,
+      "loss": 0.0499,
+      "step": 100685
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.065304017626928e-05,
+      "loss": 0.0274,
+      "step": 100690
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.0652571374994141e-05,
+      "loss": 0.0225,
+      "step": 100695
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.0652102573719001e-05,
+      "loss": 0.0368,
+      "step": 100700
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.0651633772443861e-05,
+      "loss": 0.039,
+      "step": 100705
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.0651164971168721e-05,
+      "loss": 0.0947,
+      "step": 100710
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.0650696169893585e-05,
+      "loss": 0.1109,
+      "step": 100715
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.0650227368618444e-05,
+      "loss": 0.1628,
+      "step": 100720
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.0649758567343304e-05,
+      "loss": 0.2349,
+      "step": 100725
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.0649289766068164e-05,
+      "loss": 0.0657,
+      "step": 100730
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.0648820964793026e-05,
+      "loss": 0.0673,
+      "step": 100735
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.0648352163517886e-05,
+      "loss": 0.0122,
+      "step": 100740
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.0647883362242746e-05,
+      "loss": 0.0133,
+      "step": 100745
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.0647414560967606e-05,
+      "loss": 0.1202,
+      "step": 100750
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.0646945759692466e-05,
+      "loss": 0.0651,
+      "step": 100755
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.0646476958417329e-05,
+      "loss": 0.0506,
+      "step": 100760
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.0646008157142189e-05,
+      "loss": 0.1803,
+      "step": 100765
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.0645539355867049e-05,
+      "loss": 0.1909,
+      "step": 100770
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.064507055459191e-05,
+      "loss": 0.4379,
+      "step": 100775
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.064460175331677e-05,
+      "loss": 0.0823,
+      "step": 100780
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.064413295204163e-05,
+      "loss": 0.016,
+      "step": 100785
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.064366415076649e-05,
+      "loss": 0.0294,
+      "step": 100790
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.064319534949135e-05,
+      "loss": 0.0207,
+      "step": 100795
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.0642726548216212e-05,
+      "loss": 0.0547,
+      "step": 100800
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.0642257746941072e-05,
+      "loss": 0.0596,
+      "step": 100805
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.0641788945665933e-05,
+      "loss": 0.038,
+      "step": 100810
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.0641320144390795e-05,
+      "loss": 0.0886,
+      "step": 100815
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.0640851343115655e-05,
+      "loss": 0.1261,
+      "step": 100820
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.0640382541840515e-05,
+      "loss": 0.2122,
+      "step": 100825
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.0639913740565375e-05,
+      "loss": 0.0946,
+      "step": 100830
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.0639444939290235e-05,
+      "loss": 0.0335,
+      "step": 100835
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.0638976138015096e-05,
+      "loss": 0.029,
+      "step": 100840
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.0638507336739956e-05,
+      "loss": 0.113,
+      "step": 100845
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.0638038535464816e-05,
+      "loss": 0.04,
+      "step": 100850
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.063756973418968e-05,
+      "loss": 0.0688,
+      "step": 100855
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.063710093291454e-05,
+      "loss": 0.1049,
+      "step": 100860
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.06366321316394e-05,
+      "loss": 0.1635,
+      "step": 100865
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.063616333036426e-05,
+      "loss": 0.1323,
+      "step": 100870
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.063569452908912e-05,
+      "loss": 0.1909,
+      "step": 100875
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.0635225727813981e-05,
+      "loss": 0.0875,
+      "step": 100880
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.0634756926538841e-05,
+      "loss": 0.0142,
+      "step": 100885
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.06342881252637e-05,
+      "loss": 0.019,
+      "step": 100890
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.063381932398856e-05,
+      "loss": 0.0893,
+      "step": 100895
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.0633350522713424e-05,
+      "loss": 0.0698,
+      "step": 100900
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.0632881721438284e-05,
+      "loss": 0.0431,
+      "step": 100905
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.0632412920163144e-05,
+      "loss": 0.1191,
+      "step": 100910
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.0631944118888004e-05,
+      "loss": 0.0925,
+      "step": 100915
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.0631475317612866e-05,
+      "loss": 0.1974,
+      "step": 100920
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.0631006516337725e-05,
+      "loss": 0.204,
+      "step": 100925
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.0630537715062585e-05,
+      "loss": 0.0802,
+      "step": 100930
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.0630068913787445e-05,
+      "loss": 0.0217,
+      "step": 100935
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.0629600112512305e-05,
+      "loss": 0.0483,
+      "step": 100940
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.0629131311237167e-05,
+      "loss": 0.042,
+      "step": 100945
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.0628662509962029e-05,
+      "loss": 0.0397,
+      "step": 100950
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.0628193708686888e-05,
+      "loss": 0.0709,
+      "step": 100955
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.062772490741175e-05,
+      "loss": 0.1507,
+      "step": 100960
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.062725610613661e-05,
+      "loss": 0.1002,
+      "step": 100965
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.062678730486147e-05,
+      "loss": 0.1177,
+      "step": 100970
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.062631850358633e-05,
+      "loss": 0.2481,
+      "step": 100975
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.062584970231119e-05,
+      "loss": 0.0972,
+      "step": 100980
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.0625380901036051e-05,
+      "loss": 0.0139,
+      "step": 100985
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.0624912099760911e-05,
+      "loss": 0.0208,
+      "step": 100990
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.0624443298485773e-05,
+      "loss": 0.0425,
+      "step": 100995
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.0623974497210635e-05,
+      "loss": 0.0514,
+      "step": 101000
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.0623505695935495e-05,
+      "loss": 0.0322,
+      "step": 101005
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.0623036894660354e-05,
+      "loss": 0.0461,
+      "step": 101010
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.0622568093385214e-05,
+      "loss": 0.093,
+      "step": 101015
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.0622099292110074e-05,
+      "loss": 0.1639,
+      "step": 101020
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.0621630490834936e-05,
+      "loss": 0.3203,
+      "step": 101025
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.0621161689559796e-05,
+      "loss": 0.1225,
+      "step": 101030
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.0620692888284656e-05,
+      "loss": 0.0396,
+      "step": 101035
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.062022408700952e-05,
+      "loss": 0.0563,
+      "step": 101040
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.0619755285734379e-05,
+      "loss": 0.0214,
+      "step": 101045
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.0619286484459239e-05,
+      "loss": 0.0621,
+      "step": 101050
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.0618817683184099e-05,
+      "loss": 0.0518,
+      "step": 101055
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.0618348881908959e-05,
+      "loss": 0.0619,
+      "step": 101060
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.061788008063382e-05,
+      "loss": 0.0415,
+      "step": 101065
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.061741127935868e-05,
+      "loss": 0.0964,
+      "step": 101070
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.061694247808354e-05,
+      "loss": 0.2579,
+      "step": 101075
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.06164736768084e-05,
+      "loss": 0.1298,
+      "step": 101080
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.0616004875533264e-05,
+      "loss": 0.007,
+      "step": 101085
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.0615536074258124e-05,
+      "loss": 0.0423,
+      "step": 101090
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.0615067272982984e-05,
+      "loss": 0.0619,
+      "step": 101095
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.0614598471707843e-05,
+      "loss": 0.031,
+      "step": 101100
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.0614129670432705e-05,
+      "loss": 0.0469,
+      "step": 101105
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.0613660869157565e-05,
+      "loss": 0.1384,
+      "step": 101110
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.0613192067882425e-05,
+      "loss": 0.0661,
+      "step": 101115
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.0612723266607285e-05,
+      "loss": 0.215,
+      "step": 101120
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.0612254465332147e-05,
+      "loss": 0.1955,
+      "step": 101125
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.0611785664057006e-05,
+      "loss": 0.0905,
+      "step": 101130
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.0611316862781868e-05,
+      "loss": 0.0382,
+      "step": 101135
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.061084806150673e-05,
+      "loss": 0.026,
+      "step": 101140
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.061037926023159e-05,
+      "loss": 0.0418,
+      "step": 101145
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.060991045895645e-05,
+      "loss": 0.032,
+      "step": 101150
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.060944165768131e-05,
+      "loss": 0.0463,
+      "step": 101155
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.060897285640617e-05,
+      "loss": 0.0731,
+      "step": 101160
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.0608504055131031e-05,
+      "loss": 0.1283,
+      "step": 101165
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.0608035253855891e-05,
+      "loss": 0.1381,
+      "step": 101170
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.0607566452580751e-05,
+      "loss": 0.3131,
+      "step": 101175
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.0607097651305614e-05,
+      "loss": 0.061,
+      "step": 101180
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.0606628850030474e-05,
+      "loss": 0.0771,
+      "step": 101185
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.0606160048755334e-05,
+      "loss": 0.0415,
+      "step": 101190
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.0605691247480194e-05,
+      "loss": 0.0172,
+      "step": 101195
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.0605222446205054e-05,
+      "loss": 0.029,
+      "step": 101200
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.0604753644929916e-05,
+      "loss": 0.0524,
+      "step": 101205
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.0604284843654776e-05,
+      "loss": 0.0929,
+      "step": 101210
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.0603816042379635e-05,
+      "loss": 0.1215,
+      "step": 101215
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.0603347241104495e-05,
+      "loss": 0.1959,
+      "step": 101220
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.0602878439829359e-05,
+      "loss": 0.4005,
+      "step": 101225
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.0602409638554219e-05,
+      "loss": 0.0942,
+      "step": 101230
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.0601940837279079e-05,
+      "loss": 0.0179,
+      "step": 101235
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.0601472036003939e-05,
+      "loss": 0.0323,
+      "step": 101240
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.06010032347288e-05,
+      "loss": 0.0408,
+      "step": 101245
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.060053443345366e-05,
+      "loss": 0.0497,
+      "step": 101250
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.060006563217852e-05,
+      "loss": 0.0602,
+      "step": 101255
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.059959683090338e-05,
+      "loss": 0.1121,
+      "step": 101260
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.059912802962824e-05,
+      "loss": 0.097,
+      "step": 101265
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.0598659228353102e-05,
+      "loss": 0.1617,
+      "step": 101270
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.0598190427077963e-05,
+      "loss": 0.3105,
+      "step": 101275
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.0597721625802823e-05,
+      "loss": 0.0835,
+      "step": 101280
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.0597252824527685e-05,
+      "loss": 0.0294,
+      "step": 101285
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.0596784023252545e-05,
+      "loss": 0.0231,
+      "step": 101290
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.0596315221977405e-05,
+      "loss": 0.0334,
+      "step": 101295
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.0595846420702265e-05,
+      "loss": 0.0379,
+      "step": 101300
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.0595377619427124e-05,
+      "loss": 0.1497,
+      "step": 101305
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.0594908818151986e-05,
+      "loss": 0.0553,
+      "step": 101310
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.0594440016876846e-05,
+      "loss": 0.1355,
+      "step": 101315
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.0593971215601708e-05,
+      "loss": 0.1098,
+      "step": 101320
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.059350241432657e-05,
+      "loss": 0.3215,
+      "step": 101325
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.059303361305143e-05,
+      "loss": 0.0571,
+      "step": 101330
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.0592564811776289e-05,
+      "loss": 0.0312,
+      "step": 101335
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.0592096010501149e-05,
+      "loss": 0.0226,
+      "step": 101340
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.0591627209226009e-05,
+      "loss": 0.023,
+      "step": 101345
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.059115840795087e-05,
+      "loss": 0.0544,
+      "step": 101350
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.059068960667573e-05,
+      "loss": 0.0402,
+      "step": 101355
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.059022080540059e-05,
+      "loss": 0.0666,
+      "step": 101360
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.0589752004125454e-05,
+      "loss": 0.1517,
+      "step": 101365
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.0589283202850314e-05,
+      "loss": 0.2026,
+      "step": 101370
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.0588814401575174e-05,
+      "loss": 0.2658,
+      "step": 101375
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.0588345600300034e-05,
+      "loss": 0.068,
+      "step": 101380
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.0587876799024894e-05,
+      "loss": 0.0279,
+      "step": 101385
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.0587407997749755e-05,
+      "loss": 0.035,
+      "step": 101390
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.0586939196474615e-05,
+      "loss": 0.0521,
+      "step": 101395
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.0586470395199475e-05,
+      "loss": 0.0612,
+      "step": 101400
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.0586001593924335e-05,
+      "loss": 0.0438,
+      "step": 101405
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.0585532792649198e-05,
+      "loss": 0.0862,
+      "step": 101410
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.0585063991374058e-05,
+      "loss": 0.1029,
+      "step": 101415
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.0584595190098918e-05,
+      "loss": 0.2249,
+      "step": 101420
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.0584126388823778e-05,
+      "loss": 0.3748,
+      "step": 101425
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.058365758754864e-05,
+      "loss": 0.0915,
+      "step": 101430
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.05831887862735e-05,
+      "loss": 0.0284,
+      "step": 101435
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.058271998499836e-05,
+      "loss": 0.0441,
+      "step": 101440
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.058225118372322e-05,
+      "loss": 0.0797,
+      "step": 101445
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.058178238244808e-05,
+      "loss": 0.0886,
+      "step": 101450
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.0581313581172941e-05,
+      "loss": 0.0483,
+      "step": 101455
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.0580844779897803e-05,
+      "loss": 0.0716,
+      "step": 101460
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.0580375978622663e-05,
+      "loss": 0.0923,
+      "step": 101465
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.0579907177347524e-05,
+      "loss": 0.1421,
+      "step": 101470
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.0579438376072384e-05,
+      "loss": 0.1857,
+      "step": 101475
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.0578969574797244e-05,
+      "loss": 0.0826,
+      "step": 101480
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.0578500773522104e-05,
+      "loss": 0.0218,
+      "step": 101485
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.0578031972246966e-05,
+      "loss": 0.0298,
+      "step": 101490
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.0577563170971826e-05,
+      "loss": 0.0628,
