diff --git "a/checkpoint-48000/trainer_state.json" "b/checkpoint-48000/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-48000/trainer_state.json"
@@ -0,0 +1,57625 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.536,
+  "eval_steps": 500,
+  "global_step": 48000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2800000000000001e-09,
+      "loss": 2.0804,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.400000000000001e-09,
+      "loss": 2.0692,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2800000000000002e-08,
+      "loss": 2.1047,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9200000000000003e-08,
+      "loss": 2.0927,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5600000000000004e-08,
+      "loss": 2.0802,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2e-08,
+      "loss": 2.0879,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8400000000000006e-08,
+      "loss": 2.0852,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.48e-08,
+      "loss": 2.0654,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.120000000000001e-08,
+      "loss": 2.1006,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.7600000000000006e-08,
+      "loss": 2.0717,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.4e-08,
+      "loss": 2.0658,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.040000000000001e-08,
+      "loss": 2.036,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.680000000000001e-08,
+      "loss": 2.0428,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.32e-08,
+      "loss": 2.0419,
+      "step": 65
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.96e-08,
+      "loss": 2.0317,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.6e-08,
+      "loss": 2.0456,
+      "step": 75
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0240000000000002e-07,
+      "loss": 2.0136,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0880000000000002e-07,
+      "loss": 2.0241,
+      "step": 85
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1520000000000001e-07,
+      "loss": 2.0083,
+      "step": 90
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.216e-07,
+      "loss": 2.0187,
+      "step": 95
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.28e-07,
+      "loss": 2.0067,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.344e-07,
+      "loss": 2.0019,
+      "step": 105
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4080000000000002e-07,
+      "loss": 1.9899,
+      "step": 110
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4720000000000002e-07,
+      "loss": 1.9648,
+      "step": 115
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5360000000000003e-07,
+      "loss": 1.9652,
+      "step": 120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6e-07,
+      "loss": 1.9284,
+      "step": 125
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.664e-07,
+      "loss": 1.9491,
+      "step": 130
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.728e-07,
+      "loss": 1.9442,
+      "step": 135
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.792e-07,
+      "loss": 1.9239,
+      "step": 140
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8560000000000002e-07,
+      "loss": 1.9454,
+      "step": 145
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.92e-07,
+      "loss": 1.914,
+      "step": 150
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.984e-07,
+      "loss": 1.8963,
+      "step": 155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0480000000000003e-07,
+      "loss": 1.8846,
+      "step": 160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.112e-07,
+      "loss": 1.8761,
+      "step": 165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1760000000000004e-07,
+      "loss": 1.8882,
+      "step": 170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2400000000000002e-07,
+      "loss": 1.8794,
+      "step": 175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3040000000000002e-07,
+      "loss": 1.8585,
+      "step": 180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.368e-07,
+      "loss": 1.8616,
+      "step": 185
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.432e-07,
+      "loss": 1.8567,
+      "step": 190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.496e-07,
+      "loss": 1.8642,
+      "step": 195
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.56e-07,
+      "loss": 1.8508,
+      "step": 200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.624e-07,
+      "loss": 1.8372,
+      "step": 205
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.688e-07,
+      "loss": 1.8225,
+      "step": 210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7520000000000003e-07,
+      "loss": 1.8232,
+      "step": 215
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8160000000000003e-07,
+      "loss": 1.8183,
+      "step": 220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8800000000000004e-07,
+      "loss": 1.8174,
+      "step": 225
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9440000000000004e-07,
+      "loss": 1.8196,
+      "step": 230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.008e-07,
+      "loss": 1.8158,
+      "step": 235
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0720000000000005e-07,
+      "loss": 1.7998,
+      "step": 240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.136e-07,
+      "loss": 1.7931,
+      "step": 245
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2e-07,
+      "loss": 1.8028,
+      "step": 250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2640000000000006e-07,
+      "loss": 1.7743,
+      "step": 255
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.328e-07,
+      "loss": 1.8076,
+      "step": 260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.392e-07,
+      "loss": 1.7776,
+      "step": 265
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.456e-07,
+      "loss": 1.788,
+      "step": 270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5200000000000003e-07,
+      "loss": 1.7723,
+      "step": 275
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.584e-07,
+      "loss": 1.7644,
+      "step": 280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6480000000000004e-07,
+      "loss": 1.7607,
+      "step": 285
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7120000000000004e-07,
+      "loss": 1.7548,
+      "step": 290
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7760000000000005e-07,
+      "loss": 1.7537,
+      "step": 295
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.84e-07,
+      "loss": 1.7551,
+      "step": 300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.904e-07,
+      "loss": 1.742,
+      "step": 305
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.968e-07,
+      "loss": 1.7261,
+      "step": 310
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0320000000000006e-07,
+      "loss": 1.7364,
+      "step": 315
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0960000000000007e-07,
+      "loss": 1.7286,
+      "step": 320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.16e-07,
+      "loss": 1.7373,
+      "step": 325
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.224e-07,
+      "loss": 1.7206,
+      "step": 330
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2880000000000003e-07,
+      "loss": 1.6963,
+      "step": 335
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.352000000000001e-07,
+      "loss": 1.7102,
+      "step": 340
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4160000000000004e-07,
+      "loss": 1.7082,
+      "step": 345
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4800000000000004e-07,
+      "loss": 1.6996,
+      "step": 350
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5440000000000005e-07,
+      "loss": 1.7064,
+      "step": 355
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6080000000000005e-07,
+      "loss": 1.6969,
+      "step": 360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.672e-07,
+      "loss": 1.703,
+      "step": 365
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.736e-07,
+      "loss": 1.6827,
+      "step": 370
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.800000000000001e-07,
+      "loss": 1.692,
+      "step": 375
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.864e-07,
+      "loss": 1.6673,
+      "step": 380
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.928e-07,
+      "loss": 1.6776,
+      "step": 385
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.992e-07,
+      "loss": 1.6732,
+      "step": 390
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.056000000000001e-07,
+      "loss": 1.6806,
+      "step": 395
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.12e-07,
+      "loss": 1.6687,
+      "step": 400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.184e-07,
+      "loss": 1.664,
+      "step": 405
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.248e-07,
+      "loss": 1.6637,
+      "step": 410
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.312000000000001e-07,
+      "loss": 1.6457,
+      "step": 415
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.376e-07,
+      "loss": 1.6616,
+      "step": 420
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.44e-07,
+      "loss": 1.6566,
+      "step": 425
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.504000000000001e-07,
+      "loss": 1.6495,
+      "step": 430
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.568e-07,
+      "loss": 1.6377,
+      "step": 435
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.632000000000001e-07,
+      "loss": 1.66,
+      "step": 440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.696e-07,
+      "loss": 1.6432,
+      "step": 445
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.760000000000001e-07,
+      "loss": 1.6198,
+      "step": 450
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.824e-07,
+      "loss": 1.6216,
+      "step": 455
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.888000000000001e-07,
+      "loss": 1.6218,
+      "step": 460
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.952e-07,
+      "loss": 1.6223,
+      "step": 465
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.016e-07,
+      "loss": 1.6177,
+      "step": 470
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.08e-07,
+      "loss": 1.6115,
+      "step": 475
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.144000000000001e-07,
+      "loss": 1.6242,
+      "step": 480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.208000000000001e-07,
+      "loss": 1.6061,
+      "step": 485
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.272e-07,
+      "loss": 1.625,
+      "step": 490
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.336000000000001e-07,
+      "loss": 1.6041,
+      "step": 495
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.4e-07,
+      "loss": 1.5859,
+      "step": 500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.464000000000001e-07,
+      "loss": 1.6088,
+      "step": 505
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.528000000000001e-07,
+      "loss": 1.5875,
+      "step": 510
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.592000000000001e-07,
+      "loss": 1.597,
+      "step": 515
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.656e-07,
+      "loss": 1.5934,
+      "step": 520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.72e-07,
+      "loss": 1.5865,
+      "step": 525
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.784e-07,
+      "loss": 1.5897,
+      "step": 530
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.848000000000001e-07,
+      "loss": 1.5934,
+      "step": 535
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.912e-07,
+      "loss": 1.5699,
+      "step": 540
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.976000000000001e-07,
+      "loss": 1.5822,
+      "step": 545
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.040000000000001e-07,
+      "loss": 1.5747,
+      "step": 550
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.104000000000001e-07,
+      "loss": 1.564,
+      "step": 555
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.168e-07,
+      "loss": 1.5655,
+      "step": 560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.232e-07,
+      "loss": 1.5757,
+      "step": 565
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.296000000000001e-07,
+      "loss": 1.5593,
+      "step": 570
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.36e-07,
+      "loss": 1.5645,
+      "step": 575
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.424000000000001e-07,
+      "loss": 1.5617,
+      "step": 580
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.488e-07,
+      "loss": 1.5675,
+      "step": 585
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.552000000000001e-07,
+      "loss": 1.5632,
+      "step": 590
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.616000000000001e-07,
+      "loss": 1.5554,
+      "step": 595
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.68e-07,
+      "loss": 1.5714,
+      "step": 600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.744e-07,
+      "loss": 1.5601,
+      "step": 605
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.808e-07,
+      "loss": 1.5641,
+      "step": 610
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.872000000000001e-07,
+      "loss": 1.5384,
+      "step": 615
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.936e-07,
+      "loss": 1.5356,
+      "step": 620
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 1.5413,
+      "step": 625
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.064000000000001e-07,
+      "loss": 1.5428,
+      "step": 630
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.128000000000001e-07,
+      "loss": 1.5215,
+      "step": 635
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.192000000000001e-07,
+      "loss": 1.5349,
+      "step": 640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.256e-07,
+      "loss": 1.531,
+      "step": 645
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.32e-07,
+      "loss": 1.5235,
+      "step": 650
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.384000000000001e-07,
+      "loss": 1.5263,
+      "step": 655
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.448e-07,
+      "loss": 1.5258,
+      "step": 660
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.512000000000001e-07,
+      "loss": 1.5198,
+      "step": 665
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.576000000000001e-07,
+      "loss": 1.509,
+      "step": 670
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.640000000000001e-07,
+      "loss": 1.5356,
+      "step": 675
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.704000000000002e-07,
+      "loss": 1.5134,
+      "step": 680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.768e-07,
+      "loss": 1.5173,
+      "step": 685
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.832000000000001e-07,
+      "loss": 1.5056,
+      "step": 690
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.896e-07,
+      "loss": 1.5093,
+      "step": 695
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.960000000000001e-07,
+      "loss": 1.5077,
+      "step": 700
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.024e-07,
+      "loss": 1.5128,
+      "step": 705
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.088000000000001e-07,
+      "loss": 1.5049,
+      "step": 710
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.152000000000001e-07,
+      "loss": 1.4956,
+      "step": 715
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.216000000000001e-07,
+      "loss": 1.5162,
+      "step": 720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.28e-07,
+      "loss": 1.496,
+      "step": 725
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.344e-07,
+      "loss": 1.4865,
+      "step": 730
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.408000000000001e-07,
+      "loss": 1.4901,
+      "step": 735
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.472e-07,
+      "loss": 1.5181,
+      "step": 740
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.536000000000001e-07,
+      "loss": 1.4874,
+      "step": 745
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.600000000000001e-07,
+      "loss": 1.4866,
+      "step": 750
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.664000000000002e-07,
+      "loss": 1.4838,
+      "step": 755
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.728e-07,
+      "loss": 1.48,
+      "step": 760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.792e-07,
+      "loss": 1.4878,
+      "step": 765
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.856e-07,
+      "loss": 1.4835,
+      "step": 770
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.92e-07,
+      "loss": 1.4858,
+      "step": 775
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.984e-07,
+      "loss": 1.4906,
+      "step": 780
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0048e-06,
+      "loss": 1.4854,
+      "step": 785
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0112000000000002e-06,
+      "loss": 1.4715,
+      "step": 790
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0176000000000002e-06,
+      "loss": 1.4845,
+      "step": 795
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.024e-06,
+      "loss": 1.4707,
+      "step": 800
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0304000000000001e-06,
+      "loss": 1.4713,
+      "step": 805
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0368e-06,
+      "loss": 1.4625,
+      "step": 810
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0432e-06,
+      "loss": 1.4681,
+      "step": 815
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0496e-06,
+      "loss": 1.4508,
+      "step": 820
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0560000000000001e-06,
+      "loss": 1.4705,
+      "step": 825
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0624000000000002e-06,
+      "loss": 1.4628,
+      "step": 830
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0688e-06,
+      "loss": 1.4703,
+      "step": 835
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0752e-06,
+      "loss": 1.4615,
+      "step": 840
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0816000000000002e-06,
+      "loss": 1.4573,
+      "step": 845
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.088e-06,
+      "loss": 1.477,
+      "step": 850
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0944e-06,
+      "loss": 1.4623,
+      "step": 855
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1008000000000001e-06,
+      "loss": 1.4557,
+      "step": 860
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1072000000000002e-06,
+      "loss": 1.453,
+      "step": 865
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1136e-06,
+      "loss": 1.4659,
+      "step": 870
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.12e-06,
+      "loss": 1.452,
+      "step": 875
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1264000000000001e-06,
+      "loss": 1.4496,
+      "step": 880
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1328000000000002e-06,
+      "loss": 1.4559,
+      "step": 885
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1392e-06,
+      "loss": 1.4529,
+      "step": 890
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1456e-06,
+      "loss": 1.4408,
+      "step": 895
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1520000000000002e-06,
+      "loss": 1.4372,
+      "step": 900
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1584e-06,
+      "loss": 1.4433,
+      "step": 905
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1648e-06,
+      "loss": 1.4471,
+      "step": 910
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1712000000000001e-06,
+      "loss": 1.4568,
+      "step": 915
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1776000000000002e-06,
+      "loss": 1.4407,
+      "step": 920
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1840000000000002e-06,
+      "loss": 1.4464,
+      "step": 925
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1904e-06,
+      "loss": 1.4253,
+      "step": 930
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1968000000000001e-06,
+      "loss": 1.4401,
+      "step": 935
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.2032e-06,
+      "loss": 1.43,
+      "step": 940
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.2096e-06,
+      "loss": 1.4409,
+      "step": 945
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.216e-06,
+      "loss": 1.429,
+      "step": 950
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.2224000000000001e-06,
+      "loss": 1.4276,
+      "step": 955
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.2288000000000002e-06,
+      "loss": 1.4344,
+      "step": 960
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.2352e-06,
+      "loss": 1.4406,
+      "step": 965
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.2416000000000001e-06,
+      "loss": 1.4254,
+      "step": 970
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.248e-06,
+      "loss": 1.4122,
+      "step": 975
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.2544e-06,
+      "loss": 1.425,
+      "step": 980
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.2608e-06,
+      "loss": 1.4237,
+      "step": 985
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.2672000000000001e-06,
+      "loss": 1.4224,
+      "step": 990
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.2736000000000002e-06,
+      "loss": 1.4227,
+      "step": 995
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.28e-06,
+      "loss": 1.4325,
+      "step": 1000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.2864e-06,
+      "loss": 1.4278,
+      "step": 1005
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.2928000000000001e-06,
+      "loss": 1.4035,
+      "step": 1010
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.2992000000000002e-06,
+      "loss": 1.4231,
+      "step": 1015
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3056000000000003e-06,
+      "loss": 1.4169,
+      "step": 1020
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3120000000000003e-06,
+      "loss": 1.4035,
+      "step": 1025
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3184000000000002e-06,
+      "loss": 1.4094,
+      "step": 1030
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3248e-06,
+      "loss": 1.4097,
+      "step": 1035
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3312e-06,
+      "loss": 1.4261,
+      "step": 1040
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3376e-06,
+      "loss": 1.4113,
+      "step": 1045
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.344e-06,
+      "loss": 1.4034,
+      "step": 1050
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3504e-06,
+      "loss": 1.4096,
+      "step": 1055
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3568e-06,
+      "loss": 1.4106,
+      "step": 1060
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3632000000000001e-06,
+      "loss": 1.4023,
+      "step": 1065
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3696000000000002e-06,
+      "loss": 1.4035,
+      "step": 1070
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.376e-06,
+      "loss": 1.4146,
+      "step": 1075
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3824e-06,
+      "loss": 1.4054,
+      "step": 1080
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3888000000000002e-06,
+      "loss": 1.3996,
+      "step": 1085
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3952000000000002e-06,
+      "loss": 1.4073,
+      "step": 1090
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4016000000000003e-06,
+      "loss": 1.4119,
+      "step": 1095
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4080000000000001e-06,
+      "loss": 1.394,
+      "step": 1100
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4144000000000002e-06,
+      "loss": 1.3976,
+      "step": 1105
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4208000000000002e-06,
+      "loss": 1.4036,
+      "step": 1110
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4272000000000003e-06,
+      "loss": 1.3786,
+      "step": 1115
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4336e-06,
+      "loss": 1.4106,
+      "step": 1120
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.44e-06,
+      "loss": 1.3807,
+      "step": 1125
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4464e-06,
+      "loss": 1.3796,
+      "step": 1130
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4528e-06,
+      "loss": 1.3996,
+      "step": 1135
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4592000000000001e-06,
+      "loss": 1.3989,
+      "step": 1140
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4656e-06,
+      "loss": 1.4022,
+      "step": 1145
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.472e-06,
+      "loss": 1.3845,
+      "step": 1150
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4784000000000001e-06,
+      "loss": 1.3965,
+      "step": 1155
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4848000000000002e-06,
+      "loss": 1.3828,
+      "step": 1160
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4912000000000002e-06,
+      "loss": 1.3883,
+      "step": 1165
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4976e-06,
+      "loss": 1.3855,
+      "step": 1170
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.5040000000000001e-06,
+      "loss": 1.3835,
+      "step": 1175
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.5104000000000002e-06,
+      "loss": 1.3739,
+      "step": 1180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.5168000000000002e-06,
+      "loss": 1.3903,
+      "step": 1185
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.5232000000000003e-06,
+      "loss": 1.3775,
+      "step": 1190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.5296000000000004e-06,
+      "loss": 1.3816,
+      "step": 1195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.536e-06,
+      "loss": 1.3702,
+      "step": 1200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.5424e-06,
+      "loss": 1.3846,
+      "step": 1205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.5488e-06,
+      "loss": 1.3755,
+      "step": 1210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.5552e-06,
+      "loss": 1.382,
+      "step": 1215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.5616e-06,
+      "loss": 1.3678,
+      "step": 1220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.568e-06,
+      "loss": 1.3717,
+      "step": 1225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.5744000000000001e-06,
+      "loss": 1.3731,
+      "step": 1230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.5808000000000002e-06,
+      "loss": 1.3582,
+      "step": 1235
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.5872e-06,
+      "loss": 1.3716,
+      "step": 1240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.5936e-06,
+      "loss": 1.3805,
+      "step": 1245
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 1.3684,
+      "step": 1250
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.6064000000000002e-06,
+      "loss": 1.35,
+      "step": 1255
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.6128000000000003e-06,
+      "loss": 1.3612,
+      "step": 1260
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.6192000000000003e-06,
+      "loss": 1.351,
+      "step": 1265
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.6256000000000002e-06,
+      "loss": 1.3513,
+      "step": 1270
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.6320000000000002e-06,
+      "loss": 1.3741,
+      "step": 1275
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.6384000000000003e-06,
+      "loss": 1.3708,
+      "step": 1280
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.6448e-06,
+      "loss": 1.3671,
+      "step": 1285
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.6512e-06,
+      "loss": 1.3587,
+      "step": 1290
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.6576e-06,
+      "loss": 1.3568,
+      "step": 1295
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.664e-06,
+      "loss": 1.347,
+      "step": 1300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.6704000000000001e-06,
+      "loss": 1.3528,
+      "step": 1305
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.6768000000000002e-06,
+      "loss": 1.351,
+      "step": 1310
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.6832e-06,
+      "loss": 1.3543,
+      "step": 1315
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.6896e-06,
+      "loss": 1.3554,
+      "step": 1320
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.6960000000000002e-06,
+      "loss": 1.3562,
+      "step": 1325
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7024000000000002e-06,
+      "loss": 1.3566,
+      "step": 1330
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7088000000000003e-06,
+      "loss": 1.3526,
+      "step": 1335
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7152000000000001e-06,
+      "loss": 1.3561,
+      "step": 1340
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7216000000000002e-06,
+      "loss": 1.3588,
+      "step": 1345
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7280000000000002e-06,
+      "loss": 1.3519,
+      "step": 1350
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7344000000000003e-06,
+      "loss": 1.3386,
+      "step": 1355
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7408000000000003e-06,
+      "loss": 1.3567,
+      "step": 1360
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7472e-06,
+      "loss": 1.3535,
+      "step": 1365
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7536e-06,
+      "loss": 1.3189,
+      "step": 1370
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.76e-06,
+      "loss": 1.3458,
+      "step": 1375
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7664000000000001e-06,
+      "loss": 1.3399,
+      "step": 1380
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7728e-06,
+      "loss": 1.3406,
+      "step": 1385
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7792e-06,
+      "loss": 1.3407,
+      "step": 1390
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7856000000000001e-06,
+      "loss": 1.3516,
+      "step": 1395
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7920000000000002e-06,
+      "loss": 1.3394,
+      "step": 1400
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7984000000000002e-06,
+      "loss": 1.3431,
+      "step": 1405
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8048e-06,
+      "loss": 1.3346,
+      "step": 1410
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8112000000000001e-06,
+      "loss": 1.3364,
+      "step": 1415
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8176000000000002e-06,
+      "loss": 1.3337,
+      "step": 1420
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8240000000000002e-06,
+      "loss": 1.3255,
+      "step": 1425
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8304000000000003e-06,
+      "loss": 1.3376,
+      "step": 1430
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8368000000000004e-06,
+      "loss": 1.3674,
+      "step": 1435
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8432000000000002e-06,
+      "loss": 1.3323,
+      "step": 1440
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8496000000000003e-06,
+      "loss": 1.3244,
+      "step": 1445
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.856e-06,
+      "loss": 1.3328,
+      "step": 1450
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8624e-06,
+      "loss": 1.3299,
+      "step": 1455
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8688e-06,
+      "loss": 1.3258,
+      "step": 1460
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8752e-06,
+      "loss": 1.3274,
+      "step": 1465
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8816000000000001e-06,
+      "loss": 1.3327,
+      "step": 1470
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8880000000000002e-06,
+      "loss": 1.3311,
+      "step": 1475
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8944e-06,
+      "loss": 1.3206,
+      "step": 1480
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9008e-06,
+      "loss": 1.3196,
+      "step": 1485
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9072000000000001e-06,
+      "loss": 1.3148,
+      "step": 1490
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9136e-06,
+      "loss": 1.3196,
+      "step": 1495
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9200000000000003e-06,
+      "loss": 1.3246,
+      "step": 1500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9264000000000003e-06,
+      "loss": 1.3435,
+      "step": 1505
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9328000000000004e-06,
+      "loss": 1.3225,
+      "step": 1510
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9392000000000004e-06,
+      "loss": 1.306,
+      "step": 1515
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9456e-06,
+      "loss": 1.3141,
+      "step": 1520
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.952e-06,
+      "loss": 1.3152,
+      "step": 1525
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9584e-06,
+      "loss": 1.324,
+      "step": 1530
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9648000000000002e-06,
+      "loss": 1.3314,
+      "step": 1535
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9712e-06,
+      "loss": 1.3185,
+      "step": 1540
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9776e-06,
+      "loss": 1.313,
+      "step": 1545
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.984e-06,
+      "loss": 1.3079,
+      "step": 1550
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9904e-06,
+      "loss": 1.3119,
+      "step": 1555
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968e-06,
+      "loss": 1.3131,
+      "step": 1560
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.0032e-06,
+      "loss": 1.3072,
+      "step": 1565
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.0096e-06,
+      "loss": 1.3082,
+      "step": 1570
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.0160000000000003e-06,
+      "loss": 1.3033,
+      "step": 1575
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.0224000000000003e-06,
+      "loss": 1.3144,
+      "step": 1580
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.0288000000000004e-06,
+      "loss": 1.321,
+      "step": 1585
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.0352000000000004e-06,
+      "loss": 1.3186,
+      "step": 1590
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.0416e-06,
+      "loss": 1.3114,
+      "step": 1595
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.048e-06,
+      "loss": 1.3239,
+      "step": 1600
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.0544e-06,
+      "loss": 1.3137,
+      "step": 1605
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.0608000000000002e-06,
+      "loss": 1.3011,
+      "step": 1610
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.0672e-06,
+      "loss": 1.3122,
+      "step": 1615
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.0736e-06,
+      "loss": 1.32,
+      "step": 1620
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.08e-06,
+      "loss": 1.3039,
+      "step": 1625
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.0864e-06,
+      "loss": 1.3066,
+      "step": 1630
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.0928e-06,
+      "loss": 1.2993,
+      "step": 1635
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.0992e-06,
+      "loss": 1.3151,
+      "step": 1640
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1056000000000002e-06,
+      "loss": 1.3069,
+      "step": 1645
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1120000000000003e-06,
+      "loss": 1.287,
+      "step": 1650
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1184000000000003e-06,
+      "loss": 1.2889,
+      "step": 1655
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1248000000000004e-06,
+      "loss": 1.3052,
+      "step": 1660
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1312e-06,
+      "loss": 1.289,
+      "step": 1665
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1376e-06,
+      "loss": 1.2908,
+      "step": 1670
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.144e-06,
+      "loss": 1.2901,
+      "step": 1675
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1504e-06,
+      "loss": 1.3023,
+      "step": 1680
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1568000000000003e-06,
+      "loss": 1.2964,
+      "step": 1685
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1632000000000003e-06,
+      "loss": 1.3049,
+      "step": 1690
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1696e-06,
+      "loss": 1.2972,
+      "step": 1695
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.176e-06,
+      "loss": 1.2864,
+      "step": 1700
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1824e-06,
+      "loss": 1.2845,
+      "step": 1705
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1888e-06,
+      "loss": 1.2807,
+      "step": 1710
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1952e-06,
+      "loss": 1.2886,
+      "step": 1715
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.2016000000000002e-06,
+      "loss": 1.2848,
+      "step": 1720
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.2080000000000003e-06,
+      "loss": 1.2882,
+      "step": 1725
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.2144000000000003e-06,
+      "loss": 1.3015,
+      "step": 1730
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.2208e-06,
+      "loss": 1.2901,
+      "step": 1735
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.2272e-06,
+      "loss": 1.3016,
+      "step": 1740
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.2336e-06,
+      "loss": 1.2893,
+      "step": 1745
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.24e-06,
+      "loss": 1.2905,
+      "step": 1750
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.2464e-06,
+      "loss": 1.2711,
+      "step": 1755
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.2528000000000003e-06,
+      "loss": 1.2921,
+      "step": 1760
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.2592000000000003e-06,
+      "loss": 1.2906,
+      "step": 1765
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.2656000000000004e-06,
+      "loss": 1.2853,
+      "step": 1770
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.2720000000000004e-06,
+      "loss": 1.2856,
+      "step": 1775
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.2784e-06,
+      "loss": 1.2741,
+      "step": 1780
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.2848e-06,
+      "loss": 1.2789,
+      "step": 1785
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.2912e-06,
+      "loss": 1.2881,
+      "step": 1790
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.2976000000000002e-06,
+      "loss": 1.2792,
+      "step": 1795
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.3040000000000003e-06,
+      "loss": 1.2765,
+      "step": 1800
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.3104e-06,
+      "loss": 1.2863,
+      "step": 1805
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.3168e-06,
+      "loss": 1.2816,
+      "step": 1810
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.3232e-06,
+      "loss": 1.2886,
+      "step": 1815
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.3296e-06,
+      "loss": 1.2794,
+      "step": 1820
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.336e-06,
+      "loss": 1.2899,
+      "step": 1825
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.3424000000000002e-06,
+      "loss": 1.2895,
+      "step": 1830
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.3488000000000003e-06,
+      "loss": 1.2769,
+      "step": 1835
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.3552000000000003e-06,
+      "loss": 1.2717,
+      "step": 1840
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.3616000000000004e-06,
+      "loss": 1.286,
+      "step": 1845
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.3680000000000005e-06,
+      "loss": 1.2731,
+      "step": 1850
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.3744000000000005e-06,
+      "loss": 1.2753,
+      "step": 1855
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.3808e-06,
+      "loss": 1.2707,
+      "step": 1860
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.3872e-06,
+      "loss": 1.2686,
+      "step": 1865
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.3936000000000003e-06,
+      "loss": 1.2591,
+      "step": 1870
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 1.2777,
+      "step": 1875
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.4064e-06,
+      "loss": 1.2734,
+      "step": 1880
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.4128e-06,
+      "loss": 1.2722,
+      "step": 1885
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.4192e-06,
+      "loss": 1.2717,
+      "step": 1890
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.4256e-06,
+      "loss": 1.2663,
+      "step": 1895
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.432e-06,
+      "loss": 1.2559,
+      "step": 1900
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.4384000000000002e-06,
+      "loss": 1.2556,
+      "step": 1905
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.4448000000000003e-06,
+      "loss": 1.2613,
+      "step": 1910
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.4512000000000003e-06,
+      "loss": 1.2756,
+      "step": 1915
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.4576000000000004e-06,
+      "loss": 1.2729,
+      "step": 1920
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.4640000000000005e-06,
+      "loss": 1.2669,
+      "step": 1925
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.4704e-06,
+      "loss": 1.2571,
+      "step": 1930
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.4768e-06,
+      "loss": 1.2653,
+      "step": 1935
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.4832000000000002e-06,
+      "loss": 1.2534,
+      "step": 1940
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.4896000000000003e-06,
+      "loss": 1.2628,
+      "step": 1945
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.496e-06,
+      "loss": 1.2646,
+      "step": 1950
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.5024000000000004e-06,
+      "loss": 1.2566,
+      "step": 1955
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.5088e-06,
+      "loss": 1.2438,
+      "step": 1960
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.5152000000000005e-06,
+      "loss": 1.2411,
+      "step": 1965
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.5216e-06,
+      "loss": 1.2643,
+      "step": 1970
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.5280000000000006e-06,
+      "loss": 1.2745,
+      "step": 1975
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.5344000000000002e-06,
+      "loss": 1.2538,
+      "step": 1980
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.5408e-06,
+      "loss": 1.2471,
+      "step": 1985
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.5472000000000004e-06,
+      "loss": 1.2642,
+      "step": 1990
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.5536e-06,
+      "loss": 1.2732,
+      "step": 1995
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.56e-06,
+      "loss": 1.2555,
+      "step": 2000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.5664e-06,
+      "loss": 1.2584,
+      "step": 2005
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.5728e-06,
+      "loss": 1.2624,
+      "step": 2010
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.5792000000000002e-06,
+      "loss": 1.2666,
+      "step": 2015
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.5856000000000003e-06,
+      "loss": 1.2667,
+      "step": 2020
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.592e-06,
+      "loss": 1.2524,
+      "step": 2025
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.5984000000000004e-06,
+      "loss": 1.2653,
+      "step": 2030
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6048e-06,
+      "loss": 1.2391,
+      "step": 2035
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6112000000000005e-06,
+      "loss": 1.2605,
+      "step": 2040
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6176e-06,
+      "loss": 1.2638,
+      "step": 2045
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6240000000000006e-06,
+      "loss": 1.24,
+      "step": 2050
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6304000000000003e-06,
+      "loss": 1.2586,
+      "step": 2055
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6368000000000003e-06,
+      "loss": 1.2442,
+      "step": 2060
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6432000000000004e-06,
+      "loss": 1.2321,
+      "step": 2065
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6496e-06,
+      "loss": 1.2437,
+      "step": 2070
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.656e-06,
+      "loss": 1.23,
+      "step": 2075
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6624e-06,
+      "loss": 1.2537,
+      "step": 2080
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6688e-06,
+      "loss": 1.2525,
+      "step": 2085
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6752e-06,
+      "loss": 1.2437,
+      "step": 2090
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6816000000000003e-06,
+      "loss": 1.2541,
+      "step": 2095
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.688e-06,
+      "loss": 1.2265,
+      "step": 2100
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6944000000000004e-06,
+      "loss": 1.2541,
+      "step": 2105
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7008e-06,
+      "loss": 1.2571,
+      "step": 2110
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7072000000000005e-06,
+      "loss": 1.2479,
+      "step": 2115
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7136e-06,
+      "loss": 1.2528,
+      "step": 2120
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7200000000000002e-06,
+      "loss": 1.2432,
+      "step": 2125
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7264000000000003e-06,
+      "loss": 1.237,
+      "step": 2130
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7328000000000003e-06,
+      "loss": 1.2459,
+      "step": 2135
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7392000000000004e-06,
+      "loss": 1.2423,
+      "step": 2140
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7456000000000004e-06,
+      "loss": 1.2416,
+      "step": 2145
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.752e-06,
+      "loss": 1.2512,
+      "step": 2150
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7584e-06,
+      "loss": 1.2458,
+      "step": 2155
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7648e-06,
+      "loss": 1.2427,
+      "step": 2160
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7712e-06,
+      "loss": 1.2336,
+      "step": 2165
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7776000000000003e-06,
+      "loss": 1.2215,
+      "step": 2170
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.784e-06,
+      "loss": 1.2381,
+      "step": 2175
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7904000000000004e-06,
+      "loss": 1.2487,
+      "step": 2180
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7968e-06,
+      "loss": 1.224,
+      "step": 2185
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8032000000000005e-06,
+      "loss": 1.2525,
+      "step": 2190
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8096e-06,
+      "loss": 1.2341,
+      "step": 2195
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8160000000000002e-06,
+      "loss": 1.239,
+      "step": 2200
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8224000000000003e-06,
+      "loss": 1.2339,
+      "step": 2205
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8288000000000003e-06,
+      "loss": 1.2117,
+      "step": 2210
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8352e-06,
+      "loss": 1.2403,
+      "step": 2215
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8416000000000005e-06,
+      "loss": 1.2507,
+      "step": 2220
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.848e-06,
+      "loss": 1.2251,
+      "step": 2225
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8544000000000006e-06,
+      "loss": 1.2322,
+      "step": 2230
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8608e-06,
+      "loss": 1.2348,
+      "step": 2235
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8672e-06,
+      "loss": 1.2391,
+      "step": 2240
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8736000000000003e-06,
+      "loss": 1.2185,
+      "step": 2245
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.88e-06,
+      "loss": 1.2274,
+      "step": 2250
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8864000000000004e-06,
+      "loss": 1.2261,
+      "step": 2255
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8928e-06,
+      "loss": 1.2378,
+      "step": 2260
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8992000000000005e-06,
+      "loss": 1.2204,
+      "step": 2265
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9056e-06,
+      "loss": 1.2223,
+      "step": 2270
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9120000000000002e-06,
+      "loss": 1.2175,
+      "step": 2275
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9184000000000003e-06,
+      "loss": 1.2354,
+      "step": 2280
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9248000000000004e-06,
+      "loss": 1.2295,
+      "step": 2285
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9312e-06,
+      "loss": 1.2229,
+      "step": 2290
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9376000000000005e-06,
+      "loss": 1.233,
+      "step": 2295
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.944e-06,
+      "loss": 1.2283,
+      "step": 2300
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9504000000000006e-06,
+      "loss": 1.2305,
+      "step": 2305
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9568000000000002e-06,
+      "loss": 1.2257,
+      "step": 2310
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9632e-06,
+      "loss": 1.2202,
+      "step": 2315
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9696000000000003e-06,
+      "loss": 1.2102,
+      "step": 2320
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.976e-06,
+      "loss": 1.2338,
+      "step": 2325
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9824000000000004e-06,
+      "loss": 1.2201,
+      "step": 2330
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9888e-06,
+      "loss": 1.211,
+      "step": 2335
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9952e-06,
+      "loss": 1.2137,
+      "step": 2340
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.0016e-06,
+      "loss": 1.2248,
+      "step": 2345
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.0080000000000003e-06,
+      "loss": 1.2224,
+      "step": 2350
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.0144000000000003e-06,
+      "loss": 1.2241,
+      "step": 2355
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.0208000000000004e-06,
+      "loss": 1.2072,
+      "step": 2360
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.0272e-06,
+      "loss": 1.2214,
+      "step": 2365
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.0336000000000005e-06,
+      "loss": 1.1977,
+      "step": 2370
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.04e-06,
+      "loss": 1.2086,
+      "step": 2375
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.0464000000000006e-06,
+      "loss": 1.2131,
+      "step": 2380
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.0528000000000002e-06,
+      "loss": 1.2349,
+      "step": 2385
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.0592000000000007e-06,
+      "loss": 1.2279,
+      "step": 2390
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.0656000000000003e-06,
+      "loss": 1.2228,
+      "step": 2395
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.072e-06,
+      "loss": 1.2292,
+      "step": 2400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.0784000000000005e-06,
+      "loss": 1.2195,
+      "step": 2405
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.0848e-06,
+      "loss": 1.2181,
+      "step": 2410
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.0912e-06,
+      "loss": 1.2208,
+      "step": 2415
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.0976e-06,
+      "loss": 1.2104,
+      "step": 2420
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.1040000000000003e-06,
+      "loss": 1.2094,
+      "step": 2425
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.1104e-06,
+      "loss": 1.2127,
+      "step": 2430
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.1168000000000004e-06,
+      "loss": 1.2238,
+      "step": 2435
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.1232e-06,
+      "loss": 1.2172,
+      "step": 2440
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.1296000000000005e-06,
+      "loss": 1.2101,
+      "step": 2445
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.136e-06,
+      "loss": 1.2164,
+      "step": 2450
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.1424000000000006e-06,
+      "loss": 1.2118,
+      "step": 2455
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.1488000000000002e-06,
+      "loss": 1.2142,
+      "step": 2460
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.1552000000000003e-06,
+      "loss": 1.2087,
+      "step": 2465
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.1616000000000004e-06,
+      "loss": 1.2323,
+      "step": 2470
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.1680000000000004e-06,
+      "loss": 1.2207,
+      "step": 2475
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.1744e-06,
+      "loss": 1.2257,
+      "step": 2480
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.1808e-06,
+      "loss": 1.1972,
+      "step": 2485
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.1872e-06,
+      "loss": 1.2023,
+      "step": 2490
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.1936000000000002e-06,
+      "loss": 1.2067,
+      "step": 2495
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 1.2064,
+      "step": 2500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.2064e-06,
+      "loss": 1.2167,
+      "step": 2505
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.2128000000000004e-06,
+      "loss": 1.2108,
+      "step": 2510
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.2192e-06,
+      "loss": 1.2102,
+      "step": 2515
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.2256000000000005e-06,
+      "loss": 1.2149,
+      "step": 2520
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.232e-06,
+      "loss": 1.2085,
+      "step": 2525
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.2384000000000006e-06,
+      "loss": 1.2,
+      "step": 2530
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.2448000000000003e-06,
+      "loss": 1.2016,
+      "step": 2535
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.2512000000000003e-06,
+      "loss": 1.2142,
+      "step": 2540
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.2576000000000004e-06,
+      "loss": 1.2163,
+      "step": 2545
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.2640000000000004e-06,
+      "loss": 1.2028,
+      "step": 2550
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.2704e-06,
+      "loss": 1.1994,
+      "step": 2555
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.2768000000000005e-06,
+      "loss": 1.1973,
+      "step": 2560
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.2832e-06,
+      "loss": 1.2048,
+      "step": 2565
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.2896e-06,
+      "loss": 1.1863,
+      "step": 2570
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.2960000000000003e-06,
+      "loss": 1.2074,
+      "step": 2575
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3024e-06,
+      "loss": 1.2108,
+      "step": 2580
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3088000000000004e-06,
+      "loss": 1.2082,
+      "step": 2585
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3152e-06,
+      "loss": 1.1918,
+      "step": 2590
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3216000000000005e-06,
+      "loss": 1.2178,
+      "step": 2595
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.328e-06,
+      "loss": 1.1889,
+      "step": 2600
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3344e-06,
+      "loss": 1.2029,
+      "step": 2605
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3408000000000003e-06,
+      "loss": 1.2069,
+      "step": 2610
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3472000000000003e-06,
+      "loss": 1.2117,
+      "step": 2615
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3536000000000004e-06,
+      "loss": 1.2043,
+      "step": 2620
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3600000000000004e-06,
+      "loss": 1.1912,
+      "step": 2625
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3664e-06,
+      "loss": 1.1852,
+      "step": 2630
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3728000000000006e-06,
+      "loss": 1.2023,
+      "step": 2635
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3792e-06,
+      "loss": 1.1801,
+      "step": 2640
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3856000000000007e-06,
+      "loss": 1.204,
+      "step": 2645
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3920000000000003e-06,
+      "loss": 1.2159,
+      "step": 2650
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3984e-06,
+      "loss": 1.1923,
+      "step": 2655
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.4048000000000004e-06,
+      "loss": 1.1966,
+      "step": 2660
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.4112e-06,
+      "loss": 1.1846,
+      "step": 2665
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.4176000000000005e-06,
+      "loss": 1.1846,
+      "step": 2670
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.424e-06,
+      "loss": 1.1758,
+      "step": 2675
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.4304000000000002e-06,
+      "loss": 1.1757,
+      "step": 2680
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.4368000000000003e-06,
+      "loss": 1.1903,
+      "step": 2685
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.4432000000000003e-06,
+      "loss": 1.1876,
+      "step": 2690
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.4496e-06,
+      "loss": 1.1803,
+      "step": 2695
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.4560000000000005e-06,
+      "loss": 1.1645,
+      "step": 2700
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.4624e-06,
+      "loss": 1.1897,
+      "step": 2705
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.4688000000000006e-06,
+      "loss": 1.1921,
+      "step": 2710
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.4752e-06,
+      "loss": 1.1884,
+      "step": 2715
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.4816000000000007e-06,
+      "loss": 1.1819,
+      "step": 2720
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.4880000000000003e-06,
+      "loss": 1.1808,
+      "step": 2725
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.4944e-06,
+      "loss": 1.1762,
+      "step": 2730
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.5008000000000004e-06,
+      "loss": 1.1973,
+      "step": 2735
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.5072e-06,
+      "loss": 1.174,
+      "step": 2740
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.5136000000000005e-06,
+      "loss": 1.2014,
+      "step": 2745
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.52e-06,
+      "loss": 1.1842,
+      "step": 2750
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.5264000000000002e-06,
+      "loss": 1.1756,
+      "step": 2755
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.5328000000000003e-06,
+      "loss": 1.1707,
+      "step": 2760
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.5392000000000004e-06,
+      "loss": 1.1843,
+      "step": 2765
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.5456e-06,
+      "loss": 1.1768,
+      "step": 2770
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.5520000000000005e-06,
+      "loss": 1.1657,
+      "step": 2775
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.5584e-06,
+      "loss": 1.187,
+      "step": 2780
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.5648000000000006e-06,
+      "loss": 1.1849,
+      "step": 2785
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.5712000000000002e-06,
+      "loss": 1.1726,
+      "step": 2790
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.5776000000000007e-06,
+      "loss": 1.1715,
+      "step": 2795
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.5840000000000003e-06,
+      "loss": 1.1893,
+      "step": 2800
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.5904000000000004e-06,
+      "loss": 1.1647,
+      "step": 2805
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.5968000000000004e-06,
+      "loss": 1.1998,
+      "step": 2810
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.6032e-06,
+      "loss": 1.1876,
+      "step": 2815
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.6096e-06,
+      "loss": 1.1954,
+      "step": 2820
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.616e-06,
+      "loss": 1.1815,
+      "step": 2825
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.6224000000000002e-06,
+      "loss": 1.1883,
+      "step": 2830
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.6288000000000003e-06,
+      "loss": 1.1937,
+      "step": 2835
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.6352000000000004e-06,
+      "loss": 1.1706,
+      "step": 2840
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.6416e-06,
+      "loss": 1.1839,
+      "step": 2845
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.6480000000000005e-06,
+      "loss": 1.1614,
+      "step": 2850
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.6544e-06,
+      "loss": 1.1787,
+      "step": 2855
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.6608000000000006e-06,
+      "loss": 1.182,
+      "step": 2860
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.6672000000000002e-06,
+      "loss": 1.1826,
+      "step": 2865
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.6736000000000007e-06,
+      "loss": 1.1771,
+      "step": 2870
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.6800000000000003e-06,
+      "loss": 1.1878,
+      "step": 2875
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.6864000000000004e-06,
+      "loss": 1.1823,
+      "step": 2880
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.6928000000000005e-06,
+      "loss": 1.1782,
+      "step": 2885
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.6992000000000005e-06,
+      "loss": 1.1768,
+      "step": 2890
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.7056e-06,
+      "loss": 1.1898,
+      "step": 2895
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.712e-06,
+      "loss": 1.1775,
+      "step": 2900
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.7184000000000003e-06,
+      "loss": 1.1724,
+      "step": 2905
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.7248e-06,
+      "loss": 1.1498,
+      "step": 2910
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.7312000000000004e-06,
+      "loss": 1.1684,
+      "step": 2915
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.7376e-06,
+      "loss": 1.1571,
+      "step": 2920
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.7440000000000005e-06,
+      "loss": 1.1717,
+      "step": 2925
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.7504e-06,
+      "loss": 1.1754,
+      "step": 2930
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.7568000000000006e-06,
+      "loss": 1.155,
+      "step": 2935
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.7632000000000002e-06,
+      "loss": 1.1626,
+      "step": 2940
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.7696000000000003e-06,
+      "loss": 1.1724,
+      "step": 2945
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.7760000000000004e-06,
+      "loss": 1.1635,
+      "step": 2950
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.7824000000000004e-06,
+      "loss": 1.1543,
+      "step": 2955
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.7888e-06,
+      "loss": 1.1759,
+      "step": 2960
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.7952000000000005e-06,
+      "loss": 1.1561,
+      "step": 2965
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.8016e-06,
+      "loss": 1.1785,
+      "step": 2970
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.8080000000000006e-06,
+      "loss": 1.1688,
+      "step": 2975
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.8144000000000003e-06,
+      "loss": 1.1759,
+      "step": 2980
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.8208e-06,
+      "loss": 1.1705,
+      "step": 2985
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.8272e-06,
+      "loss": 1.1588,
+      "step": 2990
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.8336e-06,
+      "loss": 1.1631,
+      "step": 2995
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.8400000000000005e-06,
+      "loss": 1.1649,
+      "step": 3000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.8464e-06,
+      "loss": 1.1778,
+      "step": 3005
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.852800000000001e-06,
+      "loss": 1.1691,
+      "step": 3010
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.8592e-06,
+      "loss": 1.1603,
+      "step": 3015
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.865600000000001e-06,
+      "loss": 1.1672,
+      "step": 3020
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.872e-06,
+      "loss": 1.1635,
+      "step": 3025
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.878400000000001e-06,
+      "loss": 1.1661,
+      "step": 3030
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.8848000000000005e-06,
+      "loss": 1.1561,
+      "step": 3035
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.8912e-06,
+      "loss": 1.1639,
+      "step": 3040
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.897600000000001e-06,
+      "loss": 1.17,
+      "step": 3045
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.904e-06,
+      "loss": 1.1514,
+      "step": 3050
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.910400000000001e-06,
+      "loss": 1.1704,
+      "step": 3055
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9168e-06,
+      "loss": 1.1664,
+      "step": 3060
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9232e-06,
+      "loss": 1.1543,
+      "step": 3065
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9296000000000005e-06,
+      "loss": 1.1462,
+      "step": 3070
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.936e-06,
+      "loss": 1.1662,
+      "step": 3075
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9424e-06,
+      "loss": 1.1567,
+      "step": 3080
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9488e-06,
+      "loss": 1.1523,
+      "step": 3085
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9552e-06,
+      "loss": 1.1354,
+      "step": 3090
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9616e-06,
+      "loss": 1.1616,
+      "step": 3095
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.968e-06,
+      "loss": 1.1584,
+      "step": 3100
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9744000000000004e-06,
+      "loss": 1.1478,
+      "step": 3105
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9808e-06,
+      "loss": 1.1562,
+      "step": 3110
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9872000000000006e-06,
+      "loss": 1.157,
+      "step": 3115
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9936e-06,
+      "loss": 1.1639,
+      "step": 3120
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 1.1421,
+      "step": 3125
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.0064e-06,
+      "loss": 1.1459,
+      "step": 3130
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.012800000000001e-06,
+      "loss": 1.159,
+      "step": 3135
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.0192e-06,
+      "loss": 1.1462,
+      "step": 3140
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.0256e-06,
+      "loss": 1.1504,
+      "step": 3145
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.0320000000000005e-06,
+      "loss": 1.1537,
+      "step": 3150
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.0384e-06,
+      "loss": 1.1426,
+      "step": 3155
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.044800000000001e-06,
+      "loss": 1.1658,
+      "step": 3160
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.0512e-06,
+      "loss": 1.1394,
+      "step": 3165
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.057600000000001e-06,
+      "loss": 1.1534,
+      "step": 3170
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.064e-06,
+      "loss": 1.1461,
+      "step": 3175
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.070400000000001e-06,
+      "loss": 1.1534,
+      "step": 3180
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.0768000000000005e-06,
+      "loss": 1.1479,
+      "step": 3185
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.0832e-06,
+      "loss": 1.1521,
+      "step": 3190
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.089600000000001e-06,
+      "loss": 1.1643,
+      "step": 3195
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.096e-06,
+      "loss": 1.1591,
+      "step": 3200
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.1024e-06,
+      "loss": 1.1489,
+      "step": 3205
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.1088e-06,
+      "loss": 1.1445,
+      "step": 3210
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.1152e-06,
+      "loss": 1.1613,
+      "step": 3215
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.1216000000000005e-06,
+      "loss": 1.1436,
+      "step": 3220
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.128e-06,
+      "loss": 1.1523,
+      "step": 3225
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.1344e-06,
+      "loss": 1.1571,
+      "step": 3230
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.1408e-06,
+      "loss": 1.1565,
+      "step": 3235
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.1472e-06,
+      "loss": 1.1406,
+      "step": 3240
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.1536e-06,
+      "loss": 1.1499,
+      "step": 3245
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.16e-06,
+      "loss": 1.1654,
+      "step": 3250
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.1664000000000005e-06,
+      "loss": 1.156,
+      "step": 3255
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.1728e-06,
+      "loss": 1.1513,
+      "step": 3260
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.179200000000001e-06,
+      "loss": 1.1289,
+      "step": 3265
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.1856e-06,
+      "loss": 1.1543,
+      "step": 3270
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.192000000000001e-06,
+      "loss": 1.1429,
+      "step": 3275
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.1984e-06,
+      "loss": 1.1467,
+      "step": 3280
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.204800000000001e-06,
+      "loss": 1.1308,
+      "step": 3285
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.2112000000000004e-06,
+      "loss": 1.141,
+      "step": 3290
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.217600000000001e-06,
+      "loss": 1.1566,
+      "step": 3295
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.2240000000000006e-06,
+      "loss": 1.1493,
+      "step": 3300
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.230400000000001e-06,
+      "loss": 1.1438,
+      "step": 3305
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.236800000000001e-06,
+      "loss": 1.1448,
+      "step": 3310
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.2432e-06,
+      "loss": 1.1433,
+      "step": 3315
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.249600000000001e-06,
+      "loss": 1.1533,
+      "step": 3320
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.256e-06,
+      "loss": 1.1174,
+      "step": 3325
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.2624e-06,
+      "loss": 1.1375,
+      "step": 3330
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.2688000000000005e-06,
+      "loss": 1.1547,
+      "step": 3335
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.2752e-06,
+      "loss": 1.1384,
+      "step": 3340
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.2816e-06,
+      "loss": 1.1429,
+      "step": 3345
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.288e-06,
+      "loss": 1.1531,
+      "step": 3350
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.2944e-06,
+      "loss": 1.1459,
+      "step": 3355
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.3008e-06,
+      "loss": 1.1463,