+      "step": 101495
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.0577094369696686e-05,
+      "loss": 0.0613,
+      "step": 101500
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.0576625568421547e-05,
+      "loss": 0.0306,
+      "step": 101505
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.0576156767146409e-05,
+      "loss": 0.0818,
+      "step": 101510
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.0575687965871269e-05,
+      "loss": 0.0592,
+      "step": 101515
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.0575219164596129e-05,
+      "loss": 0.1266,
+      "step": 101520
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.0574750363320989e-05,
+      "loss": 0.217,
+      "step": 101525
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.057428156204585e-05,
+      "loss": 0.0769,
+      "step": 101530
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.057381276077071e-05,
+      "loss": 0.0268,
+      "step": 101535
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.057334395949557e-05,
+      "loss": 0.0274,
+      "step": 101540
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.057287515822043e-05,
+      "loss": 0.0371,
+      "step": 101545
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.0572406356945293e-05,
+      "loss": 0.0786,
+      "step": 101550
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.0571937555670153e-05,
+      "loss": 0.0879,
+      "step": 101555
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.0571468754395013e-05,
+      "loss": 0.0993,
+      "step": 101560
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.0570999953119873e-05,
+      "loss": 0.1341,
+      "step": 101565
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.0570531151844735e-05,
+      "loss": 0.238,
+      "step": 101570
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.0570062350569595e-05,
+      "loss": 0.2976,
+      "step": 101575
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.0569593549294455e-05,
+      "loss": 0.0692,
+      "step": 101580
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.0569124748019315e-05,
+      "loss": 0.03,
+      "step": 101585
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.0568655946744175e-05,
+      "loss": 0.0146,
+      "step": 101590
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.0568187145469036e-05,
+      "loss": 0.0406,
+      "step": 101595
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.0567718344193898e-05,
+      "loss": 0.0327,
+      "step": 101600
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.0567249542918758e-05,
+      "loss": 0.0249,
+      "step": 101605
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.056678074164362e-05,
+      "loss": 0.0787,
+      "step": 101610
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.056631194036848e-05,
+      "loss": 0.0633,
+      "step": 101615
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.056584313909334e-05,
+      "loss": 0.0889,
+      "step": 101620
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.05653743378182e-05,
+      "loss": 0.1909,
+      "step": 101625
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.0564905536543059e-05,
+      "loss": 0.0471,
+      "step": 101630
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.056443673526792e-05,
+      "loss": 0.029,
+      "step": 101635
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.056396793399278e-05,
+      "loss": 0.0324,
+      "step": 101640
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.0563499132717642e-05,
+      "loss": 0.1051,
+      "step": 101645
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.0563030331442504e-05,
+      "loss": 0.0491,
+      "step": 101650
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.0562561530167364e-05,
+      "loss": 0.053,
+      "step": 101655
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.0562092728892224e-05,
+      "loss": 0.0921,
+      "step": 101660
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.0561623927617084e-05,
+      "loss": 0.1537,
+      "step": 101665
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.0561155126341944e-05,
+      "loss": 0.1917,
+      "step": 101670
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.0560686325066805e-05,
+      "loss": 0.2712,
+      "step": 101675
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.0560217523791665e-05,
+      "loss": 0.0696,
+      "step": 101680
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.0559748722516525e-05,
+      "loss": 0.0133,
+      "step": 101685
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.0559279921241388e-05,
+      "loss": 0.0415,
+      "step": 101690
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.0558811119966248e-05,
+      "loss": 0.0612,
+      "step": 101695
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.0558342318691108e-05,
+      "loss": 0.0546,
+      "step": 101700
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.0557873517415968e-05,
+      "loss": 0.0456,
+      "step": 101705
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.0557404716140828e-05,
+      "loss": 0.1161,
+      "step": 101710
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.055693591486569e-05,
+      "loss": 0.0825,
+      "step": 101715
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.055646711359055e-05,
+      "loss": 0.1404,
+      "step": 101720
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.055599831231541e-05,
+      "loss": 0.3184,
+      "step": 101725
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.055552951104027e-05,
+      "loss": 0.0752,
+      "step": 101730
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.0555060709765133e-05,
+      "loss": 0.0162,
+      "step": 101735
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.0554591908489993e-05,
+      "loss": 0.0307,
+      "step": 101740
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.0554123107214853e-05,
+      "loss": 0.0248,
+      "step": 101745
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.0553654305939713e-05,
+      "loss": 0.0339,
+      "step": 101750
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.0553185504664574e-05,
+      "loss": 0.0374,
+      "step": 101755
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.0552716703389434e-05,
+      "loss": 0.0871,
+      "step": 101760
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.0552247902114294e-05,
+      "loss": 0.0959,
+      "step": 101765
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.0551779100839154e-05,
+      "loss": 0.1684,
+      "step": 101770
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.0551310299564014e-05,
+      "loss": 0.2322,
+      "step": 101775
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.0550841498288876e-05,
+      "loss": 0.08,
+      "step": 101780
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.0550372697013737e-05,
+      "loss": 0.0207,
+      "step": 101785
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.0549903895738597e-05,
+      "loss": 0.0214,
+      "step": 101790
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.0549435094463459e-05,
+      "loss": 0.0576,
+      "step": 101795
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.0548966293188319e-05,
+      "loss": 0.0363,
+      "step": 101800
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.0548497491913179e-05,
+      "loss": 0.0223,
+      "step": 101805
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.0548028690638039e-05,
+      "loss": 0.073,
+      "step": 101810
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.0547559889362899e-05,
+      "loss": 0.0442,
+      "step": 101815
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.054709108808776e-05,
+      "loss": 0.1845,
+      "step": 101820
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.054662228681262e-05,
+      "loss": 0.2365,
+      "step": 101825
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.0546153485537482e-05,
+      "loss": 0.1072,
+      "step": 101830
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.0545684684262343e-05,
+      "loss": 0.0228,
+      "step": 101835
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.0545215882987203e-05,
+      "loss": 0.0466,
+      "step": 101840
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.0544747081712063e-05,
+      "loss": 0.0442,
+      "step": 101845
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.0544278280436923e-05,
+      "loss": 0.0952,
+      "step": 101850
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.0543809479161783e-05,
+      "loss": 0.0889,
+      "step": 101855
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.0543340677886645e-05,
+      "loss": 0.0531,
+      "step": 101860
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.0542871876611505e-05,
+      "loss": 0.055,
+      "step": 101865
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.0542403075336365e-05,
+      "loss": 0.1037,
+      "step": 101870
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.0541934274061228e-05,
+      "loss": 0.2653,
+      "step": 101875
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.0541465472786088e-05,
+      "loss": 0.0719,
+      "step": 101880
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.0540996671510948e-05,
+      "loss": 0.0188,
+      "step": 101885
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.0540527870235808e-05,
+      "loss": 0.0263,
+      "step": 101890
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.0540059068960668e-05,
+      "loss": 0.0209,
+      "step": 101895
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.053959026768553e-05,
+      "loss": 0.0561,
+      "step": 101900
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.053912146641039e-05,
+      "loss": 0.0401,
+      "step": 101905
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.053865266513525e-05,
+      "loss": 0.0429,
+      "step": 101910
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.053818386386011e-05,
+      "loss": 0.187,
+      "step": 101915
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.053771506258497e-05,
+      "loss": 0.1474,
+      "step": 101920
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.0537246261309832e-05,
+      "loss": 0.2741,
+      "step": 101925
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.0536777460034692e-05,
+      "loss": 0.0729,
+      "step": 101930
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.0536308658759552e-05,
+      "loss": 0.0145,
+      "step": 101935
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.0535839857484414e-05,
+      "loss": 0.0065,
+      "step": 101940
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.0535371056209274e-05,
+      "loss": 0.0341,
+      "step": 101945
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.0534902254934134e-05,
+      "loss": 0.0818,
+      "step": 101950
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.0534433453658994e-05,
+      "loss": 0.0284,
+      "step": 101955
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.0533964652383855e-05,
+      "loss": 0.1548,
+      "step": 101960
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.0533495851108715e-05,
+      "loss": 0.1037,
+      "step": 101965
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.0533027049833577e-05,
+      "loss": 0.1463,
+      "step": 101970
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.0532558248558437e-05,
+      "loss": 0.2978,
+      "step": 101975
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.0532089447283299e-05,
+      "loss": 0.0927,
+      "step": 101980
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.0531620646008158e-05,
+      "loss": 0.0104,
+      "step": 101985
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.0531151844733018e-05,
+      "loss": 0.0138,
+      "step": 101990
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.0530683043457878e-05,
+      "loss": 0.0604,
+      "step": 101995
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.053021424218274e-05,
+      "loss": 0.0422,
+      "step": 102000
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.05297454409076e-05,
+      "loss": 0.0416,
+      "step": 102005
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.052927663963246e-05,
+      "loss": 0.0786,
+      "step": 102010
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.0528807838357321e-05,
+      "loss": 0.0921,
+      "step": 102015
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.0528339037082183e-05,
+      "loss": 0.118,
+      "step": 102020
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.0527870235807043e-05,
+      "loss": 0.2083,
+      "step": 102025
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.0527401434531903e-05,
+      "loss": 0.0936,
+      "step": 102030
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.0526932633256763e-05,
+      "loss": 0.0237,
+      "step": 102035
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.0526463831981624e-05,
+      "loss": 0.0362,
+      "step": 102040
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.0525995030706484e-05,
+      "loss": 0.027,
+      "step": 102045
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.0525526229431344e-05,
+      "loss": 0.1199,
+      "step": 102050
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.0525057428156204e-05,
+      "loss": 0.0709,
+      "step": 102055
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.0524588626881068e-05,
+      "loss": 0.1292,
+      "step": 102060
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.0524119825605928e-05,
+      "loss": 0.1528,
+      "step": 102065
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.0523651024330787e-05,
+      "loss": 0.183,
+      "step": 102070
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.0523182223055647e-05,
+      "loss": 0.3114,
+      "step": 102075
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.0522713421780509e-05,
+      "loss": 0.0515,
+      "step": 102080
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.0522244620505369e-05,
+      "loss": 0.0123,
+      "step": 102085
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.0521775819230229e-05,
+      "loss": 0.0325,
+      "step": 102090
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.0521307017955089e-05,
+      "loss": 0.0539,
+      "step": 102095
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.0520838216679949e-05,
+      "loss": 0.0782,
+      "step": 102100
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.052036941540481e-05,
+      "loss": 0.0703,
+      "step": 102105
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.0519900614129672e-05,
+      "loss": 0.1676,
+      "step": 102110
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.0519431812854532e-05,
+      "loss": 0.0893,
+      "step": 102115
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0518963011579394e-05,
+      "loss": 0.1772,
+      "step": 102120
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0518494210304254e-05,
+      "loss": 0.3778,
+      "step": 102125
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0518025409029113e-05,
+      "loss": 0.1143,
+      "step": 102130
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0517556607753973e-05,
+      "loss": 0.0178,
+      "step": 102135
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0517087806478833e-05,
+      "loss": 0.0604,
+      "step": 102140
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0516619005203695e-05,
+      "loss": 0.0426,
+      "step": 102145
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0516150203928555e-05,
+      "loss": 0.056,
+      "step": 102150
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0515681402653417e-05,
+      "loss": 0.0306,
+      "step": 102155
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0515212601378278e-05,
+      "loss": 0.1248,
+      "step": 102160
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0514743800103138e-05,
+      "loss": 0.1266,
+      "step": 102165
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0514274998827998e-05,
+      "loss": 0.0695,
+      "step": 102170
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0513806197552858e-05,
+      "loss": 0.2844,
+      "step": 102175
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0513337396277718e-05,
+      "loss": 0.1131,
+      "step": 102180
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.051286859500258e-05,
+      "loss": 0.0065,
+      "step": 102185
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.051239979372744e-05,
+      "loss": 0.073,
+      "step": 102190
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.05119309924523e-05,
+      "loss": 0.0746,
+      "step": 102195
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0511462191177163e-05,
+      "loss": 0.0159,
+      "step": 102200
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0510993389902023e-05,
+      "loss": 0.1154,
+      "step": 102205
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0510524588626883e-05,
+      "loss": 0.0396,
+      "step": 102210
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0510055787351742e-05,
+      "loss": 0.1486,
+      "step": 102215
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0509586986076602e-05,
+      "loss": 0.1446,
+      "step": 102220
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0509118184801464e-05,
+      "loss": 0.2505,
+      "step": 102225
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0508649383526324e-05,
+      "loss": 0.0507,
+      "step": 102230
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0508180582251184e-05,
+      "loss": 0.0063,
+      "step": 102235
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0507711780976044e-05,
+      "loss": 0.0407,
+      "step": 102240
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0507242979700904e-05,
+      "loss": 0.0539,
+      "step": 102245
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0506774178425767e-05,
+      "loss": 0.0282,
+      "step": 102250
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0506305377150627e-05,
+      "loss": 0.1024,
+      "step": 102255
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0505836575875487e-05,
+      "loss": 0.0387,
+      "step": 102260
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0505367774600349e-05,
+      "loss": 0.0949,
+      "step": 102265
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0504898973325209e-05,
+      "loss": 0.1512,
+      "step": 102270
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0504430172050068e-05,
+      "loss": 0.218,
+      "step": 102275
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0503961370774928e-05,
+      "loss": 0.061,
+      "step": 102280
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0503492569499788e-05,
+      "loss": 0.0243,
+      "step": 102285
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.050302376822465e-05,
+      "loss": 0.0249,
+      "step": 102290
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0502554966949512e-05,
+      "loss": 0.1478,
+      "step": 102295
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0502086165674372e-05,
+      "loss": 0.0684,
+      "step": 102300
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0501617364399233e-05,
+      "loss": 0.0813,
+      "step": 102305