+      "step": 3360
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.3072e-06,
+      "loss": 1.1342,
+      "step": 3365
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.3136000000000005e-06,
+      "loss": 1.1486,
+      "step": 3370
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.32e-06,
+      "loss": 1.1343,
+      "step": 3375
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.326400000000001e-06,
+      "loss": 1.1653,
+      "step": 3380
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.3328e-06,
+      "loss": 1.1446,
+      "step": 3385
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.3392e-06,
+      "loss": 1.1291,
+      "step": 3390
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.3456e-06,
+      "loss": 1.1394,
+      "step": 3395
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.352e-06,
+      "loss": 1.1315,
+      "step": 3400
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.3584000000000005e-06,
+      "loss": 1.1552,
+      "step": 3405
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.3648e-06,
+      "loss": 1.1299,
+      "step": 3410
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.371200000000001e-06,
+      "loss": 1.1531,
+      "step": 3415
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.3776e-06,
+      "loss": 1.1364,
+      "step": 3420
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.384000000000001e-06,
+      "loss": 1.1519,
+      "step": 3425
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.3904e-06,
+      "loss": 1.1627,
+      "step": 3430
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.396800000000001e-06,
+      "loss": 1.1275,
+      "step": 3435
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4032000000000005e-06,
+      "loss": 1.1364,
+      "step": 3440
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.409600000000001e-06,
+      "loss": 1.1411,
+      "step": 3445
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.416000000000001e-06,
+      "loss": 1.1468,
+      "step": 3450
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4224e-06,
+      "loss": 1.1503,
+      "step": 3455
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.428800000000001e-06,
+      "loss": 1.1438,
+      "step": 3460
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4352e-06,
+      "loss": 1.1409,
+      "step": 3465
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4416e-06,
+      "loss": 1.137,
+      "step": 3470
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4480000000000004e-06,
+      "loss": 1.1411,
+      "step": 3475
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4544e-06,
+      "loss": 1.1438,
+      "step": 3480
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4608e-06,
+      "loss": 1.1329,
+      "step": 3485
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4672e-06,
+      "loss": 1.1356,
+      "step": 3490
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4736e-06,
+      "loss": 1.1393,
+      "step": 3495
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.48e-06,
+      "loss": 1.1279,
+      "step": 3500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4864e-06,
+      "loss": 1.1393,
+      "step": 3505
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4928e-06,
+      "loss": 1.1309,
+      "step": 3510
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4992e-06,
+      "loss": 1.1427,
+      "step": 3515
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.5056000000000005e-06,
+      "loss": 1.1188,
+      "step": 3520
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.512e-06,
+      "loss": 1.1333,
+      "step": 3525
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.518400000000001e-06,
+      "loss": 1.1416,
+      "step": 3530
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.5248e-06,
+      "loss": 1.1093,
+      "step": 3535
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.531200000000001e-06,
+      "loss": 1.1195,
+      "step": 3540
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.5376e-06,
+      "loss": 1.1359,
+      "step": 3545
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.544000000000001e-06,
+      "loss": 1.144,
+      "step": 3550
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.5504000000000005e-06,
+      "loss": 1.1411,
+      "step": 3555
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.5568e-06,
+      "loss": 1.1247,
+      "step": 3560
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.563200000000001e-06,
+      "loss": 1.1356,
+      "step": 3565
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.5696e-06,
+      "loss": 1.1362,
+      "step": 3570
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.576000000000001e-06,
+      "loss": 1.1328,
+      "step": 3575
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.5824e-06,
+      "loss": 1.1421,
+      "step": 3580
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.588800000000001e-06,
+      "loss": 1.1389,
+      "step": 3585
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.5952000000000005e-06,
+      "loss": 1.1264,
+      "step": 3590
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.6016e-06,
+      "loss": 1.1168,
+      "step": 3595
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.608000000000001e-06,
+      "loss": 1.1224,
+      "step": 3600
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.6144e-06,
+      "loss": 1.1276,
+      "step": 3605
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.6208e-06,
+      "loss": 1.1343,
+      "step": 3610
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.6272e-06,
+      "loss": 1.1517,
+      "step": 3615
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.6336e-06,
+      "loss": 1.1194,
+      "step": 3620
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.6400000000000005e-06,
+      "loss": 1.1175,
+      "step": 3625
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.6464e-06,
+      "loss": 1.1301,
+      "step": 3630
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.652800000000001e-06,
+      "loss": 1.1385,
+      "step": 3635
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.6592e-06,
+      "loss": 1.1304,
+      "step": 3640
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.6656e-06,
+      "loss": 1.13,
+      "step": 3645
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.672e-06,
+      "loss": 1.136,
+      "step": 3650
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.6784e-06,
+      "loss": 1.13,
+      "step": 3655
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.6848000000000004e-06,
+      "loss": 1.1171,
+      "step": 3660
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.6912e-06,
+      "loss": 1.117,
+      "step": 3665
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.6976000000000006e-06,
+      "loss": 1.1348,
+      "step": 3670
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.704e-06,
+      "loss": 1.1202,
+      "step": 3675
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.710400000000001e-06,
+      "loss": 1.1244,
+      "step": 3680
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.7168e-06,
+      "loss": 1.129,
+      "step": 3685
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.723200000000001e-06,
+      "loss": 1.1281,
+      "step": 3690
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.7296e-06,
+      "loss": 1.1292,
+      "step": 3695
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.736000000000001e-06,
+      "loss": 1.1214,
+      "step": 3700
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.7424000000000005e-06,
+      "loss": 1.1147,
+      "step": 3705
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.748800000000001e-06,
+      "loss": 1.1221,
+      "step": 3710
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.755200000000001e-06,
+      "loss": 1.1251,
+      "step": 3715
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.7616e-06,
+      "loss": 1.1192,
+      "step": 3720
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.768000000000001e-06,
+      "loss": 1.1113,
+      "step": 3725
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.7744e-06,
+      "loss": 1.1303,
+      "step": 3730
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.7808e-06,
+      "loss": 1.1382,
+      "step": 3735
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.7872000000000005e-06,
+      "loss": 1.1132,
+      "step": 3740
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.7936e-06,
+      "loss": 1.12,
+      "step": 3745
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 1.1229,
+      "step": 3750
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8064e-06,
+      "loss": 1.1215,
+      "step": 3755
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8128e-06,
+      "loss": 1.1218,
+      "step": 3760
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8192e-06,
+      "loss": 1.117,
+      "step": 3765
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8256e-06,
+      "loss": 1.1133,
+      "step": 3770
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8320000000000005e-06,
+      "loss": 1.1067,
+      "step": 3775
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8384e-06,
+      "loss": 1.1147,
+      "step": 3780
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.844800000000001e-06,
+      "loss": 1.1081,
+      "step": 3785
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8512e-06,
+      "loss": 1.1059,
+      "step": 3790
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.857600000000001e-06,
+      "loss": 1.1188,
+      "step": 3795
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.864e-06,
+      "loss": 1.1219,
+      "step": 3800
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8704e-06,
+      "loss": 1.1125,
+      "step": 3805
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8768000000000005e-06,
+      "loss": 1.1102,
+      "step": 3810
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8832e-06,
+      "loss": 1.1176,
+      "step": 3815
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.889600000000001e-06,
+      "loss": 1.117,
+      "step": 3820
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.896e-06,
+      "loss": 1.1167,
+      "step": 3825
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.902400000000001e-06,
+      "loss": 1.1284,
+      "step": 3830
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9088e-06,
+      "loss": 1.1064,
+      "step": 3835
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.915200000000001e-06,
+      "loss": 1.1134,
+      "step": 3840
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9216000000000004e-06,
+      "loss": 1.1087,
+      "step": 3845
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.928000000000001e-06,
+      "loss": 1.1161,
+      "step": 3850
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9344000000000006e-06,
+      "loss": 1.1061,
+      "step": 3855
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9408e-06,
+      "loss": 1.1224,
+      "step": 3860
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.947200000000001e-06,
+      "loss": 1.1233,
+      "step": 3865
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9536e-06,
+      "loss": 1.1126,
+      "step": 3870
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.960000000000001e-06,
+      "loss": 1.1029,
+      "step": 3875
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9664000000000004e-06,
+      "loss": 1.1088,
+      "step": 3880
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9728e-06,
+      "loss": 1.122,
+      "step": 3885
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9792000000000005e-06,
+      "loss": 1.109,
+      "step": 3890
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9856e-06,
+      "loss": 1.1128,
+      "step": 3895
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.992e-06,
+      "loss": 1.1019,
+      "step": 3900
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9984e-06,
+      "loss": 1.1092,
+      "step": 3905
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.004800000000001e-06,
+      "loss": 1.1368,
+      "step": 3910
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.0112e-06,
+      "loss": 1.1141,
+      "step": 3915
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.0176e-06,
+      "loss": 1.1032,
+      "step": 3920
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.024e-06,
+      "loss": 1.1166,
+      "step": 3925
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.030400000000001e-06,
+      "loss": 1.1135,
+      "step": 3930
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.036800000000001e-06,
+      "loss": 1.1105,
+      "step": 3935
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.0432e-06,
+      "loss": 1.1097,
+      "step": 3940
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.0496e-06,
+      "loss": 1.1196,
+      "step": 3945
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.056000000000001e-06,
+      "loss": 1.1085,
+      "step": 3950
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.062400000000001e-06,
+      "loss": 1.1197,
+      "step": 3955
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.0688000000000005e-06,
+      "loss": 1.1153,
+      "step": 3960
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.0752e-06,
+      "loss": 1.1081,
+      "step": 3965
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.0816e-06,
+      "loss": 1.108,
+      "step": 3970
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.088000000000001e-06,
+      "loss": 1.1096,
+      "step": 3975
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.094400000000001e-06,
+      "loss": 1.098,
+      "step": 3980
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.1008e-06,
+      "loss": 1.1007,
+      "step": 3985
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.1072e-06,
+      "loss": 1.1012,
+      "step": 3990
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.1136000000000005e-06,
+      "loss": 1.1062,
+      "step": 3995
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.12e-06,
+      "loss": 1.1051,
+      "step": 4000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.126400000000001e-06,
+      "loss": 1.1022,
+      "step": 4005
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.1328e-06,
+      "loss": 1.1045,
+      "step": 4010
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.139200000000001e-06,
+      "loss": 1.0868,
+      "step": 4015
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.1456e-06,
+      "loss": 1.1038,
+      "step": 4020
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.152e-06,
+      "loss": 1.1027,
+      "step": 4025
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.1584000000000005e-06,
+      "loss": 1.0988,
+      "step": 4030
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.164800000000001e-06,
+      "loss": 1.0951,
+      "step": 4035
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.1712000000000006e-06,
+      "loss": 1.1101,
+      "step": 4040
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.1776e-06,
+      "loss": 1.113,
+      "step": 4045
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.184e-06,
+      "loss": 1.1006,
+      "step": 4050
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.1903999999999995e-06,
+      "loss": 1.0965,
+      "step": 4055
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.196800000000001e-06,
+      "loss": 1.094,
+      "step": 4060
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.2032000000000004e-06,
+      "loss": 1.1024,
+      "step": 4065
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.2096e-06,
+      "loss": 1.0949,
+      "step": 4070
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.216e-06,
+      "loss": 1.104,
+      "step": 4075
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.222400000000001e-06,
+      "loss": 1.0972,
+      "step": 4080
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.228800000000001e-06,
+      "loss": 1.1053,
+      "step": 4085
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.2352e-06,
+      "loss": 1.0882,
+      "step": 4090
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.2416e-06,
+      "loss": 1.0947,
+      "step": 4095
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.248000000000001e-06,
+      "loss": 1.0961,
+      "step": 4100
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.254400000000001e-06,
+      "loss": 1.105,
+      "step": 4105
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.2608000000000005e-06,
+      "loss": 1.1045,
+      "step": 4110
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.2672e-06,
+      "loss": 1.0944,
+      "step": 4115
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.273600000000001e-06,
+      "loss": 1.1194,
+      "step": 4120
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.28e-06,
+      "loss": 1.0988,
+      "step": 4125
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.286400000000001e-06,
+      "loss": 1.0964,
+      "step": 4130
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.2928e-06,
+      "loss": 1.0974,
+      "step": 4135
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.2992e-06,
+      "loss": 1.0932,
+      "step": 4140
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.3056000000000005e-06,
+      "loss": 1.0935,
+      "step": 4145
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.312e-06,
+      "loss": 1.1063,
+      "step": 4150
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.318400000000001e-06,
+      "loss": 1.1039,
+      "step": 4155
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.3248e-06,
+      "loss": 1.0976,
+      "step": 4160
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.331200000000001e-06,
+      "loss": 1.1055,
+      "step": 4165
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.3376e-06,
+      "loss": 1.1024,
+      "step": 4170
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.344e-06,
+      "loss": 1.0948,
+      "step": 4175
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.3504e-06,
+      "loss": 1.0916,
+      "step": 4180
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.356800000000001e-06,
+      "loss": 1.1027,
+      "step": 4185
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.363200000000001e-06,
+      "loss": 1.0911,
+      "step": 4190
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.3696e-06,
+      "loss": 1.0785,
+      "step": 4195
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.376e-06,
+      "loss": 1.0963,
+      "step": 4200
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.382400000000001e-06,
+      "loss": 1.1146,
+      "step": 4205
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.388800000000001e-06,
+      "loss": 1.1052,
+      "step": 4210
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.3952000000000005e-06,
+      "loss": 1.0896,
+      "step": 4215
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.4016e-06,
+      "loss": 1.0978,
+      "step": 4220
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.408e-06,
+      "loss": 1.0934,
+      "step": 4225
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.414400000000001e-06,
+      "loss": 1.0947,
+      "step": 4230
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.420800000000001e-06,
+      "loss": 1.0942,
+      "step": 4235
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.4272e-06,
+      "loss": 1.0831,
+      "step": 4240
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.4336e-06,
+      "loss": 1.1008,
+      "step": 4245
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.4400000000000004e-06,
+      "loss": 1.096,
+      "step": 4250
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.446400000000001e-06,
+      "loss": 1.0949,
+      "step": 4255
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.4528000000000005e-06,
+      "loss": 1.0885,
+      "step": 4260
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.4592e-06,
+      "loss": 1.0937,
+      "step": 4265
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.465600000000001e-06,
+      "loss": 1.0931,
+      "step": 4270
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.472e-06,
+      "loss": 1.0996,
+      "step": 4275
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.478400000000001e-06,
+      "loss": 1.0817,
+      "step": 4280
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.4848e-06,
+      "loss": 1.0868,
+      "step": 4285
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.491200000000001e-06,
+      "loss": 1.0908,
+      "step": 4290
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.4976000000000005e-06,
+      "loss": 1.1053,
+      "step": 4295
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.504e-06,
+      "loss": 1.0902,
+      "step": 4300
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.5104e-06,
+      "loss": 1.0843,
+      "step": 4305
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.5168e-06,
+      "loss": 1.0987,
+      "step": 4310
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.523200000000001e-06,
+      "loss": 1.1047,
+      "step": 4315
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.5296e-06,
+      "loss": 1.1021,
+      "step": 4320
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.536e-06,
+      "loss": 1.091,
+      "step": 4325
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.5424e-06,
+      "loss": 1.0851,
+      "step": 4330
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.548800000000001e-06,
+      "loss": 1.0738,
+      "step": 4335
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.555200000000001e-06,
+      "loss": 1.0986,
+      "step": 4340
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.5616e-06,
+      "loss": 1.1086,
+      "step": 4345
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.568e-06,
+      "loss": 1.0828,
+      "step": 4350
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.574400000000001e-06,
+      "loss": 1.0802,
+      "step": 4355
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.580800000000001e-06,
+      "loss": 1.0756,
+      "step": 4360
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.5872000000000005e-06,
+      "loss": 1.0716,
+      "step": 4365
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.5936e-06,
+      "loss": 1.0891,
+      "step": 4370
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 1.0753,
+      "step": 4375
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.606400000000001e-06,
+      "loss": 1.0756,
+      "step": 4380
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.612800000000001e-06,
+      "loss": 1.0906,
+      "step": 4385
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.6192e-06,
+      "loss": 1.0901,
+      "step": 4390
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.6256e-06,
+      "loss": 1.0683,
+      "step": 4395
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.6320000000000005e-06,
+      "loss": 1.0814,
+      "step": 4400
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.638400000000001e-06,
+      "loss": 1.084,
+      "step": 4405
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.6448000000000006e-06,
+      "loss": 1.0843,
+      "step": 4410
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.6512e-06,
+      "loss": 1.0862,
+      "step": 4415
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.657600000000001e-06,
+      "loss": 1.073,
+      "step": 4420
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.664e-06,
+      "loss": 1.0885,
+      "step": 4425
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.6704e-06,
+      "loss": 1.0968,
+      "step": 4430
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.6768000000000004e-06,
+      "loss": 1.0841,
+      "step": 4435
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.683200000000001e-06,
+      "loss": 1.0782,
+      "step": 4440
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.6896000000000005e-06,
+      "loss": 1.0972,
+      "step": 4445
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.696e-06,
+      "loss": 1.0786,
+      "step": 4450
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.7024e-06,
+      "loss": 1.0806,
+      "step": 4455
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.708800000000001e-06,
+      "loss": 1.0873,
+      "step": 4460
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.715200000000001e-06,
+      "loss": 1.0864,
+      "step": 4465
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.7216e-06,
+      "loss": 1.0776,
+      "step": 4470
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.728e-06,
+      "loss": 1.084,
+      "step": 4475
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.7344e-06,
+      "loss": 1.0791,
+      "step": 4480
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.740800000000001e-06,
+      "loss": 1.083,
+      "step": 4485
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.747200000000001e-06,
+      "loss": 1.0859,
+      "step": 4490
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.7536e-06,
+      "loss": 1.0917,
+      "step": 4495
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.76e-06,
+      "loss": 1.0824,
+      "step": 4500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.766400000000001e-06,
+      "loss": 1.0849,
+      "step": 4505
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.772800000000001e-06,
+      "loss": 1.0772,
+      "step": 4510
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.7792000000000005e-06,
+      "loss": 1.0824,
+      "step": 4515
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.7856e-06,
+      "loss": 1.0661,
+      "step": 4520
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.792000000000001e-06,
+      "loss": 1.0654,
+      "step": 4525
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.798400000000001e-06,
+      "loss": 1.0809,
+      "step": 4530
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.804800000000001e-06,
+      "loss": 1.0822,
+      "step": 4535
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.8112e-06,
+      "loss": 1.0759,
+      "step": 4540
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.817600000000001e-06,
+      "loss": 1.0764,
+      "step": 4545
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.8240000000000005e-06,
+      "loss": 1.084,
+      "step": 4550
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.8304e-06,
+      "loss": 1.0857,
+      "step": 4555
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.836800000000001e-06,
+      "loss": 1.0769,
+      "step": 4560
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.8432e-06,
+      "loss": 1.0812,
+      "step": 4565
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.849600000000001e-06,
+      "loss": 1.0736,
+      "step": 4570
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.856e-06,
+      "loss": 1.0708,
+      "step": 4575
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.8624e-06,
+      "loss": 1.0782,
+      "step": 4580
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.8688000000000005e-06,
+      "loss": 1.0681,
+      "step": 4585
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.875200000000001e-06,
+      "loss": 1.0737,
+      "step": 4590
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.8816000000000006e-06,
+      "loss": 1.0563,
+      "step": 4595
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.888e-06,
+      "loss": 1.0874,
+      "step": 4600
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.8944e-06,
+      "loss": 1.0702,
+      "step": 4605
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.900800000000001e-06,
+      "loss": 1.0759,
+      "step": 4610
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.907200000000001e-06,
+      "loss": 1.069,
+      "step": 4615
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9136000000000004e-06,
+      "loss": 1.0854,
+      "step": 4620
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.92e-06,
+      "loss": 1.0817,
+      "step": 4625
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9264e-06,
+      "loss": 1.0715,
+      "step": 4630
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.932800000000001e-06,
+      "loss": 1.0658,
+      "step": 4635
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.939200000000001e-06,
+      "loss": 1.0956,
+      "step": 4640
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9456e-06,
+      "loss": 1.0769,
+      "step": 4645
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.952e-06,
+      "loss": 1.0765,
+      "step": 4650
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.958400000000001e-06,
+      "loss": 1.0685,
+      "step": 4655
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.964800000000001e-06,
+      "loss": 1.0838,
+      "step": 4660
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9712000000000005e-06,
+      "loss": 1.0689,
+      "step": 4665
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9776e-06,
+      "loss": 1.0739,
+      "step": 4670
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.984000000000001e-06,
+      "loss": 1.0598,
+      "step": 4675
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9904e-06,
+      "loss": 1.0591,
+      "step": 4680
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.996800000000001e-06,
+      "loss": 1.0691,
+      "step": 4685
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.0032e-06,
+      "loss": 1.0649,
+      "step": 4690
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.009600000000001e-06,
+      "loss": 1.0686,
+      "step": 4695
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.0160000000000005e-06,
+      "loss": 1.0903,
+      "step": 4700
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.0224e-06,
+      "loss": 1.0572,
+      "step": 4705
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.028800000000001e-06,
+      "loss": 1.0776,
+      "step": 4710
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.0352e-06,
+      "loss": 1.0562,
+      "step": 4715
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.041600000000001e-06,
+      "loss": 1.0634,
+      "step": 4720
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.048e-06,
+      "loss": 1.0607,
+      "step": 4725
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.0544e-06,
+      "loss": 1.065,
+      "step": 4730
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.0608e-06,
+      "loss": 1.0768,
+      "step": 4735
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.067200000000001e-06,
+      "loss": 1.0714,
+      "step": 4740
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.073600000000001e-06,
+      "loss": 1.064,
+      "step": 4745
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.08e-06,
+      "loss": 1.0538,
+      "step": 4750
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.0864e-06,
+      "loss": 1.0824,
+      "step": 4755
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.092800000000001e-06,
+      "loss": 1.0694,
+      "step": 4760
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.099200000000001e-06,
+      "loss": 1.0748,
+      "step": 4765
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.1056000000000005e-06,
+      "loss": 1.0536,
+      "step": 4770
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.112e-06,
+      "loss": 1.0521,
+      "step": 4775
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.1184000000000014e-06,
+      "loss": 1.0677,
+      "step": 4780
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.124800000000001e-06,
+      "loss": 1.0648,
+      "step": 4785
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.131200000000001e-06,
+      "loss": 1.0741,
+      "step": 4790
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.1376e-06,
+      "loss": 1.0583,
+      "step": 4795
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.144e-06,
+      "loss": 1.0671,
+      "step": 4800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.1504000000000004e-06,
+      "loss": 1.062,
+      "step": 4805
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.156800000000001e-06,
+      "loss": 1.0676,
+      "step": 4810
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.1632000000000006e-06,
+      "loss": 1.068,
+      "step": 4815
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.1696e-06,
+      "loss": 1.0704,
+      "step": 4820
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.176000000000001e-06,
+      "loss": 1.06,
+      "step": 4825
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.1824e-06,
+      "loss": 1.0835,
+      "step": 4830
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.188800000000001e-06,
+      "loss": 1.0541,
+      "step": 4835
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.1952e-06,
+      "loss": 1.0636,
+      "step": 4840
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.201600000000001e-06,
+      "loss": 1.0613,
+      "step": 4845
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.2080000000000005e-06,
+      "loss": 1.0516,
+      "step": 4850
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.2144e-06,
+      "loss": 1.0537,
+      "step": 4855
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.2208e-06,
+      "loss": 1.0563,
+      "step": 4860
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.227200000000001e-06,
+      "loss": 1.0686,
+      "step": 4865
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.233600000000001e-06,
+      "loss": 1.0635,
+      "step": 4870
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.24e-06,
+      "loss": 1.0724,
+      "step": 4875
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.2464e-06,
+      "loss": 1.0648,
+      "step": 4880
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.2528e-06,
+      "loss": 1.0773,
+      "step": 4885
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.259200000000001e-06,
+      "loss": 1.0591,
+      "step": 4890
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.265600000000001e-06,
+      "loss": 1.0571,
+      "step": 4895
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.272e-06,
+      "loss": 1.0419,
+      "step": 4900
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.2784e-06,
+      "loss": 1.0715,
+      "step": 4905
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.284800000000001e-06,
+      "loss": 1.0617,
+      "step": 4910
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.291200000000001e-06,
+      "loss": 1.072,
+      "step": 4915
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.2976000000000005e-06,
+      "loss": 1.0733,
+      "step": 4920
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.304e-06,
+      "loss": 1.0468,
+      "step": 4925
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.310400000000001e-06,
+      "loss": 1.0618,
+      "step": 4930
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.316800000000001e-06,
+      "loss": 1.0567,
+      "step": 4935
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.323200000000001e-06,
+      "loss": 1.076,
+      "step": 4940
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.3296e-06,
+      "loss": 1.0676,
+      "step": 4945
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.336000000000001e-06,
+      "loss": 1.0654,
+      "step": 4950
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.3424000000000005e-06,
+      "loss": 1.056,
+      "step": 4955
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.3488e-06,
+      "loss": 1.0741,
+      "step": 4960
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.355200000000001e-06,
+      "loss": 1.0556,
+      "step": 4965
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.3616e-06,
+      "loss": 1.0664,
+      "step": 4970
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.368000000000001e-06,
+      "loss": 1.0494,
+      "step": 4975
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.3744e-06,
+      "loss": 1.0628,
+      "step": 4980
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.3808e-06,
+      "loss": 1.0528,
+      "step": 4985
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.3872000000000004e-06,
+      "loss": 1.0606,
+      "step": 4990
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.393600000000001e-06,
+      "loss": 1.0548,
+      "step": 4995
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 1.0434,
+      "step": 5000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.4064e-06,
+      "loss": 1.0298,
+      "step": 5005
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.4128e-06,
+      "loss": 1.0451,
+      "step": 5010
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.419200000000001e-06,
+      "loss": 1.0565,
+      "step": 5015
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.425600000000001e-06,
+      "loss": 1.0611,
+      "step": 5020
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.432e-06,
+      "loss": 1.0505,
+      "step": 5025
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.4384e-06,
+      "loss": 1.0563,
+      "step": 5030
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.444800000000001e-06,
+      "loss": 1.0625,
+      "step": 5035