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0501148563124093e-05,
+      "loss": 0.2099,
+      "step": 102310
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0500679761848953e-05,
+      "loss": 0.08,
+      "step": 102315
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0500210960573813e-05,
+      "loss": 0.1541,
+      "step": 102320
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0499742159298673e-05,
+      "loss": 0.177,
+      "step": 102325
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0499273358023535e-05,
+      "loss": 0.0886,
+      "step": 102330
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0498804556748394e-05,
+      "loss": 0.0221,
+      "step": 102335
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0498335755473256e-05,
+      "loss": 0.029,
+      "step": 102340
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0497866954198118e-05,
+      "loss": 0.0281,
+      "step": 102345
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0497398152922978e-05,
+      "loss": 0.2023,
+      "step": 102350
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0496929351647838e-05,
+      "loss": 0.1092,
+      "step": 102355
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0496460550372698e-05,
+      "loss": 0.0955,
+      "step": 102360
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0495991749097557e-05,
+      "loss": 0.0616,
+      "step": 102365
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0495522947822419e-05,
+      "loss": 0.1498,
+      "step": 102370
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0495054146547279e-05,
+      "loss": 0.3811,
+      "step": 102375
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0494585345272139e-05,
+      "loss": 0.0465,
+      "step": 102380
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0494116543997002e-05,
+      "loss": 0.0102,
+      "step": 102385
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0493647742721862e-05,
+      "loss": 0.0027,
+      "step": 102390
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0493178941446722e-05,
+      "loss": 0.0889,
+      "step": 102395
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0492710140171582e-05,
+      "loss": 0.039,
+      "step": 102400
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0492241338896442e-05,
+      "loss": 0.0563,
+      "step": 102405
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0491772537621304e-05,
+      "loss": 0.103,
+      "step": 102410
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0491303736346164e-05,
+      "loss": 0.0921,
+      "step": 102415
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0490834935071023e-05,
+      "loss": 0.2047,
+      "step": 102420
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0490366133795883e-05,
+      "loss": 0.2644,
+      "step": 102425
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0489897332520745e-05,
+      "loss": 0.0547,
+      "step": 102430
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0489428531245607e-05,
+      "loss": 0.0294,
+      "step": 102435
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0488959729970467e-05,
+      "loss": 0.0247,
+      "step": 102440
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0488490928695327e-05,
+      "loss": 0.0283,
+      "step": 102445
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0488022127420188e-05,
+      "loss": 0.102,
+      "step": 102450
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0487553326145048e-05,
+      "loss": 0.06,
+      "step": 102455
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0487084524869908e-05,
+      "loss": 0.0462,
+      "step": 102460
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0486615723594768e-05,
+      "loss": 0.0689,
+      "step": 102465
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.048614692231963e-05,
+      "loss": 0.1073,
+      "step": 102470
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.048567812104449e-05,
+      "loss": 0.13,
+      "step": 102475
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0485209319769351e-05,
+      "loss": 0.063,
+      "step": 102480
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0484740518494211e-05,
+      "loss": 0.0185,
+      "step": 102485
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0484271717219073e-05,
+      "loss": 0.0306,
+      "step": 102490
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0483802915943933e-05,
+      "loss": 0.0213,
+      "step": 102495
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0483334114668793e-05,
+      "loss": 0.0409,
+      "step": 102500
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0482865313393653e-05,
+      "loss": 0.0853,
+      "step": 102505
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0482396512118514e-05,
+      "loss": 0.1192,
+      "step": 102510
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0481927710843374e-05,
+      "loss": 0.1563,
+      "step": 102515
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0481458909568234e-05,
+      "loss": 0.1211,
+      "step": 102520
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0480990108293096e-05,
+      "loss": 0.2028,
+      "step": 102525
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0480521307017957e-05,
+      "loss": 0.056,
+      "step": 102530
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0480052505742817e-05,
+      "loss": 0.0188,
+      "step": 102535
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0479583704467677e-05,
+      "loss": 0.0183,
+      "step": 102540
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0479114903192537e-05,
+      "loss": 0.0444,
+      "step": 102545
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.0478646101917399e-05,
+      "loss": 0.0253,
+      "step": 102550
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.0478177300642259e-05,
+      "loss": 0.0455,
+      "step": 102555
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.0477708499367119e-05,
+      "loss": 0.0746,
+      "step": 102560
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.0477239698091979e-05,
+      "loss": 0.0873,
+      "step": 102565
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.0476770896816838e-05,
+      "loss": 0.1166,
+      "step": 102570
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.0476302095541702e-05,
+      "loss": 0.3312,
+      "step": 102575
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.0475833294266562e-05,
+      "loss": 0.0987,
+      "step": 102580
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.0475364492991422e-05,
+      "loss": 0.0258,
+      "step": 102585
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.0474895691716283e-05,
+      "loss": 0.055,
+      "step": 102590
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.0474426890441143e-05,
+      "loss": 0.0206,
+      "step": 102595
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.0473958089166003e-05,
+      "loss": 0.0802,
+      "step": 102600
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.0473489287890863e-05,
+      "loss": 0.0479,
+      "step": 102605
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.0473020486615723e-05,
+      "loss": 0.204,
+      "step": 102610
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.0472551685340585e-05,
+      "loss": 0.2111,
+      "step": 102615
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.0472082884065446e-05,
+      "loss": 0.1316,
+      "step": 102620
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.0471614082790306e-05,
+      "loss": 0.2781,
+      "step": 102625
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.0471145281515168e-05,
+      "loss": 0.0586,
+      "step": 102630
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.0470676480240028e-05,
+      "loss": 0.0007,
+      "step": 102635
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.0470207678964888e-05,
+      "loss": 0.0742,
+      "step": 102640
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.0469738877689748e-05,
+      "loss": 0.0156,
+      "step": 102645
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.0469270076414608e-05,
+      "loss": 0.0449,
+      "step": 102650
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.046880127513947e-05,
+      "loss": 0.0215,
+      "step": 102655
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.0468332473864329e-05,
+      "loss": 0.0911,
+      "step": 102660
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.046786367258919e-05,
+      "loss": 0.0647,
+      "step": 102665
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.0467394871314052e-05,
+      "loss": 0.1279,
+      "step": 102670
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.0466926070038912e-05,
+      "loss": 0.4815,
+      "step": 102675
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.0466457268763772e-05,
+      "loss": 0.0562,
+      "step": 102680
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.0465988467488632e-05,
+      "loss": 0.0061,
+      "step": 102685
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.0465519666213492e-05,
+      "loss": 0.0062,
+      "step": 102690
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.0465050864938354e-05,
+      "loss": 0.0523,
+      "step": 102695
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.0464582063663214e-05,
+      "loss": 0.0474,
+      "step": 102700
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.0464113262388074e-05,
+      "loss": 0.0315,
+      "step": 102705
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.0463644461112937e-05,
+      "loss": 0.0587,
+      "step": 102710
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.0463175659837797e-05,
+      "loss": 0.0522,
+      "step": 102715
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.0462706858562657e-05,
+      "loss": 0.185,
+      "step": 102720
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.0462238057287517e-05,
+      "loss": 0.3021,
+      "step": 102725
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.0461769256012377e-05,
+      "loss": 0.0592,
+      "step": 102730
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.0461300454737238e-05,
+      "loss": 0.0363,
+      "step": 102735
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.0460831653462098e-05,
+      "loss": 0.0202,
+      "step": 102740
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.0460362852186958e-05,
+      "loss": 0.0779,
+      "step": 102745
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.0459894050911818e-05,
+      "loss": 0.0314,
+      "step": 102750
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.0459425249636678e-05,
+      "loss": 0.0416,
+      "step": 102755
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.0458956448361541e-05,
+      "loss": 0.0725,
+      "step": 102760
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.0458487647086401e-05,
+      "loss": 0.1599,
+      "step": 102765
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.0458018845811261e-05,
+      "loss": 0.1278,
+      "step": 102770
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.0457550044536123e-05,
+      "loss": 0.2902,
+      "step": 102775
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.0457081243260983e-05,
+      "loss": 0.0624,
+      "step": 102780
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.0456612441985843e-05,
+      "loss": 0.0076,
+      "step": 102785
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.0456143640710703e-05,
+      "loss": 0.0258,
+      "step": 102790
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.0455674839435563e-05,
+      "loss": 0.0954,
+      "step": 102795
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.0455206038160424e-05,
+      "loss": 0.0258,
+      "step": 102800
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.0454737236885286e-05,
+      "loss": 0.0543,
+      "step": 102805
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.0454268435610146e-05,
+      "loss": 0.0732,
+      "step": 102810
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.0453799634335007e-05,
+      "loss": 0.1822,
+      "step": 102815
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.0453330833059867e-05,
+      "loss": 0.2422,
+      "step": 102820
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.0452862031784727e-05,
+      "loss": 0.48,
+      "step": 102825
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.0452393230509587e-05,
+      "loss": 0.0662,
+      "step": 102830
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.0451924429234447e-05,
+      "loss": 0.0167,
+      "step": 102835
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.0451455627959309e-05,
+      "loss": 0.0499,
+      "step": 102840
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.0450986826684169e-05,
+      "loss": 0.0641,
+      "step": 102845
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.045051802540903e-05,
+      "loss": 0.0767,
+      "step": 102850
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.0450049224133892e-05,
+      "loss": 0.0573,
+      "step": 102855
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.0449580422858752e-05,
+      "loss": 0.0918,
+      "step": 102860
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.0449111621583612e-05,
+      "loss": 0.1074,
+      "step": 102865
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.0448642820308472e-05,
+      "loss": 0.1516,
+      "step": 102870
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.0448174019033332e-05,
+      "loss": 0.1747,
+      "step": 102875
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.0447705217758193e-05,
+      "loss": 0.0691,
+      "step": 102880
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.0447236416483053e-05,
+      "loss": 0.0151,
+      "step": 102885
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.0446767615207913e-05,
+      "loss": 0.0072,
+      "step": 102890
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.0446298813932773e-05,
+      "loss": 0.0563,
+      "step": 102895
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.0445830012657636e-05,
+      "loss": 0.0934,
+      "step": 102900
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.0445361211382496e-05,
+      "loss": 0.0263,
+      "step": 102905
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.0444892410107356e-05,
+      "loss": 0.1091,
+      "step": 102910
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.0444423608832218e-05,
+      "loss": 0.0825,
+      "step": 102915
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.0443954807557078e-05,
+      "loss": 0.1376,
+      "step": 102920
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.0443486006281938e-05,
+      "loss": 0.2562,
+      "step": 102925
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.0443017205006798e-05,
+      "loss": 0.0418,
+      "step": 102930
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.0442548403731658e-05,
+      "loss": 0.0191,
+      "step": 102935
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.044207960245652e-05,
+      "loss": 0.0611,
+      "step": 102940
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.0441610801181381e-05,
+      "loss": 0.0371,
+      "step": 102945
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.044114199990624e-05,
+      "loss": 0.0461,
+      "step": 102950
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.0440673198631102e-05,
+      "loss": 0.0477,
+      "step": 102955
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.0440204397355962e-05,
+      "loss": 0.067,
+      "step": 102960
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.0439735596080822e-05,
+      "loss": 0.1263,
+      "step": 102965
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.0439266794805682e-05,
+      "loss": 0.1754,
+      "step": 102970
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.0438797993530542e-05,
+      "loss": 0.2918,
+      "step": 102975
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0438329192255404e-05,
+      "loss": 0.0594,
+      "step": 102980
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0437860390980264e-05,
+      "loss": 0.0457,
+      "step": 102985
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0437391589705125e-05,
+      "loss": 0.0798,
+      "step": 102990
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0436922788429987e-05,
+      "loss": 0.059,
+      "step": 102995
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0436453987154847e-05,
+      "loss": 0.0832,
+      "step": 103000
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0435985185879707e-05,
+      "loss": 0.0854,
+      "step": 103005
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0435516384604567e-05,
+      "loss": 0.0542,
+      "step": 103010
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0435047583329427e-05,
+      "loss": 0.1214,
+      "step": 103015
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0434578782054288e-05,
+      "loss": 0.1712,
+      "step": 103020
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0434109980779148e-05,
+      "loss": 0.2041,
+      "step": 103025
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0433641179504008e-05,
+      "loss": 0.0414,
+      "step": 103030
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0433172378228872e-05,
+      "loss": 0.0617,
+      "step": 103035
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0432703576953731e-05,
+      "loss": 0.0553,
+      "step": 103040
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0432234775678591e-05,
+      "loss": 0.0496,
+      "step": 103045
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0431765974403451e-05,
+      "loss": 0.052,
+      "step": 103050
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0431297173128311e-05,
+      "loss": 0.0692,
+      "step": 103055
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0430828371853173e-05,
+      "loss": 0.0978,
+      "step": 103060
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0430359570578033e-05,
+      "loss": 0.1229,
+      "step": 103065
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0429890769302893e-05,
+      "loss": 0.0788,
+      "step": 103070
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0429421968027753e-05,
+      "loss": 0.249,
+      "step": 103075
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0428953166752613e-05,
+      "loss": 0.0913,
+      "step": 103080
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0428484365477476e-05,
+      "loss": 0.0702,
+      "step": 103085
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0428015564202336e-05,
+      "loss": 0.0502,
+      "step": 103090
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0427546762927196e-05,
+      "loss": 0.083,
+      "step": 103095
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0427077961652057e-05,
+      "loss": 0.053,
+      "step": 103100
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0426609160376917e-05,
+      "loss": 0.1018,
+      "step": 103105
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0426140359101777e-05,
+      "loss": 0.126,
+      "step": 103110
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0425671557826637e-05,
+      "loss": 0.0771,
+      "step": 103115
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0425202756551497e-05,
+      "loss": 0.1753,
+      "step": 103120
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0424733955276359e-05,
+      "loss": 0.1639,
+      "step": 103125
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.042426515400122e-05,
+      "loss": 0.0716,
+      "step": 103130
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.042379635272608e-05,
+      "loss": 0.0139,
+      "step": 103135
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0423327551450942e-05,
+      "loss": 0.0337,
+      "step": 103140
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0422858750175802e-05,
+      "loss": 0.0685,
+      "step": 103145
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0422389948900662e-05,
+      "loss": 0.0434,
+      "step": 103150
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0421921147625522e-05,
+      "loss": 0.0754,
+      "step": 103155
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0421452346350382e-05,
+      "loss": 0.0806,
+      "step": 103160
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0420983545075243e-05,
+      "loss": 0.1579,
+      "step": 103165
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0420514743800103e-05,
+      "loss": 0.2571,
+      "step": 103170
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0420045942524965e-05,
+      "loss": 0.2651,
+      "step": 103175
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0419577141249827e-05,
+      "loss": 0.1012,
+      "step": 103180
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0419108339974687e-05,
+      "loss": 0.0798,
+      "step": 103185
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0418639538699546e-05,
+      "loss": 0.0407,
+      "step": 103190
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.0418170737424406e-05,
+      "loss": 0.0716,
+      "step": 103195
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.0417701936149266e-05,
+      "loss": 0.0265,
+      "step": 103200
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.0417233134874128e-05,
+      "loss": 0.0514,
+      "step": 103205
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.0416764333598988e-05,
+      "loss": 0.1115,
+      "step": 103210
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.0416295532323848e-05,
+      "loss": 0.0714,
+      "step": 103215
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.0415826731048708e-05,
+      "loss": 0.1274,
+      "step": 103220
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.0415357929773571e-05,
+      "loss": 0.3215,
+      "step": 103225
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.0414889128498431e-05,
+      "loss": 0.1016,
+      "step": 103230
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.0414420327223291e-05,
+      "loss": 0.025,
+      "step": 103235
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.0413951525948151e-05,
+      "loss": 0.0103,
+      "step": 103240
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.0413482724673012e-05,
+      "loss": 0.0244,
+      "step": 103245
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.0413013923397872e-05,
+      "loss": 0.0437,
+      "step": 103250
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.0412545122122732e-05,
+      "loss": 0.0632,
+      "step": 103255
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.0412076320847592e-05,
+      "loss": 0.076,
+      "step": 103260
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.0411607519572454e-05,
+      "loss": 0.0733,
+      "step": 103265
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.0411138718297316e-05,
+      "loss": 0.1895,
+      "step": 103270
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.0410669917022175e-05,
+      "loss": 0.2007,
+      "step": 103275
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.0410201115747035e-05,
+      "loss": 0.0924,
+      "step": 103280
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.0409732314471897e-05,
+      "loss": 0.0607,
+      "step": 103285
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.0409263513196757e-05,
+      "loss": 0.0297,
+      "step": 103290
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.0408794711921617e-05,
+      "loss": 0.0709,
+      "step": 103295
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.0408325910646477e-05,
+      "loss": 0.0464,
+      "step": 103300
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.0407857109371338e-05,
+      "loss": 0.0611,
+      "step": 103305
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.0407388308096198e-05,
+      "loss": 0.0293,
+      "step": 103310
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.040691950682106e-05,
+      "loss": 0.0788,
+      "step": 103315
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.040645070554592e-05,
+      "loss": 0.2029,
+      "step": 103320
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.0405981904270782e-05,
+      "loss": 0.3586,
+      "step": 103325
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.0405513102995642e-05,
+      "loss": 0.0467,
+      "step": 103330
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.0405044301720501e-05,
+      "loss": 0.0179,
+      "step": 103335
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.0404575500445361e-05,
+      "loss": 0.0412,
+      "step": 103340
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.0404106699170223e-05,
+      "loss": 0.0508,
+      "step": 103345
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.0403637897895083e-05,
+      "loss": 0.0399,
+      "step": 103350
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.0403169096619943e-05,
+      "loss": 0.0471,
+      "step": 103355
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.0402700295344803e-05,
+      "loss": 0.1354,
+      "step": 103360
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.0402231494069666e-05,
+      "loss": 0.1801,
+      "step": 103365
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.0401762692794526e-05,
+      "loss": 0.1218,
+      "step": 103370
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.0401293891519386e-05,
+      "loss": 0.2205,
+      "step": 103375
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.0400825090244246e-05,
+      "loss": 0.0513,
+      "step": 103380
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.0400356288969108e-05,
+      "loss": 0.0253,
+      "step": 103385
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.0399887487693968e-05,
+      "loss": 0.0256,
+      "step": 103390
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.0399418686418827e-05,
+      "loss": 0.0203,
+      "step": 103395
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.0398949885143687e-05,
+      "loss": 0.1352,
+      "step": 103400
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.0398481083868547e-05,
+      "loss": 0.0724,
+      "step": 103405
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.039801228259341e-05,
+      "loss": 0.0821,
+      "step": 103410
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.039754348131827e-05,
+      "loss": 0.0651,
+      "step": 103415
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.039707468004313e-05,
+      "loss": 0.2306,
+      "step": 103420
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.0396605878767992e-05,
+      "loss": 0.1944,
+      "step": 103425
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.0396137077492852e-05,
+      "loss": 0.0688,
+      "step": 103430
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.0395668276217712e-05,
+      "loss": 0.0145,
+      "step": 103435
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.0395199474942572e-05,
+      "loss": 0.0411,
+      "step": 103440
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.0394730673667432e-05,
+      "loss": 0.0541,
+      "step": 103445
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.0394261872392293e-05,
+      "loss": 0.0867,
+      "step": 103450
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.0393793071117155e-05,
+      "loss": 0.0432,
+      "step": 103455
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.0393324269842015e-05,
+      "loss": 0.068,
+      "step": 103460
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.0392855468566877e-05,
+      "loss": 0.2069,
+      "step": 103465
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.0392386667291737e-05,
+      "loss": 0.1491,
+      "step": 103470
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.0391917866016597e-05,
+      "loss": 0.182,
+      "step": 103475
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.0391449064741456e-05,
+      "loss": 0.0641,
+      "step": 103480
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.0390980263466316e-05,
+      "loss": 0.0513,
+      "step": 103485
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.0390511462191178e-05,
+      "loss": 0.0285,
+      "step": 103490
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.0390042660916038e-05,
+      "loss": 0.0453,
+      "step": 103495
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.03895738596409e-05,
+      "loss": 0.029,
+      "step": 103500
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.0389105058365761e-05,
+      "loss": 0.0194,
+      "step": 103505
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.0388636257090621e-05,
+      "loss": 0.0486,
+      "step": 103510
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.0388167455815481e-05,
+      "loss": 0.0905,
+      "step": 103515
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.0387698654540341e-05,
+      "loss": 0.1716,
+      "step": 103520
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.0387229853265201e-05,
+      "loss": 0.1719,
+      "step": 103525
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.0386761051990063e-05,
+      "loss": 0.0856,
+      "step": 103530
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.0386292250714923e-05,
+      "loss": 0.0262,
+      "step": 103535
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.0385823449439782e-05,
+      "loss": 0.0259,
+      "step": 103540
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.0385354648164642e-05,
+      "loss": 0.0864,
+      "step": 103545
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.0384885846889506e-05,
+      "loss": 0.0321,
+      "step": 103550
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.0384417045614366e-05,
+      "loss": 0.0462,
+      "step": 103555
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.0383948244339226e-05,
+      "loss": 0.0652,
+      "step": 103560
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.0383479443064086e-05,
+      "loss": 0.1621,
+      "step": 103565
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.0383010641788947e-05,
+      "loss": 0.1763,
+      "step": 103570
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.0382541840513807e-05,
+      "loss": 0.3932,
+      "step": 103575
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.0382073039238667e-05,
+      "loss": 0.0898,
+      "step": 103580
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.0381604237963527e-05,
+      "loss": 0.0166,
+      "step": 103585
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.0381135436688387e-05,
+      "loss": 0.0185,
+      "step": 103590
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.038066663541325e-05,
+      "loss": 0.0515,
+      "step": 103595
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.038019783413811e-05,
+      "loss": 0.0784,
+      "step": 103600
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.037972903286297e-05,
+      "loss": 0.0554,
+      "step": 103605
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.0379260231587832e-05,
+      "loss": 0.1382,
+      "step": 103610
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.0378791430312692e-05,
+      "loss": 0.0873,
+      "step": 103615
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.0378322629037552e-05,
+      "loss": 0.1708,
+      "step": 103620
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.0377853827762411e-05,
+      "loss": 0.1861,
+      "step": 103625
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.0377385026487271e-05,
+      "loss": 0.0717,
+      "step": 103630
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.0376916225212133e-05,
+      "loss": 0.0392,
+      "step": 103635
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.0376447423936995e-05,
+      "loss": 0.02,
+      "step": 103640
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.0375978622661855e-05,
+      "loss": 0.0132,
+      "step": 103645
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.0375509821386716e-05,
+      "loss": 0.048,
+      "step": 103650
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.0375041020111576e-05,
+      "loss": 0.0678,
+      "step": 103655
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.0374572218836436e-05,
+      "loss": 0.1245,
+      "step": 103660
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.0374103417561296e-05,
+      "loss": 0.0481,
+      "step": 103665
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.0373634616286156e-05,
+      "loss": 0.1888,
+      "step": 103670
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.0373165815011018e-05,
+      "loss": 0.235,
+      "step": 103675
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.0372697013735878e-05,
+      "loss": 0.0594,
+      "step": 103680
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.0372228212460737e-05,
+      "loss": 0.0087,
+      "step": 103685
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.03717594111856e-05,
+      "loss": 0.0239,
+      "step": 103690
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.037129060991046e-05,
+      "loss": 0.0177,
+      "step": 103695
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.037082180863532e-05,
+      "loss": 0.0595,
+      "step": 103700
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.037035300736018e-05,
+      "loss": 0.0848,
+      "step": 103705
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.036988420608504e-05,
+      "loss": 0.0601,
+      "step": 103710
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.0369415404809902e-05,
+      "loss": 0.131,
+      "step": 103715
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.0368946603534762e-05,
+      "loss": 0.1303,
+      "step": 103720
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.0368477802259622e-05,
+      "loss": 0.2756,
+      "step": 103725
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.0368009000984482e-05,
+      "loss": 0.1217,
+      "step": 103730
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.0367540199709345e-05,
+      "loss": 0.0262,
+      "step": 103735
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.0367071398434205e-05,
+      "loss": 0.0098,
+      "step": 103740
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.0366602597159065e-05,
+      "loss": 0.0318,
+      "step": 103745
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.0366133795883925e-05,
+      "loss": 0.0758,
+      "step": 103750
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.0365664994608787e-05,
+      "loss": 0.0736,
+      "step": 103755
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.0365196193333647e-05,
+      "loss": 0.0925,
+      "step": 103760
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.0364727392058507e-05,
+      "loss": 0.1003,
+      "step": 103765
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.0364258590783367e-05,
+      "loss": 0.1505,
+      "step": 103770
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.0363789789508228e-05,
+      "loss": 0.1504,
+      "step": 103775
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.036332098823309e-05,
+      "loss": 0.0782,
+      "step": 103780
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.036285218695795e-05,
+      "loss": 0.0325,
+      "step": 103785
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.036238338568281e-05,
+      "loss": 0.0306,
+      "step": 103790
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.0361914584407671e-05,
+      "loss": 0.0332,
+      "step": 103795
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.0361445783132531e-05,
+      "loss": 0.0566,
+      "step": 103800
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.0360976981857391e-05,
+      "loss": 0.052,
+      "step": 103805
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.0360508180582251e-05,
+      "loss": 0.0845,
+      "step": 103810
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.0360039379307113e-05,
+      "loss": 0.0776,
+      "step": 103815
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.0359570578031973e-05,
+      "loss": 0.1386,
+      "step": 103820
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.0359101776756834e-05,
+      "loss": 0.3277,
+      "step": 103825
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.0358632975481694e-05,
+      "loss": 0.0518,
+      "step": 103830
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.0358164174206556e-05,
+      "loss": 0.0133,
+      "step": 103835
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.0357695372931416e-05,
+      "loss": 0.0351,
+      "step": 103840
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.0357226571656276e-05,
+      "loss": 0.0212,
+      "step": 103845
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.0356757770381136e-05,
+      "loss": 0.084,
+      "step": 103850
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.0356288969105997e-05,
+      "loss": 0.0242,
+      "step": 103855
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.0355820167830857e-05,
+      "loss": 0.0712,
+      "step": 103860
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.0355351366555717e-05,
+      "loss": 0.0509,
+      "step": 103865
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.0354882565280577e-05,
+      "loss": 0.1514,
+      "step": 103870
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.035441376400544e-05,
+      "loss": 0.2023,
+      "step": 103875
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.03539449627303e-05,
+      "loss": 0.0733,
+      "step": 103880
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.035347616145516e-05,
+      "loss": 0.0018,
+      "step": 103885
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.035300736018002e-05,
+      "loss": 0.003,
+      "step": 103890
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.0352538558904882e-05,
+      "loss": 0.0272,
+      "step": 103895
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.0352069757629742e-05,
+      "loss": 0.0233,
+      "step": 103900
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.0351600956354602e-05,
+      "loss": 0.1071,
+      "step": 103905
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.0351132155079462e-05,
+      "loss": 0.0926,
+      "step": 103910
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.0350663353804322e-05,
+      "loss": 0.1107,
+      "step": 103915
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.0350194552529185e-05,
+      "loss": 0.1934,
+      "step": 103920
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.0349725751254045e-05,
+      "loss": 0.2821,
+      "step": 103925
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.0349256949978905e-05,
+      "loss": 0.0885,