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.451200000000001e-06,
+      "loss": 1.0601,
+      "step": 5040
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.457600000000001e-06,
+      "loss": 1.0514,
+      "step": 5045
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.464e-06,
+      "loss": 1.0434,
+      "step": 5050
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.4704e-06,
+      "loss": 1.037,
+      "step": 5055
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.476800000000001e-06,
+      "loss": 1.0531,
+      "step": 5060
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.483200000000001e-06,
+      "loss": 1.047,
+      "step": 5065
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.4896000000000005e-06,
+      "loss": 1.0408,
+      "step": 5070
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.496e-06,
+      "loss": 1.0489,
+      "step": 5075
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.502400000000001e-06,
+      "loss": 1.0545,
+      "step": 5080
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.5088e-06,
+      "loss": 1.0555,
+      "step": 5085
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.515200000000001e-06,
+      "loss": 1.0603,
+      "step": 5090
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.5216e-06,
+      "loss": 1.0415,
+      "step": 5095
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.528000000000001e-06,
+      "loss": 1.0547,
+      "step": 5100
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.5344000000000005e-06,
+      "loss": 1.0452,
+      "step": 5105
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.5408e-06,
+      "loss": 1.0611,
+      "step": 5110
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.547200000000001e-06,
+      "loss": 1.0604,
+      "step": 5115
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.553600000000001e-06,
+      "loss": 1.0455,
+      "step": 5120
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.560000000000001e-06,
+      "loss": 1.0583,
+      "step": 5125
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.5664e-06,
+      "loss": 1.0493,
+      "step": 5130
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.5728e-06,
+      "loss": 1.0521,
+      "step": 5135
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.5792e-06,
+      "loss": 1.0707,
+      "step": 5140
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.585600000000001e-06,
+      "loss": 1.0615,
+      "step": 5145
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.592000000000001e-06,
+      "loss": 1.0697,
+      "step": 5150
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.5984e-06,
+      "loss": 1.0473,
+      "step": 5155
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.6048e-06,
+      "loss": 1.0533,
+      "step": 5160
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.611200000000001e-06,
+      "loss": 1.0601,
+      "step": 5165
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.617600000000001e-06,
+      "loss": 1.0316,
+      "step": 5170
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.6240000000000004e-06,
+      "loss": 1.054,
+      "step": 5175
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.6304e-06,
+      "loss": 1.052,
+      "step": 5180
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.636800000000001e-06,
+      "loss": 1.0508,
+      "step": 5185
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.643200000000001e-06,
+      "loss": 1.0503,
+      "step": 5190
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.649600000000001e-06,
+      "loss": 1.044,
+      "step": 5195
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.656e-06,
+      "loss": 1.0572,
+      "step": 5200
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.662400000000001e-06,
+      "loss": 1.0439,
+      "step": 5205
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.6688e-06,
+      "loss": 1.0596,
+      "step": 5210
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.675200000000001e-06,
+      "loss": 1.0474,
+      "step": 5215
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.6816000000000005e-06,
+      "loss": 1.0536,
+      "step": 5220
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.688e-06,
+      "loss": 1.0393,
+      "step": 5225
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.694400000000001e-06,
+      "loss": 1.0396,
+      "step": 5230
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.7008e-06,
+      "loss": 1.0473,
+      "step": 5235
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.707200000000001e-06,
+      "loss": 1.0459,
+      "step": 5240
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.7136e-06,
+      "loss": 1.0405,
+      "step": 5245
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.720000000000001e-06,
+      "loss": 1.0418,
+      "step": 5250
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.7264000000000005e-06,
+      "loss": 1.0621,
+      "step": 5255
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.7328e-06,
+      "loss": 1.049,
+      "step": 5260
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.7392e-06,
+      "loss": 1.0462,
+      "step": 5265
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.745600000000001e-06,
+      "loss": 1.0452,
+      "step": 5270
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.752000000000001e-06,
+      "loss": 1.0403,
+      "step": 5275
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.7584e-06,
+      "loss": 1.052,
+      "step": 5280
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.7648e-06,
+      "loss": 1.0382,
+      "step": 5285
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.771200000000001e-06,
+      "loss": 1.0353,
+      "step": 5290
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.777600000000001e-06,
+      "loss": 1.0386,
+      "step": 5295
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.784000000000001e-06,
+      "loss": 1.0429,
+      "step": 5300
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.7904e-06,
+      "loss": 1.0452,
+      "step": 5305
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.7968e-06,
+      "loss": 1.0397,
+      "step": 5310
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.803200000000001e-06,
+      "loss": 1.0434,
+      "step": 5315
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.809600000000001e-06,
+      "loss": 1.0482,
+      "step": 5320
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.8160000000000005e-06,
+      "loss": 1.0361,
+      "step": 5325
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.8224e-06,
+      "loss": 1.0356,
+      "step": 5330
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.828800000000001e-06,
+      "loss": 1.0478,
+      "step": 5335
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.835200000000001e-06,
+      "loss": 1.048,
+      "step": 5340
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.841600000000001e-06,
+      "loss": 1.0155,
+      "step": 5345
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.848e-06,
+      "loss": 1.0397,
+      "step": 5350
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.854400000000001e-06,
+      "loss": 1.0385,
+      "step": 5355
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.8608000000000004e-06,
+      "loss": 1.0347,
+      "step": 5360
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.867200000000001e-06,
+      "loss": 1.0326,
+      "step": 5365
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.8736000000000006e-06,
+      "loss": 1.0423,
+      "step": 5370
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.88e-06,
+      "loss": 1.0348,
+      "step": 5375
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.886400000000001e-06,
+      "loss": 1.0581,
+      "step": 5380
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.8928e-06,
+      "loss": 1.0366,
+      "step": 5385
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.8992e-06,
+      "loss": 1.0362,
+      "step": 5390
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.9056000000000004e-06,
+      "loss": 1.0369,
+      "step": 5395
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.912000000000001e-06,
+      "loss": 1.031,
+      "step": 5400
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.9184000000000005e-06,
+      "loss": 1.0433,
+      "step": 5405
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.9248e-06,
+      "loss": 1.0308,
+      "step": 5410
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.9312e-06,
+      "loss": 1.0312,
+      "step": 5415
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.937600000000001e-06,
+      "loss": 1.0466,
+      "step": 5420
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.944000000000001e-06,
+      "loss": 1.0449,
+      "step": 5425
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.9504e-06,
+      "loss": 1.0314,
+      "step": 5430
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.9568e-06,
+      "loss": 1.0337,
+      "step": 5435
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.963200000000001e-06,
+      "loss": 1.0287,
+      "step": 5440
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.969600000000001e-06,
+      "loss": 1.0525,
+      "step": 5445
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.976000000000001e-06,
+      "loss": 1.0354,
+      "step": 5450
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.9824e-06,
+      "loss": 1.0241,
+      "step": 5455
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.9888e-06,
+      "loss": 1.0315,
+      "step": 5460
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.995200000000001e-06,
+      "loss": 1.029,
+      "step": 5465
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.001600000000001e-06,
+      "loss": 1.0309,
+      "step": 5470
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.0080000000000005e-06,
+      "loss": 1.0362,
+      "step": 5475
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.0144e-06,
+      "loss": 1.0282,
+      "step": 5480
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.020800000000001e-06,
+      "loss": 1.0276,
+      "step": 5485
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.027200000000001e-06,
+      "loss": 1.0375,
+      "step": 5490
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.033600000000001e-06,
+      "loss": 1.0265,
+      "step": 5495
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.04e-06,
+      "loss": 1.0319,
+      "step": 5500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.046400000000001e-06,
+      "loss": 1.0477,
+      "step": 5505
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.0528000000000005e-06,
+      "loss": 1.0397,
+      "step": 5510
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.0592e-06,
+      "loss": 1.0229,
+      "step": 5515
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.065600000000001e-06,
+      "loss": 1.0467,
+      "step": 5520
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.072000000000001e-06,
+      "loss": 1.0455,
+      "step": 5525
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.078400000000001e-06,
+      "loss": 1.0294,
+      "step": 5530
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.0848e-06,
+      "loss": 1.0293,
+      "step": 5535
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.0912e-06,
+      "loss": 1.0459,
+      "step": 5540
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.0976000000000005e-06,
+      "loss": 1.0413,
+      "step": 5545
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.104000000000001e-06,
+      "loss": 1.029,
+      "step": 5550
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.1104000000000006e-06,
+      "loss": 1.0304,
+      "step": 5555
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.1168e-06,
+      "loss": 1.0362,
+      "step": 5560
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.1232e-06,
+      "loss": 1.0438,
+      "step": 5565
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.129600000000001e-06,
+      "loss": 1.0199,
+      "step": 5570
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.136000000000001e-06,
+      "loss": 1.0331,
+      "step": 5575
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.1424000000000004e-06,
+      "loss": 1.0426,
+      "step": 5580
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.1488e-06,
+      "loss": 1.0272,
+      "step": 5585
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.155200000000001e-06,
+      "loss": 1.0324,
+      "step": 5590
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.161600000000001e-06,
+      "loss": 1.0401,
+      "step": 5595
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.168000000000001e-06,
+      "loss": 1.031,
+      "step": 5600
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.1744e-06,
+      "loss": 1.0176,
+      "step": 5605
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.180800000000001e-06,
+      "loss": 1.026,
+      "step": 5610
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.187200000000001e-06,
+      "loss": 1.0294,
+      "step": 5615
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.193600000000001e-06,
+      "loss": 1.0272,
+      "step": 5620
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.2000000000000005e-06,
+      "loss": 1.0409,
+      "step": 5625
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.2064e-06,
+      "loss": 1.0474,
+      "step": 5630
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.212800000000001e-06,
+      "loss": 1.025,
+      "step": 5635
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.2192e-06,
+      "loss": 1.0218,
+      "step": 5640
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.225600000000001e-06,
+      "loss": 1.0405,
+      "step": 5645
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.232e-06,
+      "loss": 1.017,
+      "step": 5650
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.238400000000001e-06,
+      "loss": 1.0334,
+      "step": 5655
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.2448000000000005e-06,
+      "loss": 1.0364,
+      "step": 5660
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.2512e-06,
+      "loss": 1.0327,
+      "step": 5665
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.257600000000001e-06,
+      "loss": 1.0442,
+      "step": 5670
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.264000000000001e-06,
+      "loss": 1.0396,
+      "step": 5675
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.270400000000001e-06,
+      "loss": 1.024,
+      "step": 5680
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.2768e-06,
+      "loss": 1.0282,
+      "step": 5685
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.2832e-06,
+      "loss": 1.0197,
+      "step": 5690
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.289600000000001e-06,
+      "loss": 1.0294,
+      "step": 5695
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.296000000000001e-06,
+      "loss": 1.0107,
+      "step": 5700
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.302400000000001e-06,
+      "loss": 1.0292,
+      "step": 5705
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.3088e-06,
+      "loss": 1.0261,
+      "step": 5710
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.3152e-06,
+      "loss": 1.0266,
+      "step": 5715
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.321600000000001e-06,
+      "loss": 1.0254,
+      "step": 5720
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.328000000000001e-06,
+      "loss": 1.0267,
+      "step": 5725
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.3344000000000005e-06,
+      "loss": 1.013,
+      "step": 5730
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.3408e-06,
+      "loss": 1.0352,
+      "step": 5735
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.347200000000001e-06,
+      "loss": 1.025,
+      "step": 5740
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.353600000000001e-06,
+      "loss": 1.034,
+      "step": 5745
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.360000000000001e-06,
+      "loss": 1.0261,
+      "step": 5750
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.3664e-06,
+      "loss": 1.029,
+      "step": 5755
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.372800000000001e-06,
+      "loss": 1.0113,
+      "step": 5760
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.3792000000000004e-06,
+      "loss": 1.0248,
+      "step": 5765
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.385600000000001e-06,
+      "loss": 1.0338,
+      "step": 5770
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.3920000000000005e-06,
+      "loss": 1.0273,
+      "step": 5775
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.398400000000001e-06,
+      "loss": 1.0215,
+      "step": 5780
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.404800000000001e-06,
+      "loss": 1.0135,
+      "step": 5785
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.4112e-06,
+      "loss": 1.0169,
+      "step": 5790
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.417600000000001e-06,
+      "loss": 1.0262,
+      "step": 5795
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.424e-06,
+      "loss": 1.0164,
+      "step": 5800
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.430400000000001e-06,
+      "loss": 1.0261,
+      "step": 5805
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.4368000000000005e-06,
+      "loss": 1.0103,
+      "step": 5810
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.4432e-06,
+      "loss": 1.0129,
+      "step": 5815
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.4496e-06,
+      "loss": 1.0269,
+      "step": 5820
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.456000000000001e-06,
+      "loss": 1.0242,
+      "step": 5825
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.462400000000001e-06,
+      "loss": 1.0315,
+      "step": 5830
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.4688e-06,
+      "loss": 1.0285,
+      "step": 5835
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.4752e-06,
+      "loss": 1.0257,
+      "step": 5840
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.481600000000001e-06,
+      "loss": 1.0245,
+      "step": 5845
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.488000000000001e-06,
+      "loss": 1.0216,
+      "step": 5850
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.494400000000001e-06,
+      "loss": 1.0197,
+      "step": 5855
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.5008e-06,
+      "loss": 1.0173,
+      "step": 5860
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.507200000000001e-06,
+      "loss": 1.0199,
+      "step": 5865
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.513600000000001e-06,
+      "loss": 1.0172,
+      "step": 5870
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.520000000000001e-06,
+      "loss": 1.0036,
+      "step": 5875
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.5264000000000005e-06,
+      "loss": 1.0237,
+      "step": 5880
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.5328e-06,
+      "loss": 1.0295,
+      "step": 5885
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.539200000000001e-06,
+      "loss": 1.0135,
+      "step": 5890
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.545600000000001e-06,
+      "loss": 1.0369,
+      "step": 5895
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.552000000000001e-06,
+      "loss": 1.0127,
+      "step": 5900
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.5584e-06,
+      "loss": 1.0194,
+      "step": 5905
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.564800000000001e-06,
+      "loss": 1.0172,
+      "step": 5910
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.5712000000000005e-06,
+      "loss": 1.0165,
+      "step": 5915
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.5776e-06,
+      "loss": 1.0187,
+      "step": 5920
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.5840000000000006e-06,
+      "loss": 1.0038,
+      "step": 5925
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.590400000000001e-06,
+      "loss": 1.014,
+      "step": 5930
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.596800000000001e-06,
+      "loss": 1.0166,
+      "step": 5935
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.6032e-06,
+      "loss": 1.0013,
+      "step": 5940
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.6096e-06,
+      "loss": 1.0189,
+      "step": 5945
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.616000000000001e-06,
+      "loss": 1.0127,
+      "step": 5950
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.622400000000001e-06,
+      "loss": 1.014,
+      "step": 5955
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.6288000000000005e-06,
+      "loss": 1.0311,
+      "step": 5960
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.635200000000001e-06,
+      "loss": 1.0022,
+      "step": 5965
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.6416e-06,
+      "loss": 1.003,
+      "step": 5970
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.648e-06,
+      "loss": 1.0197,
+      "step": 5975
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.6544e-06,
+      "loss": 1.0194,
+      "step": 5980
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.660800000000001e-06,
+      "loss": 1.0099,
+      "step": 5985
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.6672e-06,
+      "loss": 1.0115,
+      "step": 5990
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.6736e-06,
+      "loss": 1.0061,
+      "step": 5995
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.680000000000001e-06,
+      "loss": 1.0131,
+      "step": 6000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.6864e-06,
+      "loss": 1.02,
+      "step": 6005
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.6928e-06,
+      "loss": 1.0006,
+      "step": 6010
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.6992e-06,
+      "loss": 1.0154,
+      "step": 6015
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.705600000000001e-06,
+      "loss": 1.0283,
+      "step": 6020
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.712e-06,
+      "loss": 1.0086,
+      "step": 6025
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.7184e-06,
+      "loss": 0.9959,
+      "step": 6030
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.724800000000001e-06,
+      "loss": 1.0283,
+      "step": 6035
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.731200000000001e-06,
+      "loss": 1.0063,
+      "step": 6040
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.7376e-06,
+      "loss": 1.0115,
+      "step": 6045
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.744e-06,
+      "loss": 1.0161,
+      "step": 6050
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.7504e-06,
+      "loss": 0.9981,
+      "step": 6055
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.756800000000002e-06,
+      "loss": 1.0175,
+      "step": 6060
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.7632e-06,
+      "loss": 1.0056,
+      "step": 6065
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.769600000000001e-06,
+      "loss": 1.0104,
+      "step": 6070
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.776e-06,
+      "loss": 1.0034,
+      "step": 6075
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.7824e-06,
+      "loss": 1.0135,
+      "step": 6080
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.7888e-06,
+      "loss": 1.0196,
+      "step": 6085
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.795200000000001e-06,
+      "loss": 1.0129,
+      "step": 6090
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.8016e-06,
+      "loss": 0.9997,
+      "step": 6095
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.808e-06,
+      "loss": 1.014,
+      "step": 6100
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.814400000000001e-06,
+      "loss": 1.0185,
+      "step": 6105
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.820800000000001e-06,
+      "loss": 1.0121,
+      "step": 6110
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.8272e-06,
+      "loss": 1.0132,
+      "step": 6115
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.8336e-06,
+      "loss": 1.0053,
+      "step": 6120
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.840000000000001e-06,
+      "loss": 1.0149,
+      "step": 6125
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.8464e-06,
+      "loss": 0.9911,
+      "step": 6130
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.8528e-06,
+      "loss": 1.0041,
+      "step": 6135
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.859200000000001e-06,
+      "loss": 1.0129,
+      "step": 6140
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.865600000000001e-06,
+      "loss": 1.0186,
+      "step": 6145
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.872e-06,
+      "loss": 1.0348,
+      "step": 6150
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.8784e-06,
+      "loss": 0.9939,
+      "step": 6155
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.8848e-06,
+      "loss": 1.0114,
+      "step": 6160
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.891200000000002e-06,
+      "loss": 1.0001,
+      "step": 6165
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.8976e-06,
+      "loss": 0.9871,
+      "step": 6170
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.904000000000001e-06,
+      "loss": 1.0123,
+      "step": 6175
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.9104e-06,
+      "loss": 1.0083,
+      "step": 6180
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.916800000000002e-06,
+      "loss": 0.9998,
+      "step": 6185
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.9232e-06,
+      "loss": 1.0081,
+      "step": 6190
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.929600000000001e-06,
+      "loss": 1.0202,
+      "step": 6195
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.936e-06,
+      "loss": 1.0073,
+      "step": 6200
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.9424e-06,
+      "loss": 0.9925,
+      "step": 6205
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.948800000000001e-06,
+      "loss": 1.0175,
+      "step": 6210
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.955200000000001e-06,
+      "loss": 1.0017,
+      "step": 6215
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.9616e-06,
+      "loss": 1.0201,
+      "step": 6220
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.968e-06,
+      "loss": 1.0017,
+      "step": 6225
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.974400000000001e-06,
+      "loss": 1.0038,
+      "step": 6230
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.980800000000002e-06,
+      "loss": 1.0057,
+      "step": 6235
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.9872e-06,
+      "loss": 0.9943,
+      "step": 6240
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.993600000000001e-06,
+      "loss": 0.9988,
+      "step": 6245
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 1.0061,
+      "step": 6250
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.0064e-06,
+      "loss": 1.0027,
+      "step": 6255
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.0128e-06,
+      "loss": 1.0096,
+      "step": 6260
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.019200000000001e-06,
+      "loss": 1.0024,
+      "step": 6265
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.025600000000002e-06,
+      "loss": 1.0018,
+      "step": 6270
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.032e-06,
+      "loss": 1.0145,
+      "step": 6275
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.0384e-06,
+      "loss": 1.0073,
+      "step": 6280
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.0448e-06,
+      "loss": 1.0182,
+      "step": 6285
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.0512e-06,
+      "loss": 1.0026,
+      "step": 6290
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.0576e-06,
+      "loss": 1.0011,
+      "step": 6295
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.064000000000001e-06,
+      "loss": 1.0114,
+      "step": 6300
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.0704e-06,
+      "loss": 0.9968,
+      "step": 6305
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.0768e-06,
+      "loss": 0.9973,
+      "step": 6310
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.0832e-06,
+      "loss": 0.9983,
+      "step": 6315
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.089600000000001e-06,
+      "loss": 1.0231,
+      "step": 6320
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.096e-06,
+      "loss": 1.0109,
+      "step": 6325
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.1024e-06,
+      "loss": 1.0037,
+      "step": 6330
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.108800000000001e-06,
+      "loss": 0.9971,
+      "step": 6335
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.115200000000002e-06,
+      "loss": 0.996,
+      "step": 6340
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.1216e-06,
+      "loss": 1.0111,
+      "step": 6345
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.128e-06,
+      "loss": 0.9987,
+      "step": 6350
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.134400000000001e-06,
+      "loss": 1.0042,
+      "step": 6355
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.140800000000002e-06,
+      "loss": 1.0009,
+      "step": 6360
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.1472e-06,
+      "loss": 0.9924,
+      "step": 6365
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.153600000000001e-06,
+      "loss": 1.0061,
+      "step": 6370
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.16e-06,
+      "loss": 0.9997,
+      "step": 6375
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.1664e-06,
+      "loss": 0.9925,
+      "step": 6380
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.1728e-06,
+      "loss": 1.0199,
+      "step": 6385
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.179200000000001e-06,
+      "loss": 0.9964,
+      "step": 6390
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.1856e-06,
+      "loss": 1.0075,
+      "step": 6395
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.192e-06,
+      "loss": 1.0074,
+      "step": 6400
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.198400000000001e-06,
+      "loss": 0.9893,
+      "step": 6405
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.2048e-06,
+      "loss": 0.9958,
+      "step": 6410
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.2112e-06,
+      "loss": 0.9877,
+      "step": 6415
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.2176e-06,
+      "loss": 0.985,
+      "step": 6420
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.224000000000001e-06,
+      "loss": 0.9946,
+      "step": 6425
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.2304e-06,
+      "loss": 1.0027,
+      "step": 6430
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.2368e-06,
+      "loss": 0.9975,
+      "step": 6435
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.243200000000001e-06,
+      "loss": 0.9843,
+      "step": 6440
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.249600000000001e-06,
+      "loss": 1.0027,
+      "step": 6445
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.256e-06,
+      "loss": 0.9821,
+      "step": 6450
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.2624e-06,
+      "loss": 0.994,
+      "step": 6455
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.2688e-06,
+      "loss": 1.0038,
+      "step": 6460
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.275200000000002e-06,
+      "loss": 0.9958,
+      "step": 6465
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.2816e-06,
+      "loss": 1.0042,
+      "step": 6470
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.288000000000001e-06,
+      "loss": 1.0105,
+      "step": 6475
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.2944e-06,
+      "loss": 0.9932,
+      "step": 6480
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.300800000000002e-06,
+      "loss": 1.0046,
+      "step": 6485
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.3072e-06,
+      "loss": 0.9986,
+      "step": 6490
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.313600000000001e-06,
+      "loss": 1.0022,
+      "step": 6495
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.32e-06,
+      "loss": 0.9939,
+      "step": 6500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.3264e-06,
+      "loss": 1.0096,
+      "step": 6505
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.332800000000001e-06,
+      "loss": 1.0108,
+      "step": 6510
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.339200000000001e-06,
+      "loss": 0.9836,
+      "step": 6515
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.3456e-06,
+      "loss": 0.9977,
+      "step": 6520
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.352e-06,
+      "loss": 0.9927,
+      "step": 6525
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.358400000000001e-06,
+      "loss": 0.9983,
+      "step": 6530
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.3648e-06,
+      "loss": 1.0027,
+      "step": 6535
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.3712e-06,
+      "loss": 1.0072,
+      "step": 6540
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.377600000000001e-06,
+      "loss": 1.0042,
+      "step": 6545
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.384000000000001e-06,
+      "loss": 1.0007,
+      "step": 6550
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.3904e-06,
+      "loss": 0.9776,
+      "step": 6555
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.3968e-06,
+      "loss": 0.9978,
+      "step": 6560
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.4032e-06,
+      "loss": 0.9871,
+      "step": 6565
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.409600000000002e-06,
+      "loss": 0.9843,
+      "step": 6570
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.416e-06,
+      "loss": 0.9816,
+      "step": 6575
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.422400000000001e-06,
+      "loss": 0.9946,
+      "step": 6580
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.4288e-06,
+      "loss": 0.9942,
+      "step": 6585
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.435200000000002e-06,
+      "loss": 0.9847,
+      "step": 6590
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.4416e-06,
+      "loss": 0.9925,
+      "step": 6595
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.448000000000001e-06,
+      "loss": 0.9974,
+      "step": 6600
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.4544e-06,
+      "loss": 0.9997,
+      "step": 6605
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.460800000000002e-06,
+      "loss": 0.9877,
+      "step": 6610
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.467200000000001e-06,
+      "loss": 0.989,
+      "step": 6615
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.473600000000001e-06,
+      "loss": 0.9954,
+      "step": 6620
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.48e-06,
+      "loss": 0.9919,
+      "step": 6625
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.4864e-06,
+      "loss": 0.9787,
+      "step": 6630
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.492800000000001e-06,
+      "loss": 0.9987,
+      "step": 6635
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.499200000000002e-06,
+      "loss": 1.008,
+      "step": 6640
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.5056e-06,
+      "loss": 0.984,
+      "step": 6645
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.512e-06,
+      "loss": 1.0007,
+      "step": 6650
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.518400000000001e-06,
+      "loss": 0.9918,
+      "step": 6655
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.5248e-06,
+      "loss": 0.9753,
+      "step": 6660
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.5312e-06,
+      "loss": 0.9789,
+      "step": 6665
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.537600000000001e-06,
+      "loss": 0.9953,
+      "step": 6670
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.544000000000002e-06,
+      "loss": 1.0014,
+      "step": 6675
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.5504e-06,
+      "loss": 0.9956,
+      "step": 6680
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.5568e-06,
+      "loss": 0.9911,
+      "step": 6685
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.5632e-06,
+      "loss": 0.986,
+      "step": 6690
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.569600000000002e-06,
+      "loss": 0.9825,
+      "step": 6695
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.576e-06,
+      "loss": 1.0011,
+      "step": 6700