+      "step": 103930
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.0348788148703766e-05,
+      "loss": 0.0216,
+      "step": 103935
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.0348319347428626e-05,
+      "loss": 0.0211,
+      "step": 103940
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.0347850546153486e-05,
+      "loss": 0.0385,
+      "step": 103945
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.0347381744878346e-05,
+      "loss": 0.1131,
+      "step": 103950
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.0346912943603206e-05,
+      "loss": 0.1028,
+      "step": 103955
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.0346444142328068e-05,
+      "loss": 0.0696,
+      "step": 103960
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.034597534105293e-05,
+      "loss": 0.0906,
+      "step": 103965
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.034550653977779e-05,
+      "loss": 0.1768,
+      "step": 103970
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.0345037738502651e-05,
+      "loss": 0.2065,
+      "step": 103975
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.034456893722751e-05,
+      "loss": 0.1293,
+      "step": 103980
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.034410013595237e-05,
+      "loss": 0.0061,
+      "step": 103985
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.034363133467723e-05,
+      "loss": 0.0526,
+      "step": 103990
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.034316253340209e-05,
+      "loss": 0.0459,
+      "step": 103995
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.0342693732126952e-05,
+      "loss": 0.0759,
+      "step": 104000
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.0342224930851812e-05,
+      "loss": 0.2133,
+      "step": 104005
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.0341756129576672e-05,
+      "loss": 0.0962,
+      "step": 104010
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.0341287328301535e-05,
+      "loss": 0.1311,
+      "step": 104015
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.0340818527026395e-05,
+      "loss": 0.1578,
+      "step": 104020
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.0340349725751255e-05,
+      "loss": 0.2502,
+      "step": 104025
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.0339880924476115e-05,
+      "loss": 0.0771,
+      "step": 104030
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.0339412123200975e-05,
+      "loss": 0.0078,
+      "step": 104035
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.0338943321925837e-05,
+      "loss": 0.036,
+      "step": 104040
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.0338474520650697e-05,
+      "loss": 0.0299,
+      "step": 104045
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.0338005719375557e-05,
+      "loss": 0.0417,
+      "step": 104050
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.0337536918100417e-05,
+      "loss": 0.0297,
+      "step": 104055
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.033706811682528e-05,
+      "loss": 0.1254,
+      "step": 104060
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.033659931555014e-05,
+      "loss": 0.1065,
+      "step": 104065
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.0336130514275e-05,
+      "loss": 0.1066,
+      "step": 104070
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.033566171299986e-05,
+      "loss": 0.3243,
+      "step": 104075
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.0335192911724721e-05,
+      "loss": 0.0862,
+      "step": 104080
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.0334724110449581e-05,
+      "loss": 0.0327,
+      "step": 104085
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.0334255309174441e-05,
+      "loss": 0.0051,
+      "step": 104090
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.0333786507899301e-05,
+      "loss": 0.0404,
+      "step": 104095
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.0333317706624161e-05,
+      "loss": 0.0774,
+      "step": 104100
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.0332848905349024e-05,
+      "loss": 0.0577,
+      "step": 104105
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.0332380104073884e-05,
+      "loss": 0.153,
+      "step": 104110
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.0331911302798744e-05,
+      "loss": 0.1411,
+      "step": 104115
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.0331442501523606e-05,
+      "loss": 0.2148,
+      "step": 104120
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.0330973700248466e-05,
+      "loss": 0.2341,
+      "step": 104125
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.0330504898973326e-05,
+      "loss": 0.0696,
+      "step": 104130
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.0330036097698186e-05,
+      "loss": 0.0217,
+      "step": 104135
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.0329567296423046e-05,
+      "loss": 0.0227,
+      "step": 104140
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.0329098495147907e-05,
+      "loss": 0.0843,
+      "step": 104145
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.0328629693872769e-05,
+      "loss": 0.0313,
+      "step": 104150
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.0328160892597629e-05,
+      "loss": 0.0686,
+      "step": 104155
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.032769209132249e-05,
+      "loss": 0.1213,
+      "step": 104160
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.032722329004735e-05,
+      "loss": 0.0856,
+      "step": 104165
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.032675448877221e-05,
+      "loss": 0.1158,
+      "step": 104170
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.032628568749707e-05,
+      "loss": 0.2419,
+      "step": 104175
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.032581688622193e-05,
+      "loss": 0.0999,
+      "step": 104180
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.0325348084946792e-05,
+      "loss": 0.0753,
+      "step": 104185
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.0324879283671652e-05,
+      "loss": 0.0617,
+      "step": 104190
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.0324410482396512e-05,
+      "loss": 0.0473,
+      "step": 104195
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.0323941681121375e-05,
+      "loss": 0.0751,
+      "step": 104200
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.0323472879846235e-05,
+      "loss": 0.0753,
+      "step": 104205
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.0323004078571095e-05,
+      "loss": 0.0951,
+      "step": 104210
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.0322535277295955e-05,
+      "loss": 0.0756,
+      "step": 104215
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.0322066476020815e-05,
+      "loss": 0.0675,
+      "step": 104220
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.0321597674745676e-05,
+      "loss": 0.2918,
+      "step": 104225
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.0321128873470536e-05,
+      "loss": 0.0426,
+      "step": 104230
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.0320660072195396e-05,
+      "loss": 0.0393,
+      "step": 104235
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.0320191270920256e-05,
+      "loss": 0.0563,
+      "step": 104240
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.031972246964512e-05,
+      "loss": 0.0616,
+      "step": 104245
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.031925366836998e-05,
+      "loss": 0.0281,
+      "step": 104250
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.031878486709484e-05,
+      "loss": 0.0422,
+      "step": 104255
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.03183160658197e-05,
+      "loss": 0.1069,
+      "step": 104260
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.0317847264544561e-05,
+      "loss": 0.0781,
+      "step": 104265
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.0317378463269421e-05,
+      "loss": 0.1148,
+      "step": 104270
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.031690966199428e-05,
+      "loss": 0.2158,
+      "step": 104275
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.031644086071914e-05,
+      "loss": 0.0568,
+      "step": 104280
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.0315972059444002e-05,
+      "loss": 0.0198,
+      "step": 104285
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.0315503258168864e-05,
+      "loss": 0.0703,
+      "step": 104290
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.0315034456893724e-05,
+      "loss": 0.0625,
+      "step": 104295
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.0314565655618584e-05,
+      "loss": 0.0542,
+      "step": 104300
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.0314096854343445e-05,
+      "loss": 0.0107,
+      "step": 104305
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.0313628053068305e-05,
+      "loss": 0.195,
+      "step": 104310
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.0313159251793165e-05,
+      "loss": 0.0716,
+      "step": 104315
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.0312690450518025e-05,
+      "loss": 0.1751,
+      "step": 104320
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.0312221649242887e-05,
+      "loss": 0.294,
+      "step": 104325
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.0311752847967747e-05,
+      "loss": 0.0556,
+      "step": 104330
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.0311284046692607e-05,
+      "loss": 0.0109,
+      "step": 104335
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.031081524541747e-05,
+      "loss": 0.022,
+      "step": 104340
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.031034644414233e-05,
+      "loss": 0.0189,
+      "step": 104345
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.030987764286719e-05,
+      "loss": 0.0442,
+      "step": 104350
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.030940884159205e-05,
+      "loss": 0.0479,
+      "step": 104355
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.030894004031691e-05,
+      "loss": 0.0461,
+      "step": 104360
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.0308471239041771e-05,
+      "loss": 0.0752,
+      "step": 104365
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.0308002437766631e-05,
+      "loss": 0.1493,
+      "step": 104370
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.0307533636491491e-05,
+      "loss": 0.2768,
+      "step": 104375
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.0307064835216351e-05,
+      "loss": 0.0381,
+      "step": 104380
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.0306596033941215e-05,
+      "loss": 0.0201,
+      "step": 104385
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.0306127232666075e-05,
+      "loss": 0.0356,
+      "step": 104390
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.0305658431390934e-05,
+      "loss": 0.0462,
+      "step": 104395
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.0305189630115794e-05,
+      "loss": 0.0339,
+      "step": 104400
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.0304720828840656e-05,
+      "loss": 0.0334,
+      "step": 104405
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.0304252027565516e-05,
+      "loss": 0.1274,
+      "step": 104410
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.0303783226290376e-05,
+      "loss": 0.1074,
+      "step": 104415
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.0303314425015236e-05,
+      "loss": 0.1418,
+      "step": 104420
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.0302845623740096e-05,
+      "loss": 0.268,
+      "step": 104425
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.0302376822464959e-05,
+      "loss": 0.0409,
+      "step": 104430
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.0301908021189819e-05,
+      "loss": 0.0284,
+      "step": 104435
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.0301439219914679e-05,
+      "loss": 0.0847,
+      "step": 104440
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.030097041863954e-05,
+      "loss": 0.0451,
+      "step": 104445
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.03005016173644e-05,
+      "loss": 0.0835,
+      "step": 104450
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.030003281608926e-05,
+      "loss": 0.046,
+      "step": 104455
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.029956401481412e-05,
+      "loss": 0.042,
+      "step": 104460
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.029909521353898e-05,
+      "loss": 0.084,
+      "step": 104465
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.0298626412263842e-05,
+      "loss": 0.176,
+      "step": 104470
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.0298157610988704e-05,
+      "loss": 0.3081,
+      "step": 104475
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.0297688809713563e-05,
+      "loss": 0.0617,
+      "step": 104480
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.0297220008438425e-05,
+      "loss": 0.008,
+      "step": 104485
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.0296751207163285e-05,
+      "loss": 0.0287,
+      "step": 104490
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.0296282405888145e-05,
+      "loss": 0.0181,
+      "step": 104495
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.0295813604613005e-05,
+      "loss": 0.0593,
+      "step": 104500
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.0295344803337865e-05,
+      "loss": 0.0621,
+      "step": 104505
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.0294876002062726e-05,
+      "loss": 0.694,
+      "step": 104510
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.0294407200787586e-05,
+      "loss": 0.1163,
+      "step": 104515
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.0293938399512446e-05,
+      "loss": 0.1058,
+      "step": 104520
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.029346959823731e-05,
+      "loss": 0.322,
+      "step": 104525
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.029300079696217e-05,
+      "loss": 0.0987,
+      "step": 104530
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.029253199568703e-05,
+      "loss": 0.0057,
+      "step": 104535
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.029206319441189e-05,
+      "loss": 0.0148,
+      "step": 104540
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.029159439313675e-05,
+      "loss": 0.0796,
+      "step": 104545
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.0291125591861611e-05,
+      "loss": 0.0388,
+      "step": 104550
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.0290656790586471e-05,
+      "loss": 0.1049,
+      "step": 104555
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.0290187989311331e-05,
+      "loss": 0.0605,
+      "step": 104560
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.028971918803619e-05,
+      "loss": 0.2132,
+      "step": 104565
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.0289250386761054e-05,
+      "loss": 0.0882,
+      "step": 104570
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.0288781585485914e-05,
+      "loss": 0.2828,
+      "step": 104575
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.0288312784210774e-05,
+      "loss": 0.0809,
+      "step": 104580
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.0287843982935634e-05,
+      "loss": 0.0074,
+      "step": 104585
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.0287375181660496e-05,
+      "loss": 0.0453,
+      "step": 104590
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.0286906380385356e-05,
+      "loss": 0.0169,
+      "step": 104595
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.0286437579110215e-05,
+      "loss": 0.064,
+      "step": 104600
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.0285968777835075e-05,
+      "loss": 0.0555,
+      "step": 104605
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.0285499976559935e-05,
+      "loss": 0.0382,
+      "step": 104610
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.0285031175284799e-05,
+      "loss": 0.0682,
+      "step": 104615
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.0284562374009659e-05,
+      "loss": 0.1367,
+      "step": 104620
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.0284093572734518e-05,
+      "loss": 0.2854,
+      "step": 104625
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.028362477145938e-05,
+      "loss": 0.1447,
+      "step": 104630
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.028315597018424e-05,
+      "loss": 0.0061,
+      "step": 104635
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.02826871689091e-05,
+      "loss": 0.0346,
+      "step": 104640
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.028221836763396e-05,
+      "loss": 0.0155,
+      "step": 104645
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.0281749566358822e-05,
+      "loss": 0.0365,
+      "step": 104650
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.0281280765083681e-05,
+      "loss": 0.0685,
+      "step": 104655
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.0280811963808541e-05,
+      "loss": 0.1154,
+      "step": 104660
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.0280343162533403e-05,
+      "loss": 0.1566,
+      "step": 104665
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.0279874361258265e-05,
+      "loss": 0.0794,
+      "step": 104670
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.0279405559983125e-05,
+      "loss": 0.1238,
+      "step": 104675
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.0278936758707985e-05,
+      "loss": 0.0538,
+      "step": 104680
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.0278467957432844e-05,
+      "loss": 0.0478,
+      "step": 104685
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.0277999156157706e-05,
+      "loss": 0.0412,
+      "step": 104690
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.0277530354882566e-05,
+      "loss": 0.0181,
+      "step": 104695
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.0277061553607426e-05,
+      "loss": 0.0515,
+      "step": 104700
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.0276592752332286e-05,
+      "loss": 0.1135,
+      "step": 104705
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.027612395105715e-05,
+      "loss": 0.0426,
+      "step": 104710
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.027565514978201e-05,
+      "loss": 0.0606,
+      "step": 104715
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.0275186348506869e-05,
+      "loss": 0.1492,
+      "step": 104720
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.0274717547231729e-05,
+      "loss": 0.271,
+      "step": 104725
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.027424874595659e-05,
+      "loss": 0.0349,
+      "step": 104730
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.027377994468145e-05,
+      "loss": 0.0383,
+      "step": 104735
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.027331114340631e-05,
+      "loss": 0.085,
+      "step": 104740
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.027284234213117e-05,
+      "loss": 0.0233,
+      "step": 104745
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.027237354085603e-05,
+      "loss": 0.0605,
+      "step": 104750
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.0271904739580894e-05,
+      "loss": 0.0682,
+      "step": 104755
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.0271435938305754e-05,
+      "loss": 0.0947,
+      "step": 104760
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.0270967137030614e-05,
+      "loss": 0.1154,
+      "step": 104765
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.0270498335755475e-05,
+      "loss": 0.0902,
+      "step": 104770
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.0270029534480335e-05,
+      "loss": 0.1524,
+      "step": 104775
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.0269560733205195e-05,
+      "loss": 0.0757,
+      "step": 104780
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.0269091931930055e-05,
+      "loss": 0.0226,
+      "step": 104785
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.0268623130654915e-05,
+      "loss": 0.0261,
+      "step": 104790
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.0268154329379777e-05,
+      "loss": 0.055,
+      "step": 104795
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.0267685528104638e-05,
+      "loss": 0.092,
+      "step": 104800
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.0267216726829498e-05,
+      "loss": 0.0473,
+      "step": 104805
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.026674792555436e-05,
+      "loss": 0.1125,
+      "step": 104810
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.026627912427922e-05,
+      "loss": 0.0941,
+      "step": 104815
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.026581032300408e-05,
+      "loss": 0.0902,
+      "step": 104820
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.026534152172894e-05,
+      "loss": 0.328,
+      "step": 104825
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.02648727204538e-05,
+      "loss": 0.047,
+      "step": 104830
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.0264403919178661e-05,
+      "loss": 0.044,
+      "step": 104835
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.0263935117903521e-05,
+      "loss": 0.054,
+      "step": 104840
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.0263466316628381e-05,
+      "loss": 0.0373,
+      "step": 104845
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.0262997515353244e-05,
+      "loss": 0.0988,
+      "step": 104850
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.0262528714078104e-05,
+      "loss": 0.0431,
+      "step": 104855
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.0262059912802964e-05,
+      "loss": 0.0451,
+      "step": 104860
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.0261591111527824e-05,
+      "loss": 0.0645,
+      "step": 104865
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.0261122310252684e-05,
+      "loss": 0.1824,
+      "step": 104870
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.0260653508977546e-05,
+      "loss": 0.3749,
+      "step": 104875
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.0260184707702406e-05,
+      "loss": 0.1074,
+      "step": 104880
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.0259715906427266e-05,
+      "loss": 0.0058,
+      "step": 104885
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.0259247105152125e-05,
+      "loss": 0.0188,
+      "step": 104890
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.0258778303876989e-05,
+      "loss": 0.0376,
+      "step": 104895
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.0258309502601849e-05,
+      "loss": 0.053,
+      "step": 104900
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.0257840701326709e-05,
+      "loss": 0.052,
+      "step": 104905
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.0257371900051569e-05,
+      "loss": 0.0662,
+      "step": 104910
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.025690309877643e-05,
+      "loss": 0.0689,
+      "step": 104915
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.025643429750129e-05,
+      "loss": 0.1109,
+      "step": 104920
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.025596549622615e-05,
+      "loss": 0.3309,
+      "step": 104925
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.025549669495101e-05,
+      "loss": 0.0356,
+      "step": 104930
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.025502789367587e-05,
+      "loss": 0.0493,
+      "step": 104935
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.0254559092400733e-05,
+      "loss": 0.0227,
+      "step": 104940
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.0254090291125593e-05,
+      "loss": 0.0385,
+      "step": 104945
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.0253621489850453e-05,
+      "loss": 0.0224,
+      "step": 104950
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.0253152688575315e-05,
+      "loss": 0.1233,
+      "step": 104955
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.0252683887300175e-05,
+      "loss": 0.0783,
+      "step": 104960
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.0252215086025035e-05,
+      "loss": 0.1145,
+      "step": 104965
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.0251746284749895e-05,
+      "loss": 0.1291,
+      "step": 104970
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.0251277483474755e-05,
+      "loss": 0.2316,
+      "step": 104975
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.0250808682199616e-05,
+      "loss": 0.0912,
+      "step": 104980
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.0250339880924476e-05,
+      "loss": 0.0083,
+      "step": 104985
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.0249871079649338e-05,
+      "loss": 0.0241,
+      "step": 104990
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.02494022783742e-05,
+      "loss": 0.0891,
+      "step": 104995
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.024893347709906e-05,
+      "loss": 0.0305,
+      "step": 105000
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.024846467582392e-05,
+      "loss": 0.0235,
+      "step": 105005
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.0247995874548779e-05,
+      "loss": 0.0277,
+      "step": 105010
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.0247527073273639e-05,
+      "loss": 0.1312,
+      "step": 105015
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.02470582719985e-05,
+      "loss": 0.2417,
+      "step": 105020
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.024658947072336e-05,
+      "loss": 0.2572,
+      "step": 105025
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.024612066944822e-05,
+      "loss": 0.0608,
+      "step": 105030
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.0245651868173084e-05,
+      "loss": 0.011,
+      "step": 105035
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.0245183066897944e-05,
+      "loss": 0.0121,
+      "step": 105040
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.0244714265622804e-05,
+      "loss": 0.0739,
+      "step": 105045
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.0244245464347664e-05,
+      "loss": 0.0278,
+      "step": 105050
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.0243776663072524e-05,
+      "loss": 0.0859,
+      "step": 105055
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.0243307861797385e-05,
+      "loss": 0.1252,
+      "step": 105060
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.0242839060522245e-05,
+      "loss": 0.0627,
+      "step": 105065
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.0242370259247105e-05,
+      "loss": 0.2326,
+      "step": 105070
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.0241901457971965e-05,
+      "loss": 0.2487,
+      "step": 105075
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.0241432656696828e-05,
+      "loss": 0.0483,
+      "step": 105080
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.0240963855421688e-05,
+      "loss": 0.0036,
+      "step": 105085
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.0240495054146548e-05,
+      "loss": 0.0322,
+      "step": 105090
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.0240026252871408e-05,
+      "loss": 0.0499,
+      "step": 105095
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.023955745159627e-05,
+      "loss": 0.0546,
+      "step": 105100
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.023908865032113e-05,
+      "loss": 0.1015,
+      "step": 105105
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.023861984904599e-05,
+      "loss": 0.0561,
+      "step": 105110
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.023815104777085e-05,
+      "loss": 0.0726,
+      "step": 105115
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.0237682246495711e-05,
+      "loss": 0.1082,
+      "step": 105120
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.0237213445220573e-05,
+      "loss": 0.2577,
+      "step": 105125
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.0236744643945433e-05,
+      "loss": 0.1055,
+      "step": 105130
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.0236275842670293e-05,
+      "loss": 0.0311,
+      "step": 105135
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.0235807041395154e-05,
+      "loss": 0.0067,
+      "step": 105140
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.0235338240120014e-05,
+      "loss": 0.0624,
+      "step": 105145
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.0234869438844874e-05,
+      "loss": 0.0789,
+      "step": 105150
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.0234400637569734e-05,
+      "loss": 0.0729,
+      "step": 105155
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.0233931836294596e-05,
+      "loss": 0.0802,
+      "step": 105160
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.0233463035019456e-05,
+      "loss": 0.0696,
+      "step": 105165
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.0232994233744316e-05,
+      "loss": 0.1909,
+      "step": 105170
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.0232525432469177e-05,
+      "loss": 0.1665,
+      "step": 105175
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.0232056631194039e-05,
+      "loss": 0.0849,
+      "step": 105180
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.0231587829918899e-05,
+      "loss": 0.0045,
+      "step": 105185
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.0231119028643759e-05,
+      "loss": 0.0452,
+      "step": 105190
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.0230650227368619e-05,
+      "loss": 0.0604,
+      "step": 105195
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.023018142609348e-05,
+      "loss": 0.0435,
+      "step": 105200
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.022971262481834e-05,
+      "loss": 0.0534,
+      "step": 105205
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.02292438235432e-05,
+      "loss": 0.0723,
+      "step": 105210
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.022877502226806e-05,
+      "loss": 0.2139,
+      "step": 105215
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.0228306220992923e-05,
+      "loss": 0.1252,
+      "step": 105220
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.0227837419717783e-05,
+      "loss": 0.2006,
+      "step": 105225
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.0227368618442643e-05,
+      "loss": 0.0723,
+      "step": 105230
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.0226899817167503e-05,
+      "loss": 0.0229,
+      "step": 105235
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.0226431015892365e-05,
+      "loss": 0.0201,
+      "step": 105240
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.0225962214617225e-05,
+      "loss": 0.0269,
+      "step": 105245
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.0225493413342085e-05,
+      "loss": 0.0245,
+      "step": 105250
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.0225024612066945e-05,
+      "loss": 0.1189,
+      "step": 105255
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.0224555810791805e-05,
+      "loss": 0.0462,
+      "step": 105260
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.0224087009516668e-05,
+      "loss": 0.0845,
+      "step": 105265
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.0223618208241528e-05,
+      "loss": 0.1731,
+      "step": 105270
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.0223149406966388e-05,
+      "loss": 0.2558,
+      "step": 105275
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.022268060569125e-05,
+      "loss": 0.0636,
+      "step": 105280
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.022221180441611e-05,
+      "loss": 0.0151,
+      "step": 105285
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.022174300314097e-05,
+      "loss": 0.046,
+      "step": 105290
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.022127420186583e-05,
+      "loss": 0.0598,
+      "step": 105295
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.022080540059069e-05,
+      "loss": 0.0364,
+      "step": 105300
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.022033659931555e-05,
+      "loss": 0.0642,
+      "step": 105305
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.021986779804041e-05,
+      "loss": 0.0466,
+      "step": 105310
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.0219398996765272e-05,
+      "loss": 0.1549,
+      "step": 105315
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.0218930195490134e-05,
+      "loss": 0.1041,
+      "step": 105320
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.0218461394214994e-05,
+      "loss": 0.1938,
+      "step": 105325
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.0217992592939854e-05,
+      "loss": 0.0585,
+      "step": 105330
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.0217523791664714e-05,
+      "loss": 0.0217,
+      "step": 105335
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.0217054990389574e-05,
+      "loss": 0.0314,
+      "step": 105340
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.0216586189114435e-05,
+      "loss": 0.0546,
+      "step": 105345
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.0216117387839295e-05,
+      "loss": 0.0785,
+      "step": 105350
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.0215648586564155e-05,
+      "loss": 0.0318,
+      "step": 105355
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.0215179785289019e-05,
+      "loss": 0.1693,
+      "step": 105360
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.0214710984013878e-05,
+      "loss": 0.0898,
+      "step": 105365
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.0214242182738738e-05,
+      "loss": 0.1602,
+      "step": 105370
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.0213773381463598e-05,
+      "loss": 0.2031,
+      "step": 105375
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.0213304580188458e-05,
+      "loss": 0.0841,
+      "step": 105380
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.021283577891332e-05,
+      "loss": 0.0105,
+      "step": 105385
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.021236697763818e-05,
+      "loss": 0.0378,
+      "step": 105390
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.021189817636304e-05,
+      "loss": 0.0579,
+      "step": 105395
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.02114293750879e-05,
+      "loss": 0.0519,
+      "step": 105400
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.0210960573812763e-05,
+      "loss": 0.0504,
+      "step": 105405
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.0210491772537623e-05,
+      "loss": 0.0254,
+      "step": 105410
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.0210022971262483e-05,
+      "loss": 0.1071,
+      "step": 105415
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.0209554169987343e-05,
+      "loss": 0.1872,
+      "step": 105420
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.0209085368712204e-05,
+      "loss": 0.2193,
+      "step": 105425
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.0208616567437064e-05,
+      "loss": 0.0787,
+      "step": 105430
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.0208147766161924e-05,
+      "loss": 0.0144,
+      "step": 105435
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.0207678964886784e-05,
+      "loss": 0.0364,
+      "step": 105440
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.0207210163611644e-05,
+      "loss": 0.0291,
+      "step": 105445
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.0206741362336508e-05,
+      "loss": 0.0328,
+      "step": 105450
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.0206272561061367e-05,
+      "loss": 0.0309,
+      "step": 105455
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.0205803759786227e-05,
+      "loss": 0.0704,
+      "step": 105460
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.0205334958511089e-05,
+      "loss": 0.0925,
+      "step": 105465
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.0204866157235949e-05,
+      "loss": 0.1492,
+      "step": 105470
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.0204397355960809e-05,
+      "loss": 0.2721,
+      "step": 105475
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.0203928554685669e-05,
+      "loss": 0.0629,
+      "step": 105480
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.0203459753410529e-05,
+      "loss": 0.0471,
+      "step": 105485
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.020299095213539e-05,
+      "loss": 0.0209,
+      "step": 105490
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.020252215086025e-05,
+      "loss": 0.28,
+      "step": 105495
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.0202053349585112e-05,
+      "loss": 0.0635,
+      "step": 105500
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.0201584548309974e-05,
+      "loss": 0.0438,
+      "step": 105505
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.0201115747034833e-05,
+      "loss": 0.0661,
+      "step": 105510
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.0200646945759693e-05,
+      "loss": 0.0527,
+      "step": 105515
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.0200178144484553e-05,
+      "loss": 0.1037,
+      "step": 105520
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.0199709343209413e-05,
+      "loss": 0.1355,
+      "step": 105525
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.0199240541934275e-05,
+      "loss": 0.1083,
+      "step": 105530
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.0198771740659135e-05,
+      "loss": 0.0068,
+      "step": 105535
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.0198302939383995e-05,
+      "loss": 0.0216,
+      "step": 105540
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.0197834138108858e-05,
+      "loss": 0.0254,
+      "step": 105545
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0197365336833718e-05,
+      "loss": 0.031,
+      "step": 105550