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.582400000000001e-06,
+      "loss": 0.9717,
+      "step": 6705
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.5888e-06,
+      "loss": 0.9864,
+      "step": 6710
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.5952e-06,
+      "loss": 0.9962,
+      "step": 6715
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.6016e-06,
+      "loss": 1.0024,
+      "step": 6720
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.608000000000001e-06,
+      "loss": 0.9769,
+      "step": 6725
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.6144e-06,
+      "loss": 0.9892,
+      "step": 6730
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.6208e-06,
+      "loss": 0.9928,
+      "step": 6735
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.627200000000001e-06,
+      "loss": 0.9906,
+      "step": 6740
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.633600000000001e-06,
+      "loss": 0.9843,
+      "step": 6745
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.64e-06,
+      "loss": 0.9921,
+      "step": 6750
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.6464e-06,
+      "loss": 0.9954,
+      "step": 6755
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.652800000000001e-06,
+      "loss": 0.993,
+      "step": 6760
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.659200000000002e-06,
+      "loss": 0.9825,
+      "step": 6765
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.6656e-06,
+      "loss": 1.0045,
+      "step": 6770
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.672000000000001e-06,
+      "loss": 1.0073,
+      "step": 6775
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.6784e-06,
+      "loss": 1.0027,
+      "step": 6780
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.6848e-06,
+      "loss": 0.9958,
+      "step": 6785
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.6912e-06,
+      "loss": 0.9869,
+      "step": 6790
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.697600000000001e-06,
+      "loss": 0.9893,
+      "step": 6795
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.704e-06,
+      "loss": 0.9952,
+      "step": 6800
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.7104e-06,
+      "loss": 0.9811,
+      "step": 6805
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.716800000000001e-06,
+      "loss": 0.9807,
+      "step": 6810
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.7232e-06,
+      "loss": 0.9734,
+      "step": 6815
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.7296e-06,
+      "loss": 0.9782,
+      "step": 6820
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.736e-06,
+      "loss": 0.9795,
+      "step": 6825
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.742400000000001e-06,
+      "loss": 0.9945,
+      "step": 6830
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.7488e-06,
+      "loss": 0.9985,
+      "step": 6835
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.7552e-06,
+      "loss": 0.9905,
+      "step": 6840
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.761600000000001e-06,
+      "loss": 0.9881,
+      "step": 6845
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.768000000000001e-06,
+      "loss": 0.9986,
+      "step": 6850
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.7744e-06,
+      "loss": 0.9943,
+      "step": 6855
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.7808e-06,
+      "loss": 0.9836,
+      "step": 6860
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.7872e-06,
+      "loss": 0.9885,
+      "step": 6865
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.793600000000002e-06,
+      "loss": 0.9719,
+      "step": 6870
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.8e-06,
+      "loss": 0.978,
+      "step": 6875
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.806400000000001e-06,
+      "loss": 0.9883,
+      "step": 6880
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.8128e-06,
+      "loss": 0.9868,
+      "step": 6885
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.819200000000002e-06,
+      "loss": 0.9955,
+      "step": 6890
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.8256e-06,
+      "loss": 0.98,
+      "step": 6895
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.832000000000001e-06,
+      "loss": 0.9979,
+      "step": 6900
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.8384e-06,
+      "loss": 0.9776,
+      "step": 6905
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.8448e-06,
+      "loss": 0.9883,
+      "step": 6910
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.851200000000001e-06,
+      "loss": 0.9757,
+      "step": 6915
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.857600000000001e-06,
+      "loss": 0.9742,
+      "step": 6920
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.864e-06,
+      "loss": 0.9842,
+      "step": 6925
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.8704e-06,
+      "loss": 0.9828,
+      "step": 6930
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.876800000000001e-06,
+      "loss": 0.9903,
+      "step": 6935
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.8832e-06,
+      "loss": 0.986,
+      "step": 6940
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.8896e-06,
+      "loss": 0.9862,
+      "step": 6945
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.896000000000001e-06,
+      "loss": 0.9805,
+      "step": 6950
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.902400000000001e-06,
+      "loss": 0.9862,
+      "step": 6955
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.9088e-06,
+      "loss": 0.9696,
+      "step": 6960
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.9152e-06,
+      "loss": 0.9806,
+      "step": 6965
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.9216e-06,
+      "loss": 0.9681,
+      "step": 6970
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.928000000000002e-06,
+      "loss": 0.9698,
+      "step": 6975
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.9344e-06,
+      "loss": 0.9781,
+      "step": 6980
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.940800000000001e-06,
+      "loss": 0.983,
+      "step": 6985
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.9472e-06,
+      "loss": 0.9878,
+      "step": 6990
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.953600000000002e-06,
+      "loss": 0.9852,
+      "step": 6995
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.96e-06,
+      "loss": 0.9707,
+      "step": 7000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.966400000000001e-06,
+      "loss": 0.9822,
+      "step": 7005
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.9728e-06,
+      "loss": 0.9905,
+      "step": 7010
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.979200000000002e-06,
+      "loss": 0.9615,
+      "step": 7015
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.9856e-06,
+      "loss": 0.9782,
+      "step": 7020
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.992000000000001e-06,
+      "loss": 0.9761,
+      "step": 7025
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.9984e-06,
+      "loss": 0.9848,
+      "step": 7030
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.0048e-06,
+      "loss": 0.986,
+      "step": 7035
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.011200000000001e-06,
+      "loss": 0.9895,
+      "step": 7040
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.017600000000002e-06,
+      "loss": 0.9893,
+      "step": 7045
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.024e-06,
+      "loss": 0.9722,
+      "step": 7050
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.0304e-06,
+      "loss": 0.981,
+      "step": 7055
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.036800000000001e-06,
+      "loss": 0.9785,
+      "step": 7060
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.0432e-06,
+      "loss": 0.9644,
+      "step": 7065
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.0496e-06,
+      "loss": 0.9773,
+      "step": 7070
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.056000000000001e-06,
+      "loss": 0.9854,
+      "step": 7075
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.062400000000002e-06,
+      "loss": 0.9844,
+      "step": 7080
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.0688e-06,
+      "loss": 0.9946,
+      "step": 7085
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.0752e-06,
+      "loss": 0.9801,
+      "step": 7090
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.0816e-06,
+      "loss": 1.0026,
+      "step": 7095
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.088000000000002e-06,
+      "loss": 0.9759,
+      "step": 7100
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.0944e-06,
+      "loss": 0.9847,
+      "step": 7105
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.100800000000001e-06,
+      "loss": 0.9812,
+      "step": 7110
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.1072e-06,
+      "loss": 0.9645,
+      "step": 7115
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.1136e-06,
+      "loss": 0.9731,
+      "step": 7120
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.12e-06,
+      "loss": 0.9727,
+      "step": 7125
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.126400000000001e-06,
+      "loss": 0.9815,
+      "step": 7130
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.1328e-06,
+      "loss": 0.9722,
+      "step": 7135
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.1392e-06,
+      "loss": 0.976,
+      "step": 7140
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.145600000000001e-06,
+      "loss": 0.9782,
+      "step": 7145
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.152000000000001e-06,
+      "loss": 0.9693,
+      "step": 7150
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.1584e-06,
+      "loss": 0.9755,
+      "step": 7155
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.1648e-06,
+      "loss": 0.9765,
+      "step": 7160
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.171200000000001e-06,
+      "loss": 0.9757,
+      "step": 7165
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.177600000000002e-06,
+      "loss": 0.9711,
+      "step": 7170
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.184e-06,
+      "loss": 0.9735,
+      "step": 7175
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.190400000000001e-06,
+      "loss": 0.9923,
+      "step": 7180
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.196800000000001e-06,
+      "loss": 0.9738,
+      "step": 7185
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.2032e-06,
+      "loss": 0.9627,
+      "step": 7190
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.2096e-06,
+      "loss": 0.9715,
+      "step": 7195
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.216000000000001e-06,
+      "loss": 0.9722,
+      "step": 7200
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.2224e-06,
+      "loss": 0.9798,
+      "step": 7205
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.2288e-06,
+      "loss": 0.982,
+      "step": 7210
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.235200000000001e-06,
+      "loss": 0.9637,
+      "step": 7215
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.2416e-06,
+      "loss": 0.9772,
+      "step": 7220
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.248e-06,
+      "loss": 0.9614,
+      "step": 7225
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.2544e-06,
+      "loss": 0.9879,
+      "step": 7230
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.260800000000001e-06,
+      "loss": 0.9632,
+      "step": 7235
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.2672e-06,
+      "loss": 0.9738,
+      "step": 7240
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.2736e-06,
+      "loss": 0.9791,
+      "step": 7245
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.280000000000001e-06,
+      "loss": 0.9843,
+      "step": 7250
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.286400000000001e-06,
+      "loss": 0.9616,
+      "step": 7255
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.2928e-06,
+      "loss": 0.9788,
+      "step": 7260
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.2992e-06,
+      "loss": 0.9705,
+      "step": 7265
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.305600000000001e-06,
+      "loss": 0.9778,
+      "step": 7270
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.312000000000002e-06,
+      "loss": 0.9529,
+      "step": 7275
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.3184e-06,
+      "loss": 0.971,
+      "step": 7280
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.324800000000001e-06,
+      "loss": 0.9659,
+      "step": 7285
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.3312e-06,
+      "loss": 0.9653,
+      "step": 7290
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.337600000000002e-06,
+      "loss": 0.9605,
+      "step": 7295
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.344e-06,
+      "loss": 0.9651,
+      "step": 7300
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.350400000000001e-06,
+      "loss": 0.9772,
+      "step": 7305
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.3568e-06,
+      "loss": 0.9741,
+      "step": 7310
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.3632e-06,
+      "loss": 0.9713,
+      "step": 7315
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.369600000000001e-06,
+      "loss": 0.9743,
+      "step": 7320
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.376000000000001e-06,
+      "loss": 0.9732,
+      "step": 7325
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.3824e-06,
+      "loss": 0.9696,
+      "step": 7330
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.3888e-06,
+      "loss": 0.9776,
+      "step": 7335
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.395200000000001e-06,
+      "loss": 0.9802,
+      "step": 7340
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.4016e-06,
+      "loss": 0.9635,
+      "step": 7345
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.408e-06,
+      "loss": 0.9762,
+      "step": 7350
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.414400000000001e-06,
+      "loss": 0.9474,
+      "step": 7355
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.420800000000001e-06,
+      "loss": 0.9805,
+      "step": 7360
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.4272e-06,
+      "loss": 0.9696,
+      "step": 7365
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.4336e-06,
+      "loss": 0.9584,
+      "step": 7370
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.440000000000001e-06,
+      "loss": 0.9665,
+      "step": 7375
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.446400000000002e-06,
+      "loss": 0.9616,
+      "step": 7380
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.4528e-06,
+      "loss": 0.9598,
+      "step": 7385
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.4592e-06,
+      "loss": 0.9786,
+      "step": 7390
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.4656e-06,
+      "loss": 0.9625,
+      "step": 7395
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.472000000000002e-06,
+      "loss": 0.9626,
+      "step": 7400
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.4784e-06,
+      "loss": 0.9677,
+      "step": 7405
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.484800000000001e-06,
+      "loss": 0.9677,
+      "step": 7410
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.4912e-06,
+      "loss": 0.983,
+      "step": 7415
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.497600000000002e-06,
+      "loss": 0.9876,
+      "step": 7420
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.504e-06,
+      "loss": 0.9888,
+      "step": 7425
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.510400000000001e-06,
+      "loss": 0.9588,
+      "step": 7430
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.5168e-06,
+      "loss": 0.9681,
+      "step": 7435
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.5232e-06,
+      "loss": 0.9686,
+      "step": 7440
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.529600000000001e-06,
+      "loss": 0.9601,
+      "step": 7445
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.536000000000002e-06,
+      "loss": 0.9679,
+      "step": 7450
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.5424e-06,
+      "loss": 0.9599,
+      "step": 7455
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.5488e-06,
+      "loss": 0.9674,
+      "step": 7460
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.555200000000001e-06,
+      "loss": 0.967,
+      "step": 7465
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.5616e-06,
+      "loss": 0.9651,
+      "step": 7470
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.568e-06,
+      "loss": 0.9649,
+      "step": 7475
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.574400000000001e-06,
+      "loss": 0.9748,
+      "step": 7480
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.580800000000002e-06,
+      "loss": 0.9587,
+      "step": 7485
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.5872e-06,
+      "loss": 0.9587,
+      "step": 7490
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.5936e-06,
+      "loss": 0.9587,
+      "step": 7495
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 0.9726,
+      "step": 7500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.606400000000002e-06,
+      "loss": 0.9519,
+      "step": 7505
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.6128e-06,
+      "loss": 0.9618,
+      "step": 7510
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.619200000000001e-06,
+      "loss": 0.9678,
+      "step": 7515
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.6256e-06,
+      "loss": 0.9728,
+      "step": 7520
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.632e-06,
+      "loss": 0.9613,
+      "step": 7525
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.6384e-06,
+      "loss": 0.9567,
+      "step": 7530
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.644800000000001e-06,
+      "loss": 0.9592,
+      "step": 7535
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.6512e-06,
+      "loss": 0.9705,
+      "step": 7540
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.6576e-06,
+      "loss": 0.9721,
+      "step": 7545
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.664000000000001e-06,
+      "loss": 0.9618,
+      "step": 7550
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.670400000000001e-06,
+      "loss": 0.9668,
+      "step": 7555
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.6768e-06,
+      "loss": 0.9686,
+      "step": 7560
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.6832e-06,
+      "loss": 0.9758,
+      "step": 7565
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.689600000000001e-06,
+      "loss": 0.9709,
+      "step": 7570
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.696000000000002e-06,
+      "loss": 0.9422,
+      "step": 7575
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.7024e-06,
+      "loss": 0.9618,
+      "step": 7580
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.708800000000001e-06,
+      "loss": 0.9605,
+      "step": 7585
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.715200000000001e-06,
+      "loss": 0.9707,
+      "step": 7590
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.7216e-06,
+      "loss": 0.9623,
+      "step": 7595
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.728e-06,
+      "loss": 0.981,
+      "step": 7600
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.734400000000001e-06,
+      "loss": 0.9803,
+      "step": 7605
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.7408e-06,
+      "loss": 0.9603,
+      "step": 7610
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.7472e-06,
+      "loss": 0.9603,
+      "step": 7615
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.753600000000001e-06,
+      "loss": 0.9659,
+      "step": 7620
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.760000000000001e-06,
+      "loss": 0.9534,
+      "step": 7625
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.7664e-06,
+      "loss": 0.9542,
+      "step": 7630
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.7728e-06,
+      "loss": 0.9696,
+      "step": 7635
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.779200000000001e-06,
+      "loss": 0.9621,
+      "step": 7640
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.7856e-06,
+      "loss": 0.9605,
+      "step": 7645
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.792e-06,
+      "loss": 0.9636,
+      "step": 7650
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.798400000000001e-06,
+      "loss": 0.9728,
+      "step": 7655
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.804800000000001e-06,
+      "loss": 0.963,
+      "step": 7660
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.8112e-06,
+      "loss": 0.9634,
+      "step": 7665
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.8176e-06,
+      "loss": 0.9702,
+      "step": 7670
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.824000000000001e-06,
+      "loss": 0.9435,
+      "step": 7675
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.830400000000002e-06,
+      "loss": 0.9528,
+      "step": 7680
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.8368e-06,
+      "loss": 0.957,
+      "step": 7685
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.843200000000001e-06,
+      "loss": 0.9733,
+      "step": 7690
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.8496e-06,
+      "loss": 0.9593,
+      "step": 7695
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.856000000000002e-06,
+      "loss": 0.9534,
+      "step": 7700
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.8624e-06,
+      "loss": 0.9466,
+      "step": 7705
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.868800000000001e-06,
+      "loss": 0.969,
+      "step": 7710
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.8752e-06,
+      "loss": 0.9587,
+      "step": 7715
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.8816e-06,
+      "loss": 0.9704,
+      "step": 7720
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.888000000000001e-06,
+      "loss": 0.9499,
+      "step": 7725
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.894400000000001e-06,
+      "loss": 0.9482,
+      "step": 7730
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.9008e-06,
+      "loss": 0.9545,
+      "step": 7735
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.9072e-06,
+      "loss": 0.9682,
+      "step": 7740
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.913600000000001e-06,
+      "loss": 0.9582,
+      "step": 7745
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.920000000000002e-06,
+      "loss": 0.9719,
+      "step": 7750
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.9264e-06,
+      "loss": 0.9541,
+      "step": 7755
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.932800000000001e-06,
+      "loss": 0.9784,
+      "step": 7760
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.939200000000001e-06,
+      "loss": 0.9601,
+      "step": 7765
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.9456e-06,
+      "loss": 0.9605,
+      "step": 7770
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.952e-06,
+      "loss": 0.9573,
+      "step": 7775
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.958400000000001e-06,
+      "loss": 0.9599,
+      "step": 7780
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.964800000000002e-06,
+      "loss": 0.9672,
+      "step": 7785
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.9712e-06,
+      "loss": 0.9496,
+      "step": 7790
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.9776e-06,
+      "loss": 0.95,
+      "step": 7795
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.984e-06,
+      "loss": 0.9563,
+      "step": 7800
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.990400000000002e-06,
+      "loss": 0.9676,
+      "step": 7805
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.9968e-06,
+      "loss": 0.9635,
+      "step": 7810
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.0003200000000001e-05,
+      "loss": 0.9633,
+      "step": 7815
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.0009600000000002e-05,
+      "loss": 0.9481,
+      "step": 7820
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.0016000000000002e-05,
+      "loss": 0.9518,
+      "step": 7825
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.00224e-05,
+      "loss": 0.9683,
+      "step": 7830
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.0028800000000001e-05,
+      "loss": 0.9573,
+      "step": 7835
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.00352e-05,
+      "loss": 0.9724,
+      "step": 7840
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.00416e-05,
+      "loss": 0.9588,
+      "step": 7845
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.0048e-05,
+      "loss": 0.9449,
+      "step": 7850
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.00544e-05,
+      "loss": 0.9617,
+      "step": 7855
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.0060800000000002e-05,
+      "loss": 0.9597,
+      "step": 7860
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.00672e-05,
+      "loss": 0.9517,
+      "step": 7865
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.0073600000000001e-05,
+      "loss": 0.946,
+      "step": 7870
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.008e-05,
+      "loss": 0.9783,
+      "step": 7875
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.00864e-05,
+      "loss": 0.9491,
+      "step": 7880
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.0092800000000001e-05,
+      "loss": 0.9552,
+      "step": 7885
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.00992e-05,
+      "loss": 0.9549,
+      "step": 7890
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.01056e-05,
+      "loss": 0.9529,
+      "step": 7895
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.0112000000000002e-05,
+      "loss": 0.9556,
+      "step": 7900
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.0118400000000001e-05,
+      "loss": 0.9455,
+      "step": 7905
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.0124800000000002e-05,
+      "loss": 0.9593,
+      "step": 7910
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.01312e-05,
+      "loss": 0.9527,
+      "step": 7915
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.0137600000000001e-05,
+      "loss": 0.9659,
+      "step": 7920
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.0144e-05,
+      "loss": 0.958,
+      "step": 7925
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.01504e-05,
+      "loss": 0.9371,
+      "step": 7930
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.01568e-05,
+      "loss": 0.9482,
+      "step": 7935
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.01632e-05,
+      "loss": 0.9508,
+      "step": 7940
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.0169600000000002e-05,
+      "loss": 0.9343,
+      "step": 7945
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.0176000000000002e-05,
+      "loss": 0.9557,
+      "step": 7950
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.0182400000000001e-05,
+      "loss": 0.9717,
+      "step": 7955
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.0188800000000001e-05,
+      "loss": 0.9672,
+      "step": 7960
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.01952e-05,
+      "loss": 0.9703,
+      "step": 7965
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.02016e-05,
+      "loss": 0.9607,
+      "step": 7970
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.0208e-05,
+      "loss": 0.9587,
+      "step": 7975
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.02144e-05,
+      "loss": 0.9574,
+      "step": 7980
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.0220800000000002e-05,
+      "loss": 0.9557,
+      "step": 7985
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.0227200000000001e-05,
+      "loss": 0.9593,
+      "step": 7990
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.0233600000000001e-05,
+      "loss": 0.9451,
+      "step": 7995
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.024e-05,
+      "loss": 0.9518,
+      "step": 8000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.02464e-05,
+      "loss": 0.9555,
+      "step": 8005
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.0252800000000001e-05,
+      "loss": 0.9423,
+      "step": 8010
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.02592e-05,
+      "loss": 0.945,
+      "step": 8015
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.02656e-05,
+      "loss": 0.9433,
+      "step": 8020
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.0272e-05,
+      "loss": 0.966,
+      "step": 8025
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.0278400000000001e-05,
+      "loss": 0.9671,
+      "step": 8030
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.0284800000000002e-05,
+      "loss": 0.953,
+      "step": 8035
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.02912e-05,
+      "loss": 0.9661,
+      "step": 8040
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.0297600000000001e-05,
+      "loss": 0.9642,
+      "step": 8045
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.0304e-05,
+      "loss": 0.9437,
+      "step": 8050
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.03104e-05,
+      "loss": 0.964,
+      "step": 8055
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.0316800000000001e-05,
+      "loss": 0.9514,
+      "step": 8060
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.03232e-05,
+      "loss": 0.9459,
+      "step": 8065
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.0329600000000002e-05,
+      "loss": 0.9568,
+      "step": 8070
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.0336000000000002e-05,
+      "loss": 0.9602,
+      "step": 8075
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.0342400000000001e-05,
+      "loss": 0.9486,
+      "step": 8080
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.0348800000000002e-05,
+      "loss": 0.958,
+      "step": 8085
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.03552e-05,
+      "loss": 0.955,
+      "step": 8090
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.0361600000000001e-05,
+      "loss": 0.9596,
+      "step": 8095
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.0368e-05,
+      "loss": 0.949,
+      "step": 8100
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.03744e-05,
+      "loss": 0.9473,
+      "step": 8105
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.0380799999999999e-05,
+      "loss": 0.9491,
+      "step": 8110
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.0387200000000001e-05,
+      "loss": 0.9548,
+      "step": 8115
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.0393600000000002e-05,
+      "loss": 0.9595,
+      "step": 8120
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.04e-05,
+      "loss": 0.9522,
+      "step": 8125
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.0406400000000001e-05,
+      "loss": 0.9483,
+      "step": 8130
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.0412800000000001e-05,
+      "loss": 0.9468,
+      "step": 8135
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.04192e-05,
+      "loss": 0.9621,
+      "step": 8140
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.04256e-05,
+      "loss": 0.9509,
+      "step": 8145
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.0432e-05,
+      "loss": 0.9377,
+      "step": 8150
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.0438400000000002e-05,
+      "loss": 0.9569,
+      "step": 8155
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.0444800000000002e-05,
+      "loss": 0.9541,
+      "step": 8160
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.04512e-05,
+      "loss": 0.9495,
+      "step": 8165
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.0457600000000001e-05,
+      "loss": 0.9559,
+      "step": 8170
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.0464e-05,
+      "loss": 0.9416,
+      "step": 8175
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.04704e-05,
+      "loss": 0.9424,
+      "step": 8180
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.0476800000000001e-05,
+      "loss": 0.9488,
+      "step": 8185
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.04832e-05,
+      "loss": 0.9415,
+      "step": 8190
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.04896e-05,
+      "loss": 0.947,
+      "step": 8195
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.0496000000000003e-05,
+      "loss": 0.9595,
+      "step": 8200
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.0502400000000001e-05,
+      "loss": 0.9499,
+      "step": 8205
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.0508800000000002e-05,
+      "loss": 0.9377,
+      "step": 8210
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.05152e-05,
+      "loss": 0.9596,
+      "step": 8215
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.0521600000000001e-05,
+      "loss": 0.9478,
+      "step": 8220
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.0528e-05,
+      "loss": 0.9482,
+      "step": 8225
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.05344e-05,
+      "loss": 0.9545,
+      "step": 8230
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.0540799999999999e-05,
+      "loss": 0.9423,
+      "step": 8235
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.0547200000000001e-05,
+      "loss": 0.9519,
+      "step": 8240
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.0553600000000002e-05,
+      "loss": 0.9448,
+      "step": 8245
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.056e-05,
+      "loss": 0.956,
+      "step": 8250
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.0566400000000001e-05,
+      "loss": 0.9515,
+      "step": 8255
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.0572800000000002e-05,
+      "loss": 0.9441,
+      "step": 8260
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.05792e-05,
+      "loss": 0.9494,
+      "step": 8265
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.05856e-05,
+      "loss": 0.947,
+      "step": 8270
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.0592e-05,
+      "loss": 0.9463,
+      "step": 8275
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.05984e-05,
+      "loss": 0.9385,
+      "step": 8280
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0604800000000002e-05,
+      "loss": 0.9471,
+      "step": 8285
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0611200000000001e-05,
+      "loss": 0.944,
+      "step": 8290
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0617600000000001e-05,
+      "loss": 0.9499,
+      "step": 8295
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0624e-05,
+      "loss": 0.9449,
+      "step": 8300
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.06304e-05,
+      "loss": 0.9607,
+      "step": 8305
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0636800000000001e-05,
+      "loss": 0.9392,
+      "step": 8310
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.06432e-05,
+      "loss": 0.9392,
+      "step": 8315
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.06496e-05,
+      "loss": 0.9497,
+      "step": 8320
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0656000000000003e-05,
+      "loss": 0.9449,
+      "step": 8325
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0662400000000001e-05,
+      "loss": 0.9511,
+      "step": 8330
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0668800000000002e-05,
+      "loss": 0.9424,
+      "step": 8335
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.06752e-05,
+      "loss": 0.9571,
+      "step": 8340
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0681600000000001e-05,
+      "loss": 0.9311,
+      "step": 8345
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0688e-05,
+      "loss": 0.9432,
+      "step": 8350
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.06944e-05,
+      "loss": 0.9471,
+      "step": 8355
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.07008e-05,
+      "loss": 0.9483,
+      "step": 8360
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.07072e-05,
+      "loss": 0.9364,
+      "step": 8365
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0713600000000002e-05,
+      "loss": 0.9446,
+      "step": 8370
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.072e-05,
+      "loss": 0.9337,
+      "step": 8375
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0726400000000001e-05,
+      "loss": 0.9391,
+      "step": 8380
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0732800000000002e-05,
+      "loss": 0.9379,
+      "step": 8385
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.07392e-05,
+      "loss": 0.9314,
+      "step": 8390
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0745600000000001e-05,
+      "loss": 0.9467,
+      "step": 8395
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0752e-05,
+      "loss": 0.9382,
+      "step": 8400
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.07584e-05,
+      "loss": 0.9311,
+      "step": 8405
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0764800000000002e-05,
+      "loss": 0.939,
+      "step": 8410
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0771200000000001e-05,
+      "loss": 0.9482,
+      "step": 8415
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0777600000000002e-05,
+      "loss": 0.9318,
+      "step": 8420
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0784e-05,
+      "loss": 0.9358,
+      "step": 8425
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0790400000000001e-05,
+      "loss": 0.9295,
+      "step": 8430
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0796800000000001e-05,
+      "loss": 0.9486,
+      "step": 8435
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.08032e-05,
+      "loss": 0.9309,
+      "step": 8440
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.08096e-05,
+      "loss": 0.9345,
+      "step": 8445
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0816e-05,
+      "loss": 0.9322,
+      "step": 8450
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0822400000000002e-05,
+      "loss": 0.9492,
+      "step": 8455
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0828800000000002e-05,
+      "loss": 0.9378,
+      "step": 8460
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0835200000000001e-05,
+      "loss": 0.9413,
+      "step": 8465
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0841600000000001e-05,
+      "loss": 0.9536,
+      "step": 8470
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0848e-05,
+      "loss": 0.9433,
+      "step": 8475
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.08544e-05,
+      "loss": 0.9465,
+      "step": 8480
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.08608e-05,
+      "loss": 0.9496,
+      "step": 8485
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.08672e-05,
+      "loss": 0.942,
+      "step": 8490
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0873600000000002e-05,
+      "loss": 0.956,
+      "step": 8495
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0880000000000001e-05,
+      "loss": 0.9357,
+      "step": 8500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0886400000000001e-05,
+      "loss": 0.9308,
+      "step": 8505
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0892800000000002e-05,
+      "loss": 0.9402,
+      "step": 8510
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.08992e-05,
+      "loss": 0.95,
+      "step": 8515
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0905600000000001e-05,
+      "loss": 0.9567,
+      "step": 8520
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0912e-05,
+      "loss": 0.9411,
+      "step": 8525
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.09184e-05,
+      "loss": 0.9344,
+      "step": 8530
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0924799999999999e-05,
+      "loss": 0.9275,
+      "step": 8535
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0931200000000001e-05,
+      "loss": 0.9222,
+      "step": 8540
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0937600000000002e-05,
+      "loss": 0.9308,
+      "step": 8545
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0944e-05,
+      "loss": 0.9452,
+      "step": 8550
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0950400000000001e-05,
+      "loss": 0.9308,
+      "step": 8555
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0956800000000002e-05,
+      "loss": 0.9473,
+      "step": 8560
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.09632e-05,
+      "loss": 0.9347,
+      "step": 8565
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.09696e-05,
+      "loss": 0.9397,
+      "step": 8570
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0976e-05,
+      "loss": 0.9494,
+      "step": 8575
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0982400000000002e-05,
+      "loss": 0.9255,
+      "step": 8580
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0988800000000002e-05,
+      "loss": 0.9379,
+      "step": 8585
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0995200000000001e-05,
+      "loss": 0.9376,
+      "step": 8590
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.1001600000000002e-05,
+      "loss": 0.9406,
+      "step": 8595
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.1008e-05,
+      "loss": 0.9508,
+      "step": 8600
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.10144e-05,
+      "loss": 0.934,
+      "step": 8605
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.10208e-05,
+      "loss": 0.935,
+      "step": 8610
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.10272e-05,
+      "loss": 0.9394,
+      "step": 8615
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.10336e-05,
+      "loss": 0.9355,
+      "step": 8620
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.1040000000000001e-05,
+      "loss": 0.9221,
+      "step": 8625
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.1046400000000002e-05,
+      "loss": 0.9369,
+      "step": 8630
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.1052800000000002e-05,
+      "loss": 0.9249,
+      "step": 8635
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.10592e-05,
+      "loss": 0.9441,
+      "step": 8640
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.1065600000000001e-05,
+      "loss": 0.9273,
+      "step": 8645
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.1072e-05,
+      "loss": 0.9354,
+      "step": 8650
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.10784e-05,
+      "loss": 0.9372,
+      "step": 8655
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.10848e-05,
+      "loss": 0.9388,
+      "step": 8660
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.1091200000000001e-05,
+      "loss": 0.9393,
+      "step": 8665
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.1097600000000002e-05,
+      "loss": 0.9377,
+      "step": 8670
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.1104e-05,
+      "loss": 0.9401,
+      "step": 8675
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.1110400000000001e-05,
+      "loss": 0.9285,
+      "step": 8680
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.1116800000000002e-05,
+      "loss": 0.9381,
+      "step": 8685
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.11232e-05,
+      "loss": 0.9389,
+      "step": 8690
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.1129600000000001e-05,
+      "loss": 0.9154,
+      "step": 8695
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.1136e-05,
+      "loss": 0.9378,
+      "step": 8700
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.11424e-05,
+      "loss": 0.9411,
+      "step": 8705
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.1148800000000002e-05,
+      "loss": 0.9287,
+      "step": 8710
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.1155200000000001e-05,
+      "loss": 0.9373,
+      "step": 8715
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.1161600000000002e-05,
+      "loss": 0.9324,
+      "step": 8720
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.1168e-05,
+      "loss": 0.9263,
+      "step": 8725
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.1174400000000001e-05,
+      "loss": 0.9363,
+      "step": 8730
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.11808e-05,
+      "loss": 0.9297,
+      "step": 8735
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.11872e-05,
+      "loss": 0.9327,
+      "step": 8740
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.11936e-05,
+      "loss": 0.9277,
+      "step": 8745
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.1200000000000001e-05,
+      "loss": 0.9308,
+      "step": 8750
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.1206400000000002e-05,
+      "loss": 0.9352,
+      "step": 8755
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.1212800000000002e-05,
+      "loss": 0.9405,
+      "step": 8760
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.1219200000000001e-05,
+      "loss": 0.9167,
+      "step": 8765
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.1225600000000001e-05,
+      "loss": 0.9364,
+      "step": 8770
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.1232e-05,
+      "loss": 0.9389,
+      "step": 8775
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.12384e-05,
+      "loss": 0.9274,
+      "step": 8780
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.12448e-05,
+      "loss": 0.935,
+      "step": 8785
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.12512e-05,
+      "loss": 0.9218,
+      "step": 8790
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.1257600000000002e-05,
+      "loss": 0.9417,
+      "step": 8795
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.1264000000000001e-05,
+      "loss": 0.9173,
+      "step": 8800
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.1270400000000001e-05,
+      "loss": 0.9304,
+      "step": 8805
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.1276800000000002e-05,
+      "loss": 0.9273,
+      "step": 8810
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.12832e-05,
+      "loss": 0.931,
+      "step": 8815
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.1289600000000001e-05,
+      "loss": 0.9311,
+      "step": 8820
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.1296e-05,
+      "loss": 0.9355,
+      "step": 8825
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.13024e-05,
+      "loss": 0.9451,
+      "step": 8830
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.1308800000000003e-05,
+      "loss": 0.9275,
+      "step": 8835
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.1315200000000001e-05,
+      "loss": 0.9271,
+      "step": 8840
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.1321600000000002e-05,
+      "loss": 0.9398,
+      "step": 8845
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.1328e-05,
+      "loss": 0.935,
+      "step": 8850
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.1334400000000001e-05,
+      "loss": 0.9322,
+      "step": 8855
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.13408e-05,
+      "loss": 0.9445,
+      "step": 8860
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.13472e-05,
+      "loss": 0.9463,
+      "step": 8865
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.1353600000000001e-05,
+      "loss": 0.9276,
+      "step": 8870
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.136e-05,
+      "loss": 0.9246,
+      "step": 8875
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.1366400000000002e-05,
+      "loss": 0.9211,
+      "step": 8880
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.1372800000000002e-05,
+      "loss": 0.9354,
+      "step": 8885
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.1379200000000001e-05,
+      "loss": 0.9196,
+      "step": 8890
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.1385600000000002e-05,
+      "loss": 0.9265,
+      "step": 8895
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.1392e-05,
+      "loss": 0.9225,
+      "step": 8900
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.13984e-05,
+      "loss": 0.9132,
+      "step": 8905
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.14048e-05,
+      "loss": 0.9381,
+      "step": 8910
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.14112e-05,
+      "loss": 0.9352,
+      "step": 8915
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.1417600000000002e-05,
+      "loss": 0.9213,
+      "step": 8920
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.1424000000000001e-05,
+      "loss": 0.9346,
+      "step": 8925
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.1430400000000002e-05,
+      "loss": 0.9241,
+      "step": 8930
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.1436800000000002e-05,
+      "loss": 0.9441,
+      "step": 8935
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.14432e-05,
+      "loss": 0.9261,
+      "step": 8940
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.1449600000000001e-05,
+      "loss": 0.948,
+      "step": 8945
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.1456e-05,
+      "loss": 0.9225,
+      "step": 8950
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.14624e-05,
+      "loss": 0.9266,
+      "step": 8955
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.14688e-05,
+      "loss": 0.9222,
+      "step": 8960
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.1475200000000002e-05,
+      "loss": 0.9162,
+      "step": 8965
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.1481600000000002e-05,
+      "loss": 0.9375,
+      "step": 8970
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.1488e-05,
+      "loss": 0.9293,
+      "step": 8975
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.1494400000000001e-05,
+      "loss": 0.9237,
+      "step": 8980
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.15008e-05,
+      "loss": 0.9372,
+      "step": 8985
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.15072e-05,
+      "loss": 0.931,
+      "step": 8990
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.1513600000000001e-05,
+      "loss": 0.9332,
+      "step": 8995
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.152e-05,
+      "loss": 0.941,
+      "step": 9000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.1526400000000002e-05,
+      "loss": 0.9396,
+      "step": 9005
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.1532800000000002e-05,
+      "loss": 0.9224,
+      "step": 9010
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.1539200000000001e-05,
+      "loss": 0.9137,
+      "step": 9015
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.1545600000000002e-05,
+      "loss": 0.926,
+      "step": 9020
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.1552e-05,
+      "loss": 0.9151,
+      "step": 9025
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.1558400000000001e-05,
+      "loss": 0.9464,
+      "step": 9030
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.15648e-05,
+      "loss": 0.9108,
+      "step": 9035
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.15712e-05,
+      "loss": 0.9154,
+      "step": 9040
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.15776e-05,
+      "loss": 0.9171,
+      "step": 9045
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.1584000000000001e-05,
+      "loss": 0.9357,
+      "step": 9050
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.1590400000000002e-05,
+      "loss": 0.937,
+      "step": 9055
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.1596800000000002e-05,
+      "loss": 0.9269,
+      "step": 9060
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.1603200000000001e-05,
+      "loss": 0.9198,
+      "step": 9065
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.1609600000000001e-05,
+      "loss": 0.9186,
+      "step": 9070
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.1616e-05,
+      "loss": 0.9168,
+      "step": 9075
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.16224e-05,
+      "loss": 0.9208,
+      "step": 9080
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.16288e-05,
+      "loss": 0.9182,
+      "step": 9085
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.1635200000000002e-05,
+      "loss": 0.8972,
+      "step": 9090
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.1641600000000002e-05,
+      "loss": 0.9152,
+      "step": 9095
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.1648000000000001e-05,
+      "loss": 0.9268,
+      "step": 9100
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.1654400000000001e-05,
+      "loss": 0.9175,
+      "step": 9105
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.16608e-05,
+      "loss": 0.929,
+      "step": 9110
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.16672e-05,
+      "loss": 0.9327,
+      "step": 9115
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.1673600000000001e-05,
+      "loss": 0.9225,
+      "step": 9120
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.168e-05,
+      "loss": 0.9307,
+      "step": 9125
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.16864e-05,
+      "loss": 0.9197,
+      "step": 9130
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.1692800000000003e-05,
+      "loss": 0.912,
+      "step": 9135
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.1699200000000001e-05,
+      "loss": 0.9253,
+      "step": 9140
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.1705600000000002e-05,
+      "loss": 0.9102,
+      "step": 9145
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.1712e-05,
+      "loss": 0.9178,
+      "step": 9150
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.1718400000000001e-05,
+      "loss": 0.9193,
+      "step": 9155
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.17248e-05,
+      "loss": 0.923,
+      "step": 9160
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.17312e-05,
+      "loss": 0.9238,
+      "step": 9165
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.1737600000000001e-05,
+      "loss": 0.9092,
+      "step": 9170
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.1744000000000001e-05,
+      "loss": 0.9113,
+      "step": 9175
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.1750400000000002e-05,
+      "loss": 0.9125,
+      "step": 9180
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.1756800000000002e-05,
+      "loss": 0.9262,
+      "step": 9185
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.1763200000000001e-05,
+      "loss": 0.9306,
+      "step": 9190
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.1769600000000002e-05,
+      "loss": 0.9174,
+      "step": 9195
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.1776e-05,
+      "loss": 0.9411,
+      "step": 9200
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.1782400000000001e-05,
+      "loss": 0.917,
+      "step": 9205
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.17888e-05,
+      "loss": 0.9258,
+      "step": 9210
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.17952e-05,
+      "loss": 0.9398,
+      "step": 9215
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.1801600000000002e-05,
+      "loss": 0.9216,
+      "step": 9220
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.1808000000000001e-05,
+      "loss": 0.9212,
+      "step": 9225
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.1814400000000002e-05,
+      "loss": 0.9243,
+      "step": 9230
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.18208e-05,
+      "loss": 0.9274,
+      "step": 9235
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.1827200000000001e-05,
+      "loss": 0.9141,
+      "step": 9240
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.1833600000000001e-05,
+      "loss": 0.9285,
+      "step": 9245
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.184e-05,
+      "loss": 0.9148,
+      "step": 9250
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.18464e-05,
+      "loss": 0.9232,
+      "step": 9255
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.18528e-05,
+      "loss": 0.9245,
+      "step": 9260
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.1859200000000002e-05,
+      "loss": 0.9254,
+      "step": 9265
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.1865600000000002e-05,
+      "loss": 0.9029,
+      "step": 9270
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.1872000000000001e-05,
+      "loss": 0.9173,
+      "step": 9275
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.1878400000000001e-05,
+      "loss": 0.9191,
+      "step": 9280
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.18848e-05,
+      "loss": 0.9216,
+      "step": 9285
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.18912e-05,
+      "loss": 0.9317,
+      "step": 9290
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.1897600000000001e-05,
+      "loss": 0.9226,
+      "step": 9295
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.1904e-05,
+      "loss": 0.9029,
+      "step": 9300
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.1910400000000002e-05,
+      "loss": 0.9145,
+      "step": 9305
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.1916800000000003e-05,
+      "loss": 0.9303,
+      "step": 9310
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.1923200000000001e-05,
+      "loss": 0.9234,
+      "step": 9315
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.1929600000000002e-05,
+      "loss": 0.9294,
+      "step": 9320
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.1936e-05,
+      "loss": 0.9136,
+      "step": 9325
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.1942400000000001e-05,
+      "loss": 0.9274,
+      "step": 9330
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.19488e-05,
+      "loss": 0.9115,
+      "step": 9335
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.19552e-05,
+      "loss": 0.9221,
+      "step": 9340
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.1961599999999999e-05,
+      "loss": 0.9223,
+      "step": 9345
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.1968000000000001e-05,
+      "loss": 0.9185,
+      "step": 9350
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.1974400000000002e-05,
+      "loss": 0.9242,
+      "step": 9355
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.19808e-05,
+      "loss": 0.9235,
+      "step": 9360
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.1987200000000001e-05,
+      "loss": 0.928,
+      "step": 9365
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.1993600000000002e-05,
+      "loss": 0.9298,
+      "step": 9370
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2e-05,
+      "loss": 0.917,
+      "step": 9375
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.20064e-05,
+      "loss": 0.9229,
+      "step": 9380
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.20128e-05,
+      "loss": 0.9261,
+      "step": 9385
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2019200000000002e-05,
+      "loss": 0.928,
+      "step": 9390
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2025600000000002e-05,
+      "loss": 0.9266,
+      "step": 9395
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2032000000000001e-05,
+      "loss": 0.9187,
+      "step": 9400
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2038400000000001e-05,
+      "loss": 0.9241,
+      "step": 9405
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.20448e-05,
+      "loss": 0.898,
+      "step": 9410
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.20512e-05,
+      "loss": 0.9211,
+      "step": 9415
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2057600000000001e-05,
+      "loss": 0.9024,
+      "step": 9420
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2064e-05,
+      "loss": 0.8996,
+      "step": 9425
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.20704e-05,
+      "loss": 0.9287,
+      "step": 9430
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2076800000000003e-05,
+      "loss": 0.9059,
+      "step": 9435
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2083200000000001e-05,
+      "loss": 0.9273,
+      "step": 9440
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2089600000000002e-05,
+      "loss": 0.9205,
+      "step": 9445
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2096e-05,
+      "loss": 0.9062,
+      "step": 9450
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2102400000000001e-05,
+      "loss": 0.9211,
+      "step": 9455
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.21088e-05,
+      "loss": 0.9095,
+      "step": 9460
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.21152e-05,
+      "loss": 0.9167,
+      "step": 9465
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.21216e-05,
+      "loss": 0.9106,
+      "step": 9470
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2128000000000001e-05,
+      "loss": 0.9264,
+      "step": 9475
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2134400000000002e-05,
+      "loss": 0.9231,
+      "step": 9480
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.21408e-05,
+      "loss": 0.916,
+      "step": 9485
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2147200000000001e-05,
+      "loss": 0.9295,
+      "step": 9490
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2153600000000002e-05,
+      "loss": 0.9193,
+      "step": 9495
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.216e-05,
+      "loss": 0.9068,
+      "step": 9500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2166400000000001e-05,
+      "loss": 0.9131,
+      "step": 9505
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.21728e-05,
+      "loss": 0.9205,
+      "step": 9510
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.21792e-05,
+      "loss": 0.9184,
+      "step": 9515
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2185600000000002e-05,
+      "loss": 0.9186,
+      "step": 9520
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2192000000000001e-05,
+      "loss": 0.9118,
+      "step": 9525
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2198400000000002e-05,
+      "loss": 0.9275,
+      "step": 9530
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.22048e-05,
+      "loss": 0.9175,
+      "step": 9535
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2211200000000001e-05,
+      "loss": 0.9116,
+      "step": 9540
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2217600000000001e-05,
+      "loss": 0.9236,
+      "step": 9545
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2224e-05,
+      "loss": 0.9215,
+      "step": 9550
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.22304e-05,
+      "loss": 0.9102,
+      "step": 9555
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2236800000000003e-05,
+      "loss": 0.9051,
+      "step": 9560
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2243200000000002e-05,
+      "loss": 0.9134,
+      "step": 9565
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2249600000000002e-05,
+      "loss": 0.9131,
+      "step": 9570
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2256000000000001e-05,
+      "loss": 0.9237,
+      "step": 9575
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2262400000000001e-05,
+      "loss": 0.9128,
+      "step": 9580
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.22688e-05,
+      "loss": 0.9137,
+      "step": 9585
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.22752e-05,
+      "loss": 0.9218,
+      "step": 9590
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.22816e-05,
+      "loss": 0.9023,
+      "step": 9595
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2288e-05,
+      "loss": 0.927,
+      "step": 9600
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2294400000000002e-05,
+      "loss": 0.932,
+      "step": 9605
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2300800000000001e-05,
+      "loss": 0.9262,
+      "step": 9610
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2307200000000001e-05,
+      "loss": 0.9077,
+      "step": 9615
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2313600000000002e-05,
+      "loss": 0.9221,
+      "step": 9620
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.232e-05,
+      "loss": 0.9222,
+      "step": 9625
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2326400000000001e-05,
+      "loss": 0.9179,
+      "step": 9630
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.23328e-05,
+      "loss": 0.9082,
+      "step": 9635
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.23392e-05,
+      "loss": 0.9231,
+      "step": 9640
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2345600000000003e-05,
+      "loss": 0.9272,
+      "step": 9645
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2352000000000001e-05,
+      "loss": 0.9067,
+      "step": 9650
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2358400000000002e-05,
+      "loss": 0.8991,
+      "step": 9655
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.23648e-05,
+      "loss": 0.9077,
+      "step": 9660
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2371200000000001e-05,
+      "loss": 0.8967,
+      "step": 9665
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2377600000000002e-05,
+      "loss": 0.9157,
+      "step": 9670
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2384e-05,
+      "loss": 0.9213,
+      "step": 9675
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.23904e-05,
+      "loss": 0.9032,
+      "step": 9680
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.23968e-05,
+      "loss": 0.9013,
+      "step": 9685
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2403200000000002e-05,
+      "loss": 0.9094,
+      "step": 9690
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2409600000000002e-05,
+      "loss": 0.9064,
+      "step": 9695
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2416000000000001e-05,
+      "loss": 0.9015,
+      "step": 9700
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2422400000000002e-05,
+      "loss": 0.9107,
+      "step": 9705
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.24288e-05,
+      "loss": 0.9085,
+      "step": 9710
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.24352e-05,
+      "loss": 0.9192,
+      "step": 9715
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.24416e-05,
+      "loss": 0.9013,
+      "step": 9720
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2448e-05,
+      "loss": 0.9125,
+      "step": 9725
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2454400000000002e-05,
+      "loss": 0.9074,
+      "step": 9730
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2460800000000001e-05,
+      "loss": 0.9249,
+      "step": 9735
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2467200000000002e-05,
+      "loss": 0.9085,
+      "step": 9740
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2473600000000002e-05,
+      "loss": 0.921,
+      "step": 9745
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.248e-05,
+      "loss": 0.9129,
+      "step": 9750
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2486400000000001e-05,
+      "loss": 0.9149,
+      "step": 9755
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.24928e-05,
+      "loss": 0.9057,
+      "step": 9760
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.24992e-05,
+      "loss": 0.8884,
+      "step": 9765
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.25056e-05,
+      "loss": 0.915,
+      "step": 9770
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2512000000000002e-05,
+      "loss": 0.9077,
+      "step": 9775
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2518400000000002e-05,
+      "loss": 0.905,
+      "step": 9780
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.25248e-05,
+      "loss": 0.9086,
+      "step": 9785
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2531200000000001e-05,
+      "loss": 0.9147,
+      "step": 9790
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.25376e-05,
+      "loss": 0.906,
+      "step": 9795
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2544e-05,
+      "loss": 0.9018,
+      "step": 9800
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2550400000000001e-05,
+      "loss": 0.8873,
+      "step": 9805
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.25568e-05,
+      "loss": 0.9107,
+      "step": 9810
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2563200000000002e-05,
+      "loss": 0.9026,
+      "step": 9815
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2569600000000002e-05,
+      "loss": 0.9167,
+      "step": 9820
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2576000000000001e-05,
+      "loss": 0.9083,
+      "step": 9825
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2582400000000002e-05,
+      "loss": 0.8961,
+      "step": 9830
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.25888e-05,
+      "loss": 0.8995,
+      "step": 9835
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2595200000000001e-05,
+      "loss": 0.9042,
+      "step": 9840
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.26016e-05,
+      "loss": 0.9055,
+      "step": 9845
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.2608e-05,
+      "loss": 0.9124,
+      "step": 9850
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.26144e-05,
+      "loss": 0.9263,
+      "step": 9855
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.2620800000000001e-05,
+      "loss": 0.9172,
+      "step": 9860
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.2627200000000002e-05,
+      "loss": 0.8868,
+      "step": 9865
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.2633600000000002e-05,
+      "loss": 0.9049,
+      "step": 9870
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.2640000000000001e-05,
+      "loss": 0.8953,
+      "step": 9875
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.2646400000000001e-05,
+      "loss": 0.9065,
+      "step": 9880
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.26528e-05,
+      "loss": 0.9143,
+      "step": 9885
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.26592e-05,
+      "loss": 0.8907,
+      "step": 9890
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.26656e-05,
+      "loss": 0.8969,
+      "step": 9895
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.2672000000000002e-05,
+      "loss": 0.8968,
+      "step": 9900
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.2678400000000002e-05,
+      "loss": 0.9009,
+      "step": 9905
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.2684800000000001e-05,
+      "loss": 0.9027,
+      "step": 9910
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.2691200000000001e-05,
+      "loss": 0.9068,
+      "step": 9915
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.26976e-05,
+      "loss": 0.8971,
+      "step": 9920
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.2704e-05,
+      "loss": 0.9023,
+      "step": 9925
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.2710400000000001e-05,
+      "loss": 0.9051,
+      "step": 9930
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.27168e-05,
+      "loss": 0.9101,
+      "step": 9935
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.27232e-05,
+      "loss": 0.905,
+      "step": 9940
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.2729600000000003e-05,
+      "loss": 0.8991,
+      "step": 9945
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.2736000000000001e-05,
+      "loss": 0.8991,
+      "step": 9950
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.2742400000000002e-05,
+      "loss": 0.9173,
+      "step": 9955
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.27488e-05,
+      "loss": 0.9042,
+      "step": 9960
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.2755200000000001e-05,
+      "loss": 0.9377,
+      "step": 9965
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.27616e-05,
+      "loss": 0.8843,
+      "step": 9970
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.2768e-05,
+      "loss": 0.9002,
+      "step": 9975
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.2774400000000001e-05,
+      "loss": 0.9132,
+      "step": 9980
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.2780800000000001e-05,
+      "loss": 0.8997,
+      "step": 9985
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.2787200000000002e-05,
+      "loss": 0.9071,
+      "step": 9990
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.2793600000000002e-05,
+      "loss": 0.9108,
+      "step": 9995
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.2800000000000001e-05,
+      "loss": 0.8915,
+      "step": 10000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.2806400000000002e-05,
+      "loss": 0.9139,
+      "step": 10005
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.28128e-05,
+      "loss": 0.8944,
+      "step": 10010
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.2819200000000001e-05,
+      "loss": 0.8891,
+      "step": 10015
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.28256e-05,
+      "loss": 0.8982,
+      "step": 10020
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.2832e-05,
+      "loss": 0.9087,
+      "step": 10025
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.2838400000000002e-05,
+      "loss": 0.9203,
+      "step": 10030