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0196896535558578e-05,
+      "loss": 0.1026,
+      "step": 105555
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0196427734283438e-05,
+      "loss": 0.0625,
+      "step": 105560
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0195958933008298e-05,
+      "loss": 0.1294,
+      "step": 105565
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.019549013173316e-05,
+      "loss": 0.0863,
+      "step": 105570
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.019502133045802e-05,
+      "loss": 0.3077,
+      "step": 105575
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.019455252918288e-05,
+      "loss": 0.0828,
+      "step": 105580
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.019408372790774e-05,
+      "loss": 0.0357,
+      "step": 105585
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0193614926632603e-05,
+      "loss": 0.0611,
+      "step": 105590
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0193146125357463e-05,
+      "loss": 0.0175,
+      "step": 105595
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0192677324082322e-05,
+      "loss": 0.0586,
+      "step": 105600
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0192208522807182e-05,
+      "loss": 0.0491,
+      "step": 105605
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0191739721532044e-05,
+      "loss": 0.1044,
+      "step": 105610
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0191270920256904e-05,
+      "loss": 0.0846,
+      "step": 105615
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0190802118981764e-05,
+      "loss": 0.0488,
+      "step": 105620
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0190333317706624e-05,
+      "loss": 0.1462,
+      "step": 105625
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0189864516431485e-05,
+      "loss": 0.1076,
+      "step": 105630
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0189395715156345e-05,
+      "loss": 0.0123,
+      "step": 105635
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0188926913881207e-05,
+      "loss": 0.027,
+      "step": 105640
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0188458112606067e-05,
+      "loss": 0.0457,
+      "step": 105645
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0187989311330929e-05,
+      "loss": 0.0454,
+      "step": 105650
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0187520510055788e-05,
+      "loss": 0.0397,
+      "step": 105655
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0187051708780648e-05,
+      "loss": 0.0971,
+      "step": 105660
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0186582907505508e-05,
+      "loss": 0.0496,
+      "step": 105665
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.018611410623037e-05,
+      "loss": 0.2406,
+      "step": 105670
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.018564530495523e-05,
+      "loss": 0.3111,
+      "step": 105675
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.018517650368009e-05,
+      "loss": 0.1134,
+      "step": 105680
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0184707702404951e-05,
+      "loss": 0.0221,
+      "step": 105685
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0184238901129813e-05,
+      "loss": 0.0478,
+      "step": 105690
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0183770099854673e-05,
+      "loss": 0.0464,
+      "step": 105695
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0183301298579533e-05,
+      "loss": 0.0237,
+      "step": 105700
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0182832497304393e-05,
+      "loss": 0.0356,
+      "step": 105705
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0182363696029255e-05,
+      "loss": 0.0724,
+      "step": 105710
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0181894894754114e-05,
+      "loss": 0.0804,
+      "step": 105715
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0181426093478974e-05,
+      "loss": 0.1141,
+      "step": 105720
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0180957292203834e-05,
+      "loss": 0.1757,
+      "step": 105725
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0180488490928698e-05,
+      "loss": 0.0447,
+      "step": 105730
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0180019689653558e-05,
+      "loss": 0.0109,
+      "step": 105735
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0179550888378418e-05,
+      "loss": 0.0557,
+      "step": 105740
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0179082087103277e-05,
+      "loss": 0.0157,
+      "step": 105745
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0178613285828139e-05,
+      "loss": 0.0685,
+      "step": 105750
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0178144484552999e-05,
+      "loss": 0.0675,
+      "step": 105755
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0177675683277859e-05,
+      "loss": 0.0339,
+      "step": 105760
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.0177206882002719e-05,
+      "loss": 0.0616,
+      "step": 105765
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.0176738080727579e-05,
+      "loss": 0.1308,
+      "step": 105770
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.0176269279452442e-05,
+      "loss": 0.2863,
+      "step": 105775
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.0175800478177302e-05,
+      "loss": 0.0831,
+      "step": 105780
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.0175331676902162e-05,
+      "loss": 0.0167,
+      "step": 105785
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.0174862875627024e-05,
+      "loss": 0.0625,
+      "step": 105790
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.0174394074351884e-05,
+      "loss": 0.0709,
+      "step": 105795
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.0173925273076744e-05,
+      "loss": 0.0417,
+      "step": 105800
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.0173456471801603e-05,
+      "loss": 0.1134,
+      "step": 105805
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.0172987670526463e-05,
+      "loss": 0.067,
+      "step": 105810
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.0172518869251325e-05,
+      "loss": 0.1224,
+      "step": 105815
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.0172050067976185e-05,
+      "loss": 0.0973,
+      "step": 105820
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.0171581266701047e-05,
+      "loss": 0.2426,
+      "step": 105825
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.0171112465425908e-05,
+      "loss": 0.0653,
+      "step": 105830
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.0170643664150768e-05,
+      "loss": 0.005,
+      "step": 105835
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.0170174862875628e-05,
+      "loss": 0.0369,
+      "step": 105840
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.0169706061600488e-05,
+      "loss": 0.0396,
+      "step": 105845
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.0169237260325348e-05,
+      "loss": 0.0607,
+      "step": 105850
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.016876845905021e-05,
+      "loss": 0.0629,
+      "step": 105855
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.016829965777507e-05,
+      "loss": 0.1218,
+      "step": 105860
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.016783085649993e-05,
+      "loss": 0.1445,
+      "step": 105865
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.0167362055224793e-05,
+      "loss": 0.1589,
+      "step": 105870
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.0166893253949653e-05,
+      "loss": 0.2263,
+      "step": 105875
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.0166424452674513e-05,
+      "loss": 0.0922,
+      "step": 105880
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.0165955651399373e-05,
+      "loss": 0.0472,
+      "step": 105885
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.0165486850124232e-05,
+      "loss": 0.0172,
+      "step": 105890
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.0165018048849094e-05,
+      "loss": 0.0513,
+      "step": 105895
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.0164549247573954e-05,
+      "loss": 0.0624,
+      "step": 105900
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.0164080446298814e-05,
+      "loss": 0.0115,
+      "step": 105905
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.0163611645023674e-05,
+      "loss": 0.0551,
+      "step": 105910
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.0163142843748537e-05,
+      "loss": 0.0976,
+      "step": 105915
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.0162674042473397e-05,
+      "loss": 0.1354,
+      "step": 105920
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.0162205241198257e-05,
+      "loss": 0.2303,
+      "step": 105925
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.0161736439923117e-05,
+      "loss": 0.0869,
+      "step": 105930
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.0161267638647979e-05,
+      "loss": 0.0097,
+      "step": 105935
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.0160798837372839e-05,
+      "loss": 0.0077,
+      "step": 105940
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.0160330036097699e-05,
+      "loss": 0.0217,
+      "step": 105945
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.0159861234822558e-05,
+      "loss": 0.0444,
+      "step": 105950
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.0159392433547418e-05,
+      "loss": 0.0707,
+      "step": 105955
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.015892363227228e-05,
+      "loss": 0.0818,
+      "step": 105960
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.0158454830997142e-05,
+      "loss": 0.0797,
+      "step": 105965
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.0157986029722002e-05,
+      "loss": 0.119,
+      "step": 105970
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.0157517228446863e-05,
+      "loss": 0.4182,
+      "step": 105975
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.0157048427171723e-05,
+      "loss": 0.116,
+      "step": 105980
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.0156579625896583e-05,
+      "loss": 0.0175,
+      "step": 105985
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.0156110824621443e-05,
+      "loss": 0.0231,
+      "step": 105990
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.0155642023346303e-05,
+      "loss": 0.063,
+      "step": 105995
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.0155173222071165e-05,
+      "loss": 0.0556,
+      "step": 106000
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.0154704420796025e-05,
+      "loss": 0.0445,
+      "step": 106005
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.0154235619520886e-05,
+      "loss": 0.0489,
+      "step": 106010
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.0153766818245748e-05,
+      "loss": 0.1033,
+      "step": 106015
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.0153298016970608e-05,
+      "loss": 0.1242,
+      "step": 106020
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.0152829215695468e-05,
+      "loss": 0.2584,
+      "step": 106025
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.0152360414420328e-05,
+      "loss": 0.0685,
+      "step": 106030
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.0151891613145187e-05,
+      "loss": 0.0105,
+      "step": 106035
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.0151422811870049e-05,
+      "loss": 0.0653,
+      "step": 106040
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.0150954010594909e-05,
+      "loss": 0.0686,
+      "step": 106045
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.0150485209319769e-05,
+      "loss": 0.0609,
+      "step": 106050
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.0150016408044632e-05,
+      "loss": 0.0506,
+      "step": 106055
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.0149547606769492e-05,
+      "loss": 0.1312,
+      "step": 106060
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.0149078805494352e-05,
+      "loss": 0.0669,
+      "step": 106065
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.0148610004219212e-05,
+      "loss": 0.1358,
+      "step": 106070
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.0148141202944074e-05,
+      "loss": 0.2325,
+      "step": 106075
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.0147672401668934e-05,
+      "loss": 0.0826,
+      "step": 106080
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.0147203600393794e-05,
+      "loss": 0.0157,
+      "step": 106085
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.0146734799118654e-05,
+      "loss": 0.036,
+      "step": 106090
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.0146265997843513e-05,
+      "loss": 0.0389,
+      "step": 106095
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.0145797196568377e-05,
+      "loss": 0.0506,
+      "step": 106100
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.0145328395293237e-05,
+      "loss": 0.0532,
+      "step": 106105
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.0144859594018097e-05,
+      "loss": 0.0792,
+      "step": 106110
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.0144390792742958e-05,
+      "loss": 0.1329,
+      "step": 106115
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.0143921991467818e-05,
+      "loss": 0.1238,
+      "step": 106120
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.0143453190192678e-05,
+      "loss": 0.1671,
+      "step": 106125
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.0142984388917538e-05,
+      "loss": 0.0674,
+      "step": 106130
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.0142515587642398e-05,
+      "loss": 0.0211,
+      "step": 106135
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.014204678636726e-05,
+      "loss": 0.0481,
+      "step": 106140
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.014157798509212e-05,
+      "loss": 0.0719,
+      "step": 106145
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.0141109183816981e-05,
+      "loss": 0.0752,
+      "step": 106150
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.0140640382541843e-05,
+      "loss": 0.0941,
+      "step": 106155
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.0140171581266703e-05,
+      "loss": 0.0856,
+      "step": 106160
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.0139702779991563e-05,
+      "loss": 0.0926,
+      "step": 106165
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.0139233978716423e-05,
+      "loss": 0.1223,
+      "step": 106170
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.0138765177441283e-05,
+      "loss": 0.2743,
+      "step": 106175
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.0138296376166144e-05,
+      "loss": 0.075,
+      "step": 106180
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.0137827574891004e-05,
+      "loss": 0.0374,
+      "step": 106185
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.0137358773615864e-05,
+      "loss": 0.0322,
+      "step": 106190
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.0136889972340727e-05,
+      "loss": 0.1074,
+      "step": 106195
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.0136421171065587e-05,
+      "loss": 0.0267,
+      "step": 106200
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.0135952369790447e-05,
+      "loss": 0.0837,
+      "step": 106205
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.0135483568515307e-05,
+      "loss": 0.096,
+      "step": 106210
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.0135014767240167e-05,
+      "loss": 0.118,
+      "step": 106215
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.0134545965965029e-05,
+      "loss": 0.3036,
+      "step": 106220
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.0134077164689889e-05,
+      "loss": 0.1294,
+      "step": 106225
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.0133608363414749e-05,
+      "loss": 0.0797,
+      "step": 106230
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.0133139562139609e-05,
+      "loss": 0.0282,
+      "step": 106235
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.0132670760864472e-05,
+      "loss": 0.0256,
+      "step": 106240
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.0132201959589332e-05,
+      "loss": 0.0425,
+      "step": 106245
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.0131733158314192e-05,
+      "loss": 0.0414,
+      "step": 106250
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.0131264357039052e-05,
+      "loss": 0.1178,
+      "step": 106255
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.0130795555763913e-05,
+      "loss": 0.0499,
+      "step": 106260
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.0130326754488773e-05,
+      "loss": 0.0522,
+      "step": 106265
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.0129857953213633e-05,
+      "loss": 0.0924,
+      "step": 106270
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.0129389151938493e-05,
+      "loss": 0.2684,
+      "step": 106275
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.0128920350663353e-05,
+      "loss": 0.0866,
+      "step": 106280
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.0128451549388215e-05,
+      "loss": 0.0249,
+      "step": 106285
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.0127982748113076e-05,
+      "loss": 0.0251,
+      "step": 106290
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.0127513946837936e-05,
+      "loss": 0.0388,
+      "step": 106295
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.0127045145562798e-05,
+      "loss": 0.0371,
+      "step": 106300
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.0126576344287658e-05,
+      "loss": 0.0321,
+      "step": 106305
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.0126107543012518e-05,
+      "loss": 0.0819,
+      "step": 106310
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.0125638741737378e-05,
+      "loss": 0.0911,
+      "step": 106315
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.0125169940462238e-05,
+      "loss": 0.0894,
+      "step": 106320
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.01247011391871e-05,
+      "loss": 0.1787,
+      "step": 106325
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.012423233791196e-05,
+      "loss": 0.0633,
+      "step": 106330
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.012376353663682e-05,
+      "loss": 0.0315,
+      "step": 106335
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.0123294735361682e-05,
+      "loss": 0.0292,
+      "step": 106340
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.0122825934086542e-05,
+      "loss": 0.0777,
+      "step": 106345
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.0122357132811402e-05,
+      "loss": 0.0588,
+      "step": 106350
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.0121888331536262e-05,
+      "loss": 0.0551,
+      "step": 106355
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.0121419530261122e-05,
+      "loss": 0.0748,
+      "step": 106360
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.0120950728985984e-05,