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.2844800000000001e-05,
+      "loss": 0.8993,
+      "step": 10035
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.2851200000000002e-05,
+      "loss": 0.8844,
+      "step": 10040
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.28576e-05,
+      "loss": 0.8984,
+      "step": 10045
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.2864e-05,
+      "loss": 0.8979,
+      "step": 10050
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.2870400000000001e-05,
+      "loss": 0.8995,
+      "step": 10055
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.28768e-05,
+      "loss": 0.9099,
+      "step": 10060
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.28832e-05,
+      "loss": 0.9072,
+      "step": 10065
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.2889600000000003e-05,
+      "loss": 0.8984,
+      "step": 10070
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.2896000000000002e-05,
+      "loss": 0.9012,
+      "step": 10075
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.2902400000000002e-05,
+      "loss": 0.8996,
+      "step": 10080
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.29088e-05,
+      "loss": 0.9115,
+      "step": 10085
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.2915200000000001e-05,
+      "loss": 0.892,
+      "step": 10090
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.29216e-05,
+      "loss": 0.9036,
+      "step": 10095
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.2928e-05,
+      "loss": 0.8999,
+      "step": 10100
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.2934400000000001e-05,
+      "loss": 0.9015,
+      "step": 10105
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.29408e-05,
+      "loss": 0.9041,
+      "step": 10110
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.2947200000000002e-05,
+      "loss": 0.8987,
+      "step": 10115
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.2953600000000002e-05,
+      "loss": 0.9014,
+      "step": 10120
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.2960000000000001e-05,
+      "loss": 0.8922,
+      "step": 10125
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.2966400000000002e-05,
+      "loss": 0.8928,
+      "step": 10130
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.29728e-05,
+      "loss": 0.8986,
+      "step": 10135
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.2979200000000001e-05,
+      "loss": 0.8968,
+      "step": 10140
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.29856e-05,
+      "loss": 0.8996,
+      "step": 10145
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.2992e-05,
+      "loss": 0.8947,
+      "step": 10150
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.2998400000000002e-05,
+      "loss": 0.902,
+      "step": 10155
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3004800000000001e-05,
+      "loss": 0.8956,
+      "step": 10160
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3011200000000002e-05,
+      "loss": 0.906,
+      "step": 10165
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.30176e-05,
+      "loss": 0.8974,
+      "step": 10170
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3024000000000001e-05,
+      "loss": 0.8932,
+      "step": 10175
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3030400000000001e-05,
+      "loss": 0.8941,
+      "step": 10180
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.30368e-05,
+      "loss": 0.8992,
+      "step": 10185
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.30432e-05,
+      "loss": 0.9014,
+      "step": 10190
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.30496e-05,
+      "loss": 0.8956,
+      "step": 10195
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3056000000000002e-05,
+      "loss": 0.9015,
+      "step": 10200
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3062400000000002e-05,
+      "loss": 0.9172,
+      "step": 10205
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3068800000000001e-05,
+      "loss": 0.8982,
+      "step": 10210
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3075200000000001e-05,
+      "loss": 0.887,
+      "step": 10215
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.30816e-05,
+      "loss": 0.8936,
+      "step": 10220
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3088e-05,
+      "loss": 0.8933,
+      "step": 10225
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3094400000000001e-05,
+      "loss": 0.8932,
+      "step": 10230
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.31008e-05,
+      "loss": 0.9066,
+      "step": 10235
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3107200000000002e-05,
+      "loss": 0.9054,
+      "step": 10240
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3113600000000003e-05,
+      "loss": 0.9054,
+      "step": 10245
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3120000000000001e-05,
+      "loss": 0.8841,
+      "step": 10250
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3126400000000002e-05,
+      "loss": 0.893,
+      "step": 10255
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.31328e-05,
+      "loss": 0.9032,
+      "step": 10260
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3139200000000001e-05,
+      "loss": 0.8861,
+      "step": 10265
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.31456e-05,
+      "loss": 0.8973,
+      "step": 10270
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3152e-05,
+      "loss": 0.8916,
+      "step": 10275
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.31584e-05,
+      "loss": 0.8998,
+      "step": 10280
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3164800000000001e-05,
+      "loss": 0.9097,
+      "step": 10285
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3171200000000002e-05,
+      "loss": 0.8925,
+      "step": 10290
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.31776e-05,
+      "loss": 0.9011,
+      "step": 10295
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3184000000000001e-05,
+      "loss": 0.9051,
+      "step": 10300
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3190400000000002e-05,
+      "loss": 0.8907,
+      "step": 10305
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.31968e-05,
+      "loss": 0.9045,
+      "step": 10310
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3203200000000001e-05,
+      "loss": 0.8973,
+      "step": 10315
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.32096e-05,
+      "loss": 0.9093,
+      "step": 10320
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3216000000000002e-05,
+      "loss": 0.9144,
+      "step": 10325
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3222400000000002e-05,
+      "loss": 0.8914,
+      "step": 10330
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3228800000000001e-05,
+      "loss": 0.8904,
+      "step": 10335
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3235200000000002e-05,
+      "loss": 0.9105,
+      "step": 10340
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.32416e-05,
+      "loss": 0.9057,
+      "step": 10345
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3248000000000001e-05,
+      "loss": 0.9008,
+      "step": 10350
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3254400000000001e-05,
+      "loss": 0.8903,
+      "step": 10355
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.32608e-05,
+      "loss": 0.9035,
+      "step": 10360
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.32672e-05,
+      "loss": 0.8682,
+      "step": 10365
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3273600000000003e-05,
+      "loss": 0.8956,
+      "step": 10370
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3280000000000002e-05,
+      "loss": 0.8888,
+      "step": 10375
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3286400000000002e-05,
+      "loss": 0.8963,
+      "step": 10380
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3292800000000001e-05,
+      "loss": 0.8928,
+      "step": 10385
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3299200000000001e-05,
+      "loss": 0.8888,
+      "step": 10390
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.33056e-05,
+      "loss": 0.8873,
+      "step": 10395
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3312e-05,
+      "loss": 0.9129,
+      "step": 10400
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.33184e-05,
+      "loss": 0.895,
+      "step": 10405
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3324800000000002e-05,
+      "loss": 0.8897,
+      "step": 10410
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3331200000000002e-05,
+      "loss": 0.8886,
+      "step": 10415
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.33376e-05,
+      "loss": 0.8916,
+      "step": 10420
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3344000000000001e-05,
+      "loss": 0.9042,
+      "step": 10425
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3350400000000002e-05,
+      "loss": 0.9021,
+      "step": 10430
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.33568e-05,
+      "loss": 0.8909,
+      "step": 10435
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3363200000000001e-05,
+      "loss": 0.9035,
+      "step": 10440
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.33696e-05,
+      "loss": 0.8906,
+      "step": 10445
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3376e-05,
+      "loss": 0.8947,
+      "step": 10450
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3382400000000003e-05,
+      "loss": 0.9037,
+      "step": 10455
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3388800000000001e-05,
+      "loss": 0.8924,
+      "step": 10460
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3395200000000002e-05,
+      "loss": 0.9094,
+      "step": 10465
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.34016e-05,
+      "loss": 0.875,
+      "step": 10470
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3408000000000001e-05,
+      "loss": 0.8961,
+      "step": 10475
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3414400000000002e-05,
+      "loss": 0.8841,
+      "step": 10480
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.34208e-05,
+      "loss": 0.8884,
+      "step": 10485
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.34272e-05,
+      "loss": 0.8942,
+      "step": 10490
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3433600000000003e-05,
+      "loss": 0.893,
+      "step": 10495
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3440000000000002e-05,
+      "loss": 0.8864,
+      "step": 10500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3446400000000002e-05,
+      "loss": 0.8647,
+      "step": 10505
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3452800000000001e-05,
+      "loss": 0.8883,
+      "step": 10510
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3459200000000002e-05,
+      "loss": 0.8849,
+      "step": 10515
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.34656e-05,
+      "loss": 0.9009,
+      "step": 10520
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3472e-05,
+      "loss": 0.8747,
+      "step": 10525
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.34784e-05,
+      "loss": 0.887,
+      "step": 10530
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.34848e-05,
+      "loss": 0.888,
+      "step": 10535
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3491200000000002e-05,
+      "loss": 0.8838,
+      "step": 10540
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3497600000000001e-05,
+      "loss": 0.8978,
+      "step": 10545
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3504000000000001e-05,
+      "loss": 0.8814,
+      "step": 10550
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3510400000000002e-05,
+      "loss": 0.8932,
+      "step": 10555
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.35168e-05,
+      "loss": 0.8947,
+      "step": 10560
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3523200000000001e-05,
+      "loss": 0.8925,
+      "step": 10565
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.35296e-05,
+      "loss": 0.8764,
+      "step": 10570
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3536e-05,
+      "loss": 0.8913,
+      "step": 10575
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3542400000000003e-05,
+      "loss": 0.8893,
+      "step": 10580
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3548800000000001e-05,
+      "loss": 0.9028,
+      "step": 10585
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3555200000000002e-05,
+      "loss": 0.8939,
+      "step": 10590
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.35616e-05,
+      "loss": 0.8937,
+      "step": 10595
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3568000000000001e-05,
+      "loss": 0.89,
+      "step": 10600
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3574400000000002e-05,
+      "loss": 0.8855,
+      "step": 10605
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.35808e-05,
+      "loss": 0.8795,
+      "step": 10610
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3587200000000001e-05,
+      "loss": 0.8955,
+      "step": 10615
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.35936e-05,
+      "loss": 0.8859,
+      "step": 10620
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3600000000000002e-05,
+      "loss": 0.8937,
+      "step": 10625
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3606400000000002e-05,
+      "loss": 0.8842,
+      "step": 10630
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3612800000000001e-05,
+      "loss": 0.9083,
+      "step": 10635
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3619200000000002e-05,
+      "loss": 0.8776,
+      "step": 10640
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.36256e-05,
+      "loss": 0.8911,
+      "step": 10645
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3632000000000001e-05,
+      "loss": 0.9006,
+      "step": 10650
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.36384e-05,
+      "loss": 0.8794,
+      "step": 10655
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.36448e-05,
+      "loss": 0.895,
+      "step": 10660
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.36512e-05,
+      "loss": 0.8938,
+      "step": 10665
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3657600000000001e-05,
+      "loss": 0.8885,
+      "step": 10670
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3664000000000002e-05,
+      "loss": 0.8899,
+      "step": 10675
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3670400000000002e-05,
+      "loss": 0.8974,
+      "step": 10680
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3676800000000001e-05,
+      "loss": 0.8844,
+      "step": 10685
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3683200000000001e-05,
+      "loss": 0.9052,
+      "step": 10690
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.36896e-05,
+      "loss": 0.8925,
+      "step": 10695
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3696e-05,
+      "loss": 0.8584,
+      "step": 10700
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.37024e-05,
+      "loss": 0.8801,
+      "step": 10705
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3708800000000002e-05,
+      "loss": 0.8746,
+      "step": 10710
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3715200000000002e-05,
+      "loss": 0.8947,
+      "step": 10715
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3721600000000001e-05,
+      "loss": 0.8897,
+      "step": 10720
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3728000000000001e-05,
+      "loss": 0.8944,
+      "step": 10725
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3734400000000002e-05,
+      "loss": 0.8913,
+      "step": 10730
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.37408e-05,
+      "loss": 0.8884,
+      "step": 10735
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3747200000000001e-05,
+      "loss": 0.8765,
+      "step": 10740
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.37536e-05,
+      "loss": 0.8909,
+      "step": 10745
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.376e-05,
+      "loss": 0.8928,
+      "step": 10750
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3766400000000003e-05,
+      "loss": 0.8694,
+      "step": 10755
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3772800000000001e-05,
+      "loss": 0.8824,
+      "step": 10760
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3779200000000002e-05,
+      "loss": 0.8904,
+      "step": 10765
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.37856e-05,
+      "loss": 0.8989,
+      "step": 10770
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3792000000000001e-05,
+      "loss": 0.8722,
+      "step": 10775
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.37984e-05,
+      "loss": 0.8918,
+      "step": 10780
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.38048e-05,
+      "loss": 0.8864,
+      "step": 10785
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.3811200000000001e-05,
+      "loss": 0.8812,
+      "step": 10790
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.3817600000000001e-05,
+      "loss": 0.8763,
+      "step": 10795
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.3824000000000002e-05,
+      "loss": 0.8849,
+      "step": 10800
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.3830400000000002e-05,
+      "loss": 0.8876,
+      "step": 10805
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.3836800000000001e-05,
+      "loss": 0.8805,
+      "step": 10810
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.3843200000000002e-05,
+      "loss": 0.8907,
+      "step": 10815
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.38496e-05,
+      "loss": 0.8727,
+      "step": 10820
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.3856e-05,
+      "loss": 0.873,
+      "step": 10825
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.38624e-05,
+      "loss": 0.8913,
+      "step": 10830
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.38688e-05,
+      "loss": 0.8873,
+      "step": 10835
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.3875200000000002e-05,
+      "loss": 0.8699,
+      "step": 10840
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.3881600000000001e-05,
+      "loss": 0.8777,
+      "step": 10845
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.3888000000000002e-05,
+      "loss": 0.8758,
+      "step": 10850
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.3894400000000002e-05,
+      "loss": 0.8966,
+      "step": 10855
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.39008e-05,
+      "loss": 0.8867,
+      "step": 10860
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.3907200000000001e-05,
+      "loss": 0.8881,
+      "step": 10865
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.39136e-05,
+      "loss": 0.887,
+      "step": 10870
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.392e-05,
+      "loss": 0.8844,
+      "step": 10875
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.3926400000000003e-05,
+      "loss": 0.8985,
+      "step": 10880
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.3932800000000002e-05,
+      "loss": 0.8694,
+      "step": 10885
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.3939200000000002e-05,
+      "loss": 0.8871,
+      "step": 10890
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.39456e-05,
+      "loss": 0.8858,
+      "step": 10895
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.3952000000000001e-05,
+      "loss": 0.8774,
+      "step": 10900
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.39584e-05,
+      "loss": 0.8787,
+      "step": 10905
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.39648e-05,
+      "loss": 0.8843,
+      "step": 10910
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.3971200000000001e-05,
+      "loss": 0.8954,
+      "step": 10915
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.39776e-05,
+      "loss": 0.8858,
+      "step": 10920
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.3984000000000002e-05,
+      "loss": 0.8776,
+      "step": 10925
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.3990400000000002e-05,
+      "loss": 0.8876,
+      "step": 10930
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.3996800000000001e-05,
+      "loss": 0.8831,
+      "step": 10935
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4003200000000002e-05,
+      "loss": 0.8885,
+      "step": 10940
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.40096e-05,
+      "loss": 0.8852,
+      "step": 10945
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4016000000000001e-05,
+      "loss": 0.8904,
+      "step": 10950
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.40224e-05,
+      "loss": 0.8809,
+      "step": 10955
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.40288e-05,
+      "loss": 0.889,
+      "step": 10960
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4035200000000002e-05,
+      "loss": 0.8751,
+      "step": 10965
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4041600000000001e-05,
+      "loss": 0.8711,
+      "step": 10970
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4048000000000002e-05,
+      "loss": 0.8929,
+      "step": 10975
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4054400000000002e-05,
+      "loss": 0.8937,
+      "step": 10980
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4060800000000001e-05,
+      "loss": 0.881,
+      "step": 10985
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4067200000000001e-05,
+      "loss": 0.8847,
+      "step": 10990
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.40736e-05,
+      "loss": 0.885,
+      "step": 10995
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.408e-05,
+      "loss": 0.8764,
+      "step": 11000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.40864e-05,
+      "loss": 0.8761,
+      "step": 11005
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4092800000000002e-05,
+      "loss": 0.8655,
+      "step": 11010
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4099200000000002e-05,
+      "loss": 0.8817,
+      "step": 11015
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4105600000000001e-05,
+      "loss": 0.8809,
+      "step": 11020
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4112000000000001e-05,
+      "loss": 0.8836,
+      "step": 11025
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.41184e-05,
+      "loss": 0.881,
+      "step": 11030
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.41248e-05,
+      "loss": 0.8952,
+      "step": 11035
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4131200000000001e-05,
+      "loss": 0.8836,
+      "step": 11040
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.41376e-05,
+      "loss": 0.8929,
+      "step": 11045
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4144000000000002e-05,
+      "loss": 0.8994,
+      "step": 11050
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4150400000000003e-05,
+      "loss": 0.8899,
+      "step": 11055
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4156800000000001e-05,
+      "loss": 0.8771,
+      "step": 11060
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4163200000000002e-05,
+      "loss": 0.8796,
+      "step": 11065
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.41696e-05,
+      "loss": 0.8878,
+      "step": 11070
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4176000000000001e-05,
+      "loss": 0.8794,
+      "step": 11075
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.41824e-05,
+      "loss": 0.883,
+      "step": 11080
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.41888e-05,
+      "loss": 0.8757,
+      "step": 11085
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4195200000000001e-05,
+      "loss": 0.878,
+      "step": 11090
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4201600000000001e-05,
+      "loss": 0.8723,
+      "step": 11095
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4208000000000002e-05,
+      "loss": 0.8807,
+      "step": 11100
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4214400000000002e-05,
+      "loss": 0.8707,
+      "step": 11105
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4220800000000001e-05,
+      "loss": 0.8782,
+      "step": 11110
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4227200000000002e-05,
+      "loss": 0.8701,
+      "step": 11115
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.42336e-05,
+      "loss": 0.8817,
+      "step": 11120
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4240000000000001e-05,
+      "loss": 0.8782,
+      "step": 11125
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.42464e-05,
+      "loss": 0.8796,
+      "step": 11130
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4252800000000002e-05,
+      "loss": 0.8741,
+      "step": 11135
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4259200000000002e-05,
+      "loss": 0.88,
+      "step": 11140
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4265600000000001e-05,
+      "loss": 0.8824,
+      "step": 11145
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4272000000000002e-05,
+      "loss": 0.8909,
+      "step": 11150
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.42784e-05,
+      "loss": 0.8851,
+      "step": 11155
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4284800000000001e-05,
+      "loss": 0.8687,
+      "step": 11160
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4291200000000001e-05,
+      "loss": 0.8837,
+      "step": 11165
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.42976e-05,
+      "loss": 0.8847,
+      "step": 11170
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4304e-05,
+      "loss": 0.8693,
+      "step": 11175
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4310400000000003e-05,
+      "loss": 0.8849,
+      "step": 11180
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4316800000000002e-05,
+      "loss": 0.8762,
+      "step": 11185
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4323200000000002e-05,
+      "loss": 0.8892,
+      "step": 11190
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.43296e-05,
+      "loss": 0.8773,
+      "step": 11195
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4336000000000001e-05,
+      "loss": 0.8719,
+      "step": 11200
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.43424e-05,
+      "loss": 0.8862,
+      "step": 11205
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.43488e-05,
+      "loss": 0.8792,
+      "step": 11210
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4355200000000001e-05,
+      "loss": 0.8937,
+      "step": 11215
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4361600000000002e-05,
+      "loss": 0.8678,
+      "step": 11220
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4368000000000002e-05,
+      "loss": 0.8753,
+      "step": 11225
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4374400000000003e-05,
+      "loss": 0.8645,
+      "step": 11230
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4380800000000001e-05,
+      "loss": 0.8675,
+      "step": 11235
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4387200000000002e-05,
+      "loss": 0.8674,
+      "step": 11240
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.43936e-05,
+      "loss": 0.8764,
+      "step": 11245
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4400000000000001e-05,
+      "loss": 0.8618,
+      "step": 11250
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.44064e-05,
+      "loss": 0.8752,
+      "step": 11255
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.44128e-05,
+      "loss": 0.8736,
+      "step": 11260
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4419200000000002e-05,
+      "loss": 0.8728,
+      "step": 11265
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4425600000000001e-05,
+      "loss": 0.8732,
+      "step": 11270
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4432000000000002e-05,
+      "loss": 0.875,
+      "step": 11275
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.44384e-05,
+      "loss": 0.8535,
+      "step": 11280
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4444800000000001e-05,
+      "loss": 0.8847,
+      "step": 11285
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4451200000000002e-05,
+      "loss": 0.8757,
+      "step": 11290
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.44576e-05,
+      "loss": 0.8608,
+      "step": 11295
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4464e-05,
+      "loss": 0.8687,
+      "step": 11300
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4470400000000003e-05,
+      "loss": 0.8982,
+      "step": 11305
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4476800000000002e-05,
+      "loss": 0.8786,
+      "step": 11310
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4483200000000002e-05,
+      "loss": 0.8846,
+      "step": 11315
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4489600000000001e-05,
+      "loss": 0.8762,
+      "step": 11320
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4496000000000001e-05,
+      "loss": 0.8704,
+      "step": 11325
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.45024e-05,
+      "loss": 0.8655,
+      "step": 11330
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.45088e-05,
+      "loss": 0.8827,
+      "step": 11335
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4515200000000001e-05,
+      "loss": 0.8808,
+      "step": 11340
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.45216e-05,
+      "loss": 0.8686,
+      "step": 11345
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4528000000000002e-05,
+      "loss": 0.871,
+      "step": 11350
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4534400000000003e-05,
+      "loss": 0.883,
+      "step": 11355
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4540800000000001e-05,
+      "loss": 0.8749,
+      "step": 11360
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4547200000000002e-05,
+      "loss": 0.8802,
+      "step": 11365
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.45536e-05,
+      "loss": 0.8818,
+      "step": 11370
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4560000000000001e-05,
+      "loss": 0.8716,
+      "step": 11375
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.45664e-05,
+      "loss": 0.8789,
+      "step": 11380
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.45728e-05,
+      "loss": 0.8783,
+      "step": 11385
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4579200000000003e-05,
+      "loss": 0.8681,
+      "step": 11390
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4585600000000001e-05,
+      "loss": 0.8707,
+      "step": 11395
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4592000000000002e-05,
+      "loss": 0.8665,
+      "step": 11400
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.45984e-05,
+      "loss": 0.8926,
+      "step": 11405
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4604800000000001e-05,
+      "loss": 0.8729,
+      "step": 11410
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4611200000000002e-05,
+      "loss": 0.8811,
+      "step": 11415
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.46176e-05,
+      "loss": 0.8823,
+      "step": 11420
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4624000000000001e-05,
+      "loss": 0.8832,
+      "step": 11425
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.46304e-05,
+      "loss": 0.8746,
+      "step": 11430
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4636800000000002e-05,
+      "loss": 0.8845,
+      "step": 11435
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4643200000000002e-05,
+      "loss": 0.8639,
+      "step": 11440
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4649600000000001e-05,
+      "loss": 0.8679,
+      "step": 11445
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4656000000000002e-05,
+      "loss": 0.8749,
+      "step": 11450
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.46624e-05,
+      "loss": 0.8602,
+      "step": 11455
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4668800000000001e-05,
+      "loss": 0.864,
+      "step": 11460
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4675200000000001e-05,
+      "loss": 0.8632,
+      "step": 11465
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.46816e-05,
+      "loss": 0.8753,
+      "step": 11470
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4688000000000002e-05,
+      "loss": 0.8707,
+      "step": 11475
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4694400000000003e-05,
+      "loss": 0.8685,
+      "step": 11480
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4700800000000002e-05,
+      "loss": 0.8751,
+      "step": 11485
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4707200000000002e-05,
+      "loss": 0.866,
+      "step": 11490
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4713600000000001e-05,
+      "loss": 0.8794,
+      "step": 11495
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4720000000000001e-05,
+      "loss": 0.8592,
+      "step": 11500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.47264e-05,
+      "loss": 0.8665,
+      "step": 11505
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.47328e-05,
+      "loss": 0.8701,
+      "step": 11510
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.47392e-05,
+      "loss": 0.8772,
+      "step": 11515
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4745600000000002e-05,
+      "loss": 0.8805,
+      "step": 11520
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4752000000000002e-05,
+      "loss": 0.8699,
+      "step": 11525
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4758400000000001e-05,
+      "loss": 0.8747,
+      "step": 11530
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4764800000000001e-05,
+      "loss": 0.8655,
+      "step": 11535
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4771200000000002e-05,
+      "loss": 0.8636,
+      "step": 11540
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.47776e-05,
+      "loss": 0.8666,
+      "step": 11545
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4784000000000001e-05,
+      "loss": 0.8676,
+      "step": 11550
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.47904e-05,
+      "loss": 0.8739,
+      "step": 11555
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4796800000000002e-05,
+      "loss": 0.8625,
+      "step": 11560
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4803200000000003e-05,
+      "loss": 0.8808,
+      "step": 11565
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4809600000000001e-05,
+      "loss": 0.8587,
+      "step": 11570
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4816000000000002e-05,
+      "loss": 0.8636,
+      "step": 11575
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.48224e-05,
+      "loss": 0.8629,
+      "step": 11580
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4828800000000001e-05,
+      "loss": 0.8564,
+      "step": 11585
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4835200000000002e-05,
+      "loss": 0.8913,
+      "step": 11590
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.48416e-05,
+      "loss": 0.8724,
+      "step": 11595
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4848e-05,
+      "loss": 0.8739,
+      "step": 11600
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4854400000000003e-05,
+      "loss": 0.8689,
+      "step": 11605
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4860800000000002e-05,
+      "loss": 0.8602,
+      "step": 11610
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4867200000000002e-05,
+      "loss": 0.8689,
+      "step": 11615
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4873600000000001e-05,
+      "loss": 0.8757,
+      "step": 11620
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4880000000000002e-05,
+      "loss": 0.8693,
+      "step": 11625
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.48864e-05,
+      "loss": 0.8738,
+      "step": 11630
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.48928e-05,
+      "loss": 0.8735,
+      "step": 11635
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.48992e-05,
+      "loss": 0.8587,
+      "step": 11640
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4905600000000002e-05,
+      "loss": 0.8732,
+      "step": 11645
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4912000000000002e-05,
+      "loss": 0.8544,
+      "step": 11650
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4918400000000001e-05,
+      "loss": 0.8576,
+      "step": 11655
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4924800000000002e-05,
+      "loss": 0.8788,
+      "step": 11660