+      "loss": 0.0935,
+      "step": 106365
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.0120481927710844e-05,
+      "loss": 0.135,
+      "step": 106370
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.0120013126435704e-05,
+      "loss": 0.3771,
+      "step": 106375
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.0119544325160567e-05,
+      "loss": 0.0698,
+      "step": 106380
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.0119075523885427e-05,
+      "loss": 0.0176,
+      "step": 106385
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.0118606722610287e-05,
+      "loss": 0.0316,
+      "step": 106390
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.0118137921335147e-05,
+      "loss": 0.062,
+      "step": 106395
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.0117669120060007e-05,
+      "loss": 0.0531,
+      "step": 106400
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.0117200318784868e-05,
+      "loss": 0.0542,
+      "step": 106405
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.0116731517509728e-05,
+      "loss": 0.0868,
+      "step": 106410
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.0116262716234588e-05,
+      "loss": 0.1046,
+      "step": 106415
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.0115793914959448e-05,
+      "loss": 0.1983,
+      "step": 106420
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.0115325113684311e-05,
+      "loss": 0.3316,
+      "step": 106425
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.0114856312409171e-05,
+      "loss": 0.0411,
+      "step": 106430
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.0114387511134031e-05,
+      "loss": 0.0243,
+      "step": 106435
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.0113918709858891e-05,
+      "loss": 0.0512,
+      "step": 106440
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.0113449908583753e-05,
+      "loss": 0.0213,
+      "step": 106445
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.0112981107308613e-05,
+      "loss": 0.0605,
+      "step": 106450
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.0112512306033473e-05,
+      "loss": 0.0454,
+      "step": 106455
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.0112043504758333e-05,
+      "loss": 0.0441,
+      "step": 106460
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.0111574703483194e-05,
+      "loss": 0.09,
+      "step": 106465
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.0111105902208054e-05,
+      "loss": 0.093,
+      "step": 106470
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.0110637100932916e-05,
+      "loss": 0.1816,
+      "step": 106475
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.0110168299657776e-05,
+      "loss": 0.0599,
+      "step": 106480
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.0109699498382637e-05,
+      "loss": 0.0688,
+      "step": 106485
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.0109230697107497e-05,
+      "loss": 0.02,
+      "step": 106490
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.0108761895832357e-05,
+      "loss": 0.0035,
+      "step": 106495
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.0108293094557217e-05,
+      "loss": 0.0438,
+      "step": 106500
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.0107824293282079e-05,
+      "loss": 0.0664,
+      "step": 106505
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.0107355492006939e-05,
+      "loss": 0.074,
+      "step": 106510
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.0106886690731799e-05,
+      "loss": 0.108,
+      "step": 106515
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.010641788945666e-05,
+      "loss": 0.2912,
+      "step": 106520
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.0105949088181522e-05,
+      "loss": 0.169,
+      "step": 106525
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.0105480286906382e-05,
+      "loss": 0.0864,
+      "step": 106530
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.0105011485631242e-05,
+      "loss": 0.0345,
+      "step": 106535
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.0104542684356102e-05,
+      "loss": 0.0443,
+      "step": 106540
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.0104073883080963e-05,
+      "loss": 0.0409,
+      "step": 106545
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.0103605081805823e-05,
+      "loss": 0.0509,
+      "step": 106550
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.0103136280530683e-05,
+      "loss": 0.0303,
+      "step": 106555
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.0102667479255543e-05,
+      "loss": 0.0752,
+      "step": 106560
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.0102198677980407e-05,
+      "loss": 0.0816,
+      "step": 106565
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.0101729876705266e-05,
+      "loss": 0.2227,
+      "step": 106570
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.0101261075430126e-05,
+      "loss": 0.2331,
+      "step": 106575
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.0100792274154986e-05,
+      "loss": 0.0779,
+      "step": 106580
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.0100323472879848e-05,
+      "loss": 0.0239,
+      "step": 106585
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.0099854671604708e-05,
+      "loss": 0.0714,
+      "step": 106590
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.0099385870329568e-05,
+      "loss": 0.0674,
+      "step": 106595
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.0098917069054428e-05,
+      "loss": 0.0846,
+      "step": 106600
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.0098448267779288e-05,
+      "loss": 0.0189,
+      "step": 106605
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.009797946650415e-05,
+      "loss": 0.1387,
+      "step": 106610
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.0097510665229011e-05,
+      "loss": 0.0821,
+      "step": 106615
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0097041863953871e-05,
+      "loss": 0.1273,
+      "step": 106620
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0096573062678733e-05,
+      "loss": 0.2383,
+      "step": 106625
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0096104261403592e-05,
+      "loss": 0.076,
+      "step": 106630
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0095635460128452e-05,
+      "loss": 0.0271,
+      "step": 106635
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0095166658853312e-05,
+      "loss": 0.0509,
+      "step": 106640
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0094697857578172e-05,
+      "loss": 0.0879,
+      "step": 106645
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0094229056303034e-05,
+      "loss": 0.0831,
+      "step": 106650
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0093760255027894e-05,
+      "loss": 0.0384,
+      "step": 106655
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0093291453752755e-05,
+      "loss": 0.0639,
+      "step": 106660
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0092822652477617e-05,
+      "loss": 0.0972,
+      "step": 106665
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0092353851202477e-05,
+      "loss": 0.1232,
+      "step": 106670
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0091885049927337e-05,
+      "loss": 0.2165,
+      "step": 106675
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0091416248652197e-05,
+      "loss": 0.0817,
+      "step": 106680
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0090947447377057e-05,
+      "loss": 0.0023,
+      "step": 106685
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0090478646101918e-05,
+      "loss": 0.0271,
+      "step": 106690
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0090009844826778e-05,
+      "loss": 0.0288,
+      "step": 106695
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0089541043551638e-05,
+      "loss": 0.0329,
+      "step": 106700
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0089072242276502e-05,
+      "loss": 0.0796,
+      "step": 106705
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0088603441001362e-05,
+      "loss": 0.0733,
+      "step": 106710
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0088134639726221e-05,
+      "loss": 0.0385,
+      "step": 106715
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0087665838451081e-05,
+      "loss": 0.1423,
+      "step": 106720
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0087197037175941e-05,
+      "loss": 0.3247,
+      "step": 106725
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0086728235900803e-05,
+      "loss": 0.1012,
+      "step": 106730
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0086259434625663e-05,
+      "loss": 0.0257,
+      "step": 106735
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0085790633350523e-05,
+      "loss": 0.023,
+      "step": 106740
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0085321832075383e-05,
+      "loss": 0.0512,
+      "step": 106745
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0084853030800246e-05,
+      "loss": 0.0609,
+      "step": 106750
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0084384229525106e-05,
+      "loss": 0.0121,
+      "step": 106755
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0083915428249966e-05,
+      "loss": 0.0427,
+      "step": 106760
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0083446626974826e-05,
+      "loss": 0.142,
+      "step": 106765
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0082977825699688e-05,
+      "loss": 0.0849,
+      "step": 106770
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0082509024424547e-05,
+      "loss": 0.3115,
+      "step": 106775
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0082040223149407e-05,
+      "loss": 0.0612,
+      "step": 106780
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0081571421874267e-05,
+      "loss": 0.0387,
+      "step": 106785
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0081102620599127e-05,
+      "loss": 0.0611,
+      "step": 106790
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0080633819323989e-05,
+      "loss": 0.051,
+      "step": 106795
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.008016501804885e-05,
+      "loss": 0.0848,
+      "step": 106800
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.007969621677371e-05,
+      "loss": 0.0298,
+      "step": 106805
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0079227415498572e-05,
+      "loss": 0.1119,
+      "step": 106810
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0078758614223432e-05,
+      "loss": 0.0883,
+      "step": 106815
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0078289812948292e-05,
+      "loss": 0.0992,
+      "step": 106820
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0077821011673152e-05,
+      "loss": 0.1845,
+      "step": 106825
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0077352210398012e-05,
+      "loss": 0.0822,
+      "step": 106830
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.0076883409122873e-05,
+      "loss": 0.0048,
+      "step": 106835
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.0076414607847733e-05,
+      "loss": 0.0381,
+      "step": 106840
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.0075945806572595e-05,
+      "loss": 0.0755,
+      "step": 106845
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.0075477005297457e-05,
+      "loss": 0.0941,
+      "step": 106850
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.0075008204022317e-05,
+      "loss": 0.0333,
+      "step": 106855
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.0074539402747177e-05,
+      "loss": 0.0741,
+      "step": 106860
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.0074070601472036e-05,
+      "loss": 0.1271,
+      "step": 106865
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.0073601800196896e-05,
+      "loss": 0.1023,
+      "step": 106870
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.0073132998921758e-05,
+      "loss": 0.378,
+      "step": 106875
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.0072664197646618e-05,
+      "loss": 0.0827,
+      "step": 106880
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.0072195396371478e-05,
+      "loss": 0.0111,
+      "step": 106885
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.0071726595096341e-05,
+      "loss": 0.0459,
+      "step": 106890
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.0071257793821201e-05,
+      "loss": 0.0547,
+      "step": 106895
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.0070788992546061e-05,
+      "loss": 0.0094,
+      "step": 106900
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.0070320191270921e-05,
+      "loss": 0.0941,
+      "step": 106905
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.0069851389995781e-05,
+      "loss": 0.1168,
+      "step": 106910
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.0069382588720643e-05,
+      "loss": 0.081,
+      "step": 106915
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.0068913787445502e-05,
+      "loss": 0.1258,
+      "step": 106920
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.0068444986170362e-05,
+      "loss": 0.3271,
+      "step": 106925
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.0067976184895222e-05,
+      "loss": 0.1,
+      "step": 106930
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.0067507383620084e-05,
+      "loss": 0.0084,
+      "step": 106935
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.0067038582344946e-05,
+      "loss": 0.0214,
+      "step": 106940
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.0066569781069806e-05,
+      "loss": 0.0321,
+      "step": 106945
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.0066100979794665e-05,
+      "loss": 0.0446,
+      "step": 106950
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.0065632178519527e-05,
+      "loss": 0.0823,
+      "step": 106955
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.0065163377244387e-05,
+      "loss": 0.0646,
+      "step": 106960
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.0064694575969247e-05,
+      "loss": 0.0987,
+      "step": 106965
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.0064225774694107e-05,
+      "loss": 0.1067,
+      "step": 106970
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.0063756973418969e-05,
+      "loss": 0.3464,
+      "step": 106975
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.0063288172143828e-05,
+      "loss": 0.0552,
+      "step": 106980
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.006281937086869e-05,
+      "loss": 0.032,
+      "step": 106985
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.006235056959355e-05,
+      "loss": 0.0183,
+      "step": 106990
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.0061881768318412e-05,
+      "loss": 0.0422,
+      "step": 106995
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.0061412967043272e-05,
+      "loss": 0.0519,
+      "step": 107000
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.0060944165768132e-05,
+      "loss": 0.0515,
+      "step": 107005
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.0060475364492991e-05,
+      "loss": 0.0477,
+      "step": 107010
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.0060006563217853e-05,
+      "loss": 0.0601,
+      "step": 107015
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.0059537761942713e-05,
+      "loss": 0.1691,
+      "step": 107020
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.0059068960667573e-05,
+      "loss": 0.1457,
+      "step": 107025
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.0058600159392435e-05,
+      "loss": 0.0356,
+      "step": 107030
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.0058131358117296e-05,
+      "loss": 0.0334,
+      "step": 107035
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.0057662556842156e-05,
+      "loss": 0.01,
+      "step": 107040
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.0057193755567016e-05,
+      "loss": 0.0349,
+      "step": 107045
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.0056724954291876e-05,
+      "loss": 0.035,
+      "step": 107050
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.0056256153016738e-05,
+      "loss": 0.018,
+      "step": 107055
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.0055787351741598e-05,
+      "loss": 0.0728,
+      "step": 107060
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.0055318550466457e-05,
+      "loss": 0.0912,
+      "step": 107065
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.0054849749191317e-05,
+      "loss": 0.1536,
+      "step": 107070
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.005438094791618e-05,
+      "loss": 0.1777,
+      "step": 107075
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.005391214664104e-05,
+      "loss": 0.0382,
+      "step": 107080
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.00534433453659e-05,
+      "loss": 0.019,
+      "step": 107085
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.005297454409076e-05,
+      "loss": 0.0254,
+      "step": 107090
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.0052505742815622e-05,
+      "loss": 0.023,
+      "step": 107095
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.0052036941540482e-05,
+      "loss": 0.0637,
+      "step": 107100
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.0051568140265342e-05,
+      "loss": 0.1354,
+      "step": 107105
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.0051099338990202e-05,
+      "loss": 0.0935,
+      "step": 107110
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.0050630537715062e-05,
+      "loss": 0.0908,
+      "step": 107115
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.0050161736439924e-05,
+      "loss": 0.1455,
+      "step": 107120
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.0049692935164785e-05,
+      "loss": 0.2236,
+      "step": 107125
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.0049224133889645e-05,
+      "loss": 0.0443,
+      "step": 107130
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.0048755332614507e-05,
+      "loss": 0.0502,
+      "step": 107135
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.0048286531339367e-05,
+      "loss": 0.0426,
+      "step": 107140
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.0047817730064227e-05,
+      "loss": 0.0321,
+      "step": 107145
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.0047348928789087e-05,
+      "loss": 0.1014,
+      "step": 107150
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.0046880127513946e-05,
+      "loss": 0.2579,
+      "step": 107155
+    },
+    {
+      "epoch": 5.0,
+      "eval_cer": 0.009420852950587375,
+      "eval_loss": 0.05503755807876587,
+      "eval_runtime": 386.7254,
+      "eval_samples_per_second": 49.26,
+      "eval_steps_per_second": 12.316,
+      "eval_wer": 0.07860459848227433,
+      "step": 107155
+    }
+  ],
+  "max_steps": 214310,
+  "num_train_epochs": 10,
+  "total_flos": 1.2873370890866225e+19,
+  "trial_name": null,
+  "trial_params": null
+}