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4931200000000002e-05,
+      "loss": 0.8654,
+      "step": 11665
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.49376e-05,
+      "loss": 0.8721,
+      "step": 11670
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4944000000000001e-05,
+      "loss": 0.8674,
+      "step": 11675
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.49504e-05,
+      "loss": 0.8778,
+      "step": 11680
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.49568e-05,
+      "loss": 0.8673,
+      "step": 11685
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4963200000000003e-05,
+      "loss": 0.8684,
+      "step": 11690
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4969600000000001e-05,
+      "loss": 0.8632,
+      "step": 11695
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4976000000000002e-05,
+      "loss": 0.8746,
+      "step": 11700
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.49824e-05,
+      "loss": 0.8715,
+      "step": 11705
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4988800000000001e-05,
+      "loss": 0.8731,
+      "step": 11710
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4995200000000002e-05,
+      "loss": 0.8759,
+      "step": 11715
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.50016e-05,
+      "loss": 0.8707,
+      "step": 11720
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.5008000000000001e-05,
+      "loss": 0.8678,
+      "step": 11725
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.5014400000000001e-05,
+      "loss": 0.8699,
+      "step": 11730
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.5020800000000002e-05,
+      "loss": 0.8649,
+      "step": 11735
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.5027200000000002e-05,
+      "loss": 0.8721,
+      "step": 11740
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.5033600000000001e-05,
+      "loss": 0.8694,
+      "step": 11745
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.5040000000000002e-05,
+      "loss": 0.8584,
+      "step": 11750
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.50464e-05,
+      "loss": 0.8805,
+      "step": 11755
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.5052800000000001e-05,
+      "loss": 0.8687,
+      "step": 11760
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.50592e-05,
+      "loss": 0.8668,
+      "step": 11765
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.50656e-05,
+      "loss": 0.846,
+      "step": 11770
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.5072000000000002e-05,
+      "loss": 0.8664,
+      "step": 11775
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.5078400000000001e-05,
+      "loss": 0.868,
+      "step": 11780
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.5084800000000002e-05,
+      "loss": 0.8618,
+      "step": 11785
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.5091200000000002e-05,
+      "loss": 0.8687,
+      "step": 11790
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.5097600000000001e-05,
+      "loss": 0.8734,
+      "step": 11795
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.5104000000000001e-05,
+      "loss": 0.8628,
+      "step": 11800
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.51104e-05,
+      "loss": 0.8628,
+      "step": 11805
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.51168e-05,
+      "loss": 0.8706,
+      "step": 11810
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.5123200000000003e-05,
+      "loss": 0.8714,
+      "step": 11815
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.5129600000000002e-05,
+      "loss": 0.8713,
+      "step": 11820
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.5136000000000002e-05,
+      "loss": 0.8615,
+      "step": 11825
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.5142400000000001e-05,
+      "loss": 0.8592,
+      "step": 11830
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.5148800000000001e-05,
+      "loss": 0.8593,
+      "step": 11835
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.51552e-05,
+      "loss": 0.8467,
+      "step": 11840
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.51616e-05,
+      "loss": 0.8556,
+      "step": 11845
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.5168000000000001e-05,
+      "loss": 0.8636,
+      "step": 11850
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.51744e-05,
+      "loss": 0.8526,
+      "step": 11855
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.5180800000000002e-05,
+      "loss": 0.8652,
+      "step": 11860
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.5187200000000003e-05,
+      "loss": 0.8567,
+      "step": 11865
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.5193600000000001e-05,
+      "loss": 0.8545,
+      "step": 11870
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.5200000000000002e-05,
+      "loss": 0.8876,
+      "step": 11875
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.52064e-05,
+      "loss": 0.873,
+      "step": 11880
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.5212800000000001e-05,
+      "loss": 0.856,
+      "step": 11885
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.52192e-05,
+      "loss": 0.8618,
+      "step": 11890
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.52256e-05,
+      "loss": 0.8571,
+      "step": 11895
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.5232000000000003e-05,
+      "loss": 0.8542,
+      "step": 11900
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.5238400000000001e-05,
+      "loss": 0.857,
+      "step": 11905
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.5244800000000002e-05,
+      "loss": 0.8711,
+      "step": 11910
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.5251200000000002e-05,
+      "loss": 0.8653,
+      "step": 11915
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.5257600000000001e-05,
+      "loss": 0.8591,
+      "step": 11920
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.5264e-05,
+      "loss": 0.8574,
+      "step": 11925
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.5270400000000002e-05,
+      "loss": 0.8592,
+      "step": 11930
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.52768e-05,
+      "loss": 0.8653,
+      "step": 11935
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.52832e-05,
+      "loss": 0.8716,
+      "step": 11940
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.5289600000000004e-05,
+      "loss": 0.8832,
+      "step": 11945
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.5296e-05,
+      "loss": 0.8701,
+      "step": 11950
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.53024e-05,
+      "loss": 0.8643,
+      "step": 11955
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.53088e-05,
+      "loss": 0.8659,
+      "step": 11960
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.5315200000000002e-05,
+      "loss": 0.8872,
+      "step": 11965
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.5321600000000003e-05,
+      "loss": 0.8671,
+      "step": 11970
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.5328e-05,
+      "loss": 0.8699,
+      "step": 11975
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.53344e-05,
+      "loss": 0.865,
+      "step": 11980
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.5340800000000004e-05,
+      "loss": 0.8639,
+      "step": 11985
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.53472e-05,
+      "loss": 0.8504,
+      "step": 11990
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.53536e-05,
+      "loss": 0.8714,
+      "step": 11995
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.5360000000000002e-05,
+      "loss": 0.8624,
+      "step": 12000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.5366400000000002e-05,
+      "loss": 0.8706,
+      "step": 12005
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.53728e-05,
+      "loss": 0.863,
+      "step": 12010
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.53792e-05,
+      "loss": 0.8778,
+      "step": 12015
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.53856e-05,
+      "loss": 0.8794,
+      "step": 12020
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.5392e-05,
+      "loss": 0.863,
+      "step": 12025
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.53984e-05,
+      "loss": 0.8576,
+      "step": 12030
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5404800000000002e-05,
+      "loss": 0.8626,
+      "step": 12035
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5411200000000002e-05,
+      "loss": 0.8556,
+      "step": 12040
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5417600000000003e-05,
+      "loss": 0.8436,
+      "step": 12045
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5424e-05,
+      "loss": 0.8655,
+      "step": 12050
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.54304e-05,
+      "loss": 0.8631,
+      "step": 12055
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.54368e-05,
+      "loss": 0.868,
+      "step": 12060
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.54432e-05,
+      "loss": 0.8657,
+      "step": 12065
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5449600000000002e-05,
+      "loss": 0.858,
+      "step": 12070
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5456000000000002e-05,
+      "loss": 0.8656,
+      "step": 12075
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5462400000000003e-05,
+      "loss": 0.8584,
+      "step": 12080
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.54688e-05,
+      "loss": 0.8471,
+      "step": 12085
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.54752e-05,
+      "loss": 0.8571,
+      "step": 12090
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.54816e-05,
+      "loss": 0.8547,
+      "step": 12095
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5488e-05,
+      "loss": 0.8729,
+      "step": 12100
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5494400000000002e-05,
+      "loss": 0.8682,
+      "step": 12105
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.55008e-05,
+      "loss": 0.8559,
+      "step": 12110
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5507200000000003e-05,
+      "loss": 0.8639,
+      "step": 12115
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5513600000000003e-05,
+      "loss": 0.8555,
+      "step": 12120
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.552e-05,
+      "loss": 0.8574,
+      "step": 12125
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.55264e-05,
+      "loss": 0.8623,
+      "step": 12130
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.55328e-05,
+      "loss": 0.8655,
+      "step": 12135
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5539200000000002e-05,
+      "loss": 0.8604,
+      "step": 12140
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.55456e-05,
+      "loss": 0.8628,
+      "step": 12145
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5552e-05,
+      "loss": 0.8471,
+      "step": 12150
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.55584e-05,
+      "loss": 0.8698,
+      "step": 12155
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.55648e-05,
+      "loss": 0.8561,
+      "step": 12160
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.55712e-05,
+      "loss": 0.8602,
+      "step": 12165
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.55776e-05,
+      "loss": 0.8536,
+      "step": 12170
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5584000000000002e-05,
+      "loss": 0.8674,
+      "step": 12175
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5590400000000002e-05,
+      "loss": 0.8667,
+      "step": 12180
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.55968e-05,
+      "loss": 0.852,
+      "step": 12185
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.56032e-05,
+      "loss": 0.8586,
+      "step": 12190
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.56096e-05,
+      "loss": 0.8561,
+      "step": 12195
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5616e-05,
+      "loss": 0.8529,
+      "step": 12200
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.56224e-05,
+      "loss": 0.8586,
+      "step": 12205
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5628800000000002e-05,
+      "loss": 0.8549,
+      "step": 12210
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5635200000000002e-05,
+      "loss": 0.8509,
+      "step": 12215
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5641600000000003e-05,
+      "loss": 0.8707,
+      "step": 12220
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5648e-05,
+      "loss": 0.8684,
+      "step": 12225
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.56544e-05,
+      "loss": 0.865,
+      "step": 12230
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.56608e-05,
+      "loss": 0.8534,
+      "step": 12235
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.56672e-05,
+      "loss": 0.8701,
+      "step": 12240
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5673600000000002e-05,
+      "loss": 0.8476,
+      "step": 12245
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5680000000000002e-05,
+      "loss": 0.8584,
+      "step": 12250
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5686400000000003e-05,
+      "loss": 0.853,
+      "step": 12255
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.56928e-05,
+      "loss": 0.8693,
+      "step": 12260
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.56992e-05,
+      "loss": 0.8523,
+      "step": 12265
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.57056e-05,
+      "loss": 0.8683,
+      "step": 12270
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5712e-05,
+      "loss": 0.8569,
+      "step": 12275
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5718400000000002e-05,
+      "loss": 0.8496,
+      "step": 12280
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5724800000000002e-05,
+      "loss": 0.8606,
+      "step": 12285
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5731200000000003e-05,
+      "loss": 0.8777,
+      "step": 12290
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5737600000000003e-05,
+      "loss": 0.8419,
+      "step": 12295
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5744e-05,
+      "loss": 0.8528,
+      "step": 12300
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.57504e-05,
+      "loss": 0.8637,
+      "step": 12305
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.57568e-05,
+      "loss": 0.8646,
+      "step": 12310
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5763200000000002e-05,
+      "loss": 0.869,
+      "step": 12315
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.57696e-05,
+      "loss": 0.8551,
+      "step": 12320
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5776e-05,
+      "loss": 0.8652,
+      "step": 12325
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5782400000000003e-05,
+      "loss": 0.8645,
+      "step": 12330
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.57888e-05,
+      "loss": 0.8595,
+      "step": 12335
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.57952e-05,
+      "loss": 0.8501,
+      "step": 12340
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.58016e-05,
+      "loss": 0.8583,
+      "step": 12345
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5808000000000002e-05,
+      "loss": 0.8744,
+      "step": 12350
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5814400000000002e-05,
+      "loss": 0.8608,
+      "step": 12355
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.58208e-05,
+      "loss": 0.8556,
+      "step": 12360
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.58272e-05,
+      "loss": 0.8554,
+      "step": 12365
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5833600000000004e-05,
+      "loss": 0.8425,
+      "step": 12370
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.584e-05,
+      "loss": 0.8558,
+      "step": 12375
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.58464e-05,
+      "loss": 0.8594,
+      "step": 12380
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5852800000000002e-05,
+      "loss": 0.8439,
+      "step": 12385
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5859200000000002e-05,
+      "loss": 0.8461,
+      "step": 12390
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.58656e-05,
+      "loss": 0.8661,
+      "step": 12395
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5872e-05,
+      "loss": 0.8695,
+      "step": 12400
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.58784e-05,
+      "loss": 0.8494,
+      "step": 12405
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.58848e-05,
+      "loss": 0.8566,
+      "step": 12410
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.58912e-05,
+      "loss": 0.8615,
+      "step": 12415
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5897600000000002e-05,
+      "loss": 0.8553,
+      "step": 12420
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5904000000000002e-05,
+      "loss": 0.8465,
+      "step": 12425
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5910400000000003e-05,
+      "loss": 0.8532,
+      "step": 12430
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.59168e-05,
+      "loss": 0.849,
+      "step": 12435
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.59232e-05,
+      "loss": 0.8599,
+      "step": 12440
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.59296e-05,
+      "loss": 0.8631,
+      "step": 12445
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5936e-05,
+      "loss": 0.8571,
+      "step": 12450
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5942400000000002e-05,
+      "loss": 0.8572,
+      "step": 12455
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5948800000000002e-05,
+      "loss": 0.8501,
+      "step": 12460
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5955200000000003e-05,
+      "loss": 0.8467,
+      "step": 12465
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5961600000000003e-05,
+      "loss": 0.8675,
+      "step": 12470
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5968e-05,
+      "loss": 0.8548,
+      "step": 12475
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.59744e-05,
+      "loss": 0.8594,
+      "step": 12480
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.59808e-05,
+      "loss": 0.8452,
+      "step": 12485
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5987200000000002e-05,
+      "loss": 0.8532,
+      "step": 12490
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.59936e-05,
+      "loss": 0.8555,
+      "step": 12495
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.8615,
+      "step": 12500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6006400000000003e-05,
+      "loss": 0.8471,
+      "step": 12505
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.60128e-05,
+      "loss": 0.8555,
+      "step": 12510
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.60192e-05,
+      "loss": 0.8453,
+      "step": 12515
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.60256e-05,
+      "loss": 0.8502,
+      "step": 12520
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6032e-05,
+      "loss": 0.8445,
+      "step": 12525
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6038400000000002e-05,
+      "loss": 0.8469,
+      "step": 12530
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.60448e-05,
+      "loss": 0.8498,
+      "step": 12535
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6051200000000003e-05,
+      "loss": 0.8576,
+      "step": 12540
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6057600000000004e-05,
+      "loss": 0.8501,
+      "step": 12545
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6064e-05,
+      "loss": 0.8534,
+      "step": 12550
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.60704e-05,
+      "loss": 0.861,
+      "step": 12555
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.60768e-05,
+      "loss": 0.8544,
+      "step": 12560
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6083200000000002e-05,
+      "loss": 0.8338,
+      "step": 12565
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.60896e-05,
+      "loss": 0.8596,
+      "step": 12570
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6096e-05,
+      "loss": 0.8688,
+      "step": 12575
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.61024e-05,
+      "loss": 0.8517,
+      "step": 12580
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.61088e-05,
+      "loss": 0.8502,
+      "step": 12585
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.61152e-05,
+      "loss": 0.851,
+      "step": 12590
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.61216e-05,
+      "loss": 0.862,
+      "step": 12595
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6128000000000002e-05,
+      "loss": 0.8616,
+      "step": 12600
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6134400000000003e-05,
+      "loss": 0.848,
+      "step": 12605
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.61408e-05,
+      "loss": 0.8418,
+      "step": 12610
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.61472e-05,
+      "loss": 0.8567,
+      "step": 12615
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.61536e-05,
+      "loss": 0.8545,
+      "step": 12620
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.616e-05,
+      "loss": 0.8522,
+      "step": 12625
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.61664e-05,
+      "loss": 0.8513,
+      "step": 12630
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6172800000000002e-05,
+      "loss": 0.8665,
+      "step": 12635
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6179200000000003e-05,
+      "loss": 0.8537,
+      "step": 12640
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.61856e-05,
+      "loss": 0.8581,
+      "step": 12645
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6192e-05,
+      "loss": 0.8471,
+      "step": 12650
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.61984e-05,
+      "loss": 0.8557,
+      "step": 12655
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.62048e-05,
+      "loss": 0.8577,
+      "step": 12660
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.62112e-05,
+      "loss": 0.8494,
+      "step": 12665
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6217600000000002e-05,
+      "loss": 0.8476,
+      "step": 12670
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6224000000000003e-05,
+      "loss": 0.8333,
+      "step": 12675
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6230400000000003e-05,
+      "loss": 0.8592,
+      "step": 12680
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.62368e-05,
+      "loss": 0.8592,
+      "step": 12685
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.62432e-05,
+      "loss": 0.8462,
+      "step": 12690
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.62496e-05,
+      "loss": 0.8662,
+      "step": 12695
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6256e-05,
+      "loss": 0.8598,
+      "step": 12700
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6262400000000002e-05,
+      "loss": 0.8409,
+      "step": 12705
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6268800000000003e-05,
+      "loss": 0.8552,
+      "step": 12710
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6275200000000003e-05,
+      "loss": 0.8476,
+      "step": 12715
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6281600000000003e-05,
+      "loss": 0.8488,
+      "step": 12720
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6288e-05,
+      "loss": 0.8546,
+      "step": 12725
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.62944e-05,
+      "loss": 0.8441,
+      "step": 12730
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.63008e-05,
+      "loss": 0.8534,
+      "step": 12735
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6307200000000002e-05,
+      "loss": 0.8496,
+      "step": 12740
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.63136e-05,
+      "loss": 0.844,
+      "step": 12745
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.632e-05,
+      "loss": 0.8457,
+      "step": 12750
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6326400000000003e-05,
+      "loss": 0.8594,
+      "step": 12755
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.63328e-05,
+      "loss": 0.8597,
+      "step": 12760
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.63392e-05,
+      "loss": 0.841,
+      "step": 12765
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.63456e-05,
+      "loss": 0.8607,
+      "step": 12770
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6352000000000002e-05,
+      "loss": 0.8301,
+      "step": 12775
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6358400000000002e-05,
+      "loss": 0.8445,
+      "step": 12780
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.63648e-05,
+      "loss": 0.8404,
+      "step": 12785
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.63712e-05,
+      "loss": 0.8545,
+      "step": 12790
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6377600000000004e-05,
+      "loss": 0.8532,
+      "step": 12795
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6384e-05,
+      "loss": 0.8558,
+      "step": 12800
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.63904e-05,
+      "loss": 0.8465,
+      "step": 12805
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6396800000000002e-05,
+      "loss": 0.8459,
+      "step": 12810
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6403200000000002e-05,
+      "loss": 0.8424,
+      "step": 12815
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.64096e-05,
+      "loss": 0.8409,
+      "step": 12820
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6416e-05,
+      "loss": 0.8441,
+      "step": 12825
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.64224e-05,
+      "loss": 0.8437,
+      "step": 12830
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.64288e-05,
+      "loss": 0.8489,
+      "step": 12835
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.64352e-05,
+      "loss": 0.8468,
+      "step": 12840
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6441600000000002e-05,
+      "loss": 0.8446,
+      "step": 12845
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6448000000000002e-05,
+      "loss": 0.8517,
+      "step": 12850
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6454400000000003e-05,
+      "loss": 0.8513,
+      "step": 12855
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.64608e-05,
+      "loss": 0.8444,
+      "step": 12860
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.64672e-05,
+      "loss": 0.8428,
+      "step": 12865
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.64736e-05,
+      "loss": 0.8445,
+      "step": 12870
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.648e-05,
+      "loss": 0.8584,
+      "step": 12875
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6486400000000002e-05,
+      "loss": 0.8393,
+      "step": 12880
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6492800000000002e-05,
+      "loss": 0.8401,
+      "step": 12885
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6499200000000003e-05,
+      "loss": 0.8489,
+      "step": 12890
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.65056e-05,
+      "loss": 0.8663,
+      "step": 12895
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6512e-05,
+      "loss": 0.8541,
+      "step": 12900
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.65184e-05,
+      "loss": 0.8419,
+      "step": 12905
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.65248e-05,
+      "loss": 0.8382,
+      "step": 12910
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6531200000000002e-05,
+      "loss": 0.8482,
+      "step": 12915
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.65376e-05,
+      "loss": 0.8531,
+      "step": 12920
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6544000000000003e-05,
+      "loss": 0.8519,
+      "step": 12925
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6550400000000003e-05,
+      "loss": 0.8506,
+      "step": 12930
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.65568e-05,
+      "loss": 0.8425,
+      "step": 12935
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.65632e-05,
+      "loss": 0.8519,
+      "step": 12940
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.65696e-05,
+      "loss": 0.8315,
+      "step": 12945
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6576000000000002e-05,
+      "loss": 0.852,
+      "step": 12950
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6582400000000002e-05,
+      "loss": 0.865,
+      "step": 12955
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.65888e-05,
+      "loss": 0.8403,
+      "step": 12960
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6595200000000003e-05,
+      "loss": 0.8565,
+      "step": 12965
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.6601600000000004e-05,
+      "loss": 0.8518,
+      "step": 12970
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.6608e-05,
+      "loss": 0.8308,
+      "step": 12975
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.66144e-05,
+      "loss": 0.8519,
+      "step": 12980
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.6620800000000002e-05,
+      "loss": 0.8368,
+      "step": 12985
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.6627200000000002e-05,
+      "loss": 0.8295,
+      "step": 12990
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.66336e-05,
+      "loss": 0.8455,
+      "step": 12995
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.664e-05,
+      "loss": 0.8491,
+      "step": 13000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.66464e-05,
+      "loss": 0.8466,
+      "step": 13005
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.66528e-05,
+      "loss": 0.8259,
+      "step": 13010
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.66592e-05,
+      "loss": 0.8609,
+      "step": 13015
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.6665600000000002e-05,
+      "loss": 0.8392,
+      "step": 13020
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.6672000000000002e-05,
+      "loss": 0.8337,
+      "step": 13025
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.6678400000000003e-05,
+      "loss": 0.8452,
+      "step": 13030
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.66848e-05,
+      "loss": 0.8362,
+      "step": 13035
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.66912e-05,
+      "loss": 0.832,
+      "step": 13040
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.66976e-05,
+      "loss": 0.8538,
+      "step": 13045
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.6704e-05,
+      "loss": 0.8519,
+      "step": 13050
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.6710400000000002e-05,
+      "loss": 0.8486,
+      "step": 13055
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.6716800000000002e-05,
+      "loss": 0.8375,
+      "step": 13060
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.6723200000000003e-05,
+      "loss": 0.8443,
+      "step": 13065
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.67296e-05,
+      "loss": 0.847,
+      "step": 13070
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.6736e-05,
+      "loss": 0.8426,
+      "step": 13075
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.67424e-05,
+      "loss": 0.8566,
+      "step": 13080
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.67488e-05,
+      "loss": 0.8459,
+      "step": 13085
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.6755200000000002e-05,
+      "loss": 0.8294,
+      "step": 13090
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.6761600000000002e-05,
+      "loss": 0.8421,
+      "step": 13095
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.6768000000000003e-05,
+      "loss": 0.8391,
+      "step": 13100
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.6774400000000003e-05,
+      "loss": 0.8439,
+      "step": 13105
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.67808e-05,
+      "loss": 0.8449,
+      "step": 13110
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.67872e-05,
+      "loss": 0.8493,
+      "step": 13115
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.67936e-05,
+      "loss": 0.8456,
+      "step": 13120
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.6800000000000002e-05,
+      "loss": 0.8449,
+      "step": 13125
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.68064e-05,
+      "loss": 0.8359,
+      "step": 13130
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.6812800000000003e-05,
+      "loss": 0.8436,
+      "step": 13135
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.6819200000000003e-05,
+      "loss": 0.84,
+      "step": 13140
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.68256e-05,
+      "loss": 0.8582,
+      "step": 13145
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.6832e-05,
+      "loss": 0.837,
+      "step": 13150
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.68384e-05,
+      "loss": 0.84,
+      "step": 13155
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.6844800000000002e-05,
+      "loss": 0.8384,
+      "step": 13160
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.6851200000000002e-05,
+      "loss": 0.8378,
+      "step": 13165
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.68576e-05,
+      "loss": 0.8475,
+      "step": 13170
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.6864e-05,
+      "loss": 0.8378,
+      "step": 13175
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.6870400000000004e-05,
+      "loss": 0.8571,
+      "step": 13180
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.68768e-05,
+      "loss": 0.8323,
+      "step": 13185
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.68832e-05,
+      "loss": 0.845,
+      "step": 13190
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.6889600000000002e-05,
+      "loss": 0.831,
+      "step": 13195
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.6896000000000002e-05,
+      "loss": 0.8358,
+      "step": 13200
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.6902400000000003e-05,
+      "loss": 0.8402,
+      "step": 13205
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.69088e-05,
+      "loss": 0.8409,
+      "step": 13210
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.69152e-05,
+      "loss": 0.8434,
+      "step": 13215
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.6921600000000004e-05,
+      "loss": 0.8425,
+      "step": 13220
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.6928e-05,
+      "loss": 0.8511,
+      "step": 13225
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.6934400000000002e-05,
+      "loss": 0.829,
+      "step": 13230
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.6940800000000002e-05,
+      "loss": 0.8404,
+      "step": 13235
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.6947200000000003e-05,
+      "loss": 0.8536,
+      "step": 13240
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.69536e-05,
+      "loss": 0.8322,
+      "step": 13245
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.696e-05,
+      "loss": 0.8599,
+      "step": 13250
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.69664e-05,
+      "loss": 0.8611,
+      "step": 13255
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.69728e-05,
+      "loss": 0.852,
+      "step": 13260
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.69792e-05,
+      "loss": 0.8429,
+      "step": 13265
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.6985600000000002e-05,
+      "loss": 0.8358,
+      "step": 13270
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.6992000000000003e-05,
+      "loss": 0.8437,
+      "step": 13275
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.6998400000000003e-05,
+      "loss": 0.8438,
+      "step": 13280
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.70048e-05,
+      "loss": 0.8245,
+      "step": 13285
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.70112e-05,
+      "loss": 0.842,
+      "step": 13290
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.70176e-05,
+      "loss": 0.8406,
+      "step": 13295
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7024e-05,
+      "loss": 0.8451,
+      "step": 13300
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7030400000000002e-05,
+      "loss": 0.839,
+      "step": 13305
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7036800000000003e-05,
+      "loss": 0.8245,
+      "step": 13310
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7043200000000003e-05,
+      "loss": 0.8367,
+      "step": 13315
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.70496e-05,
+      "loss": 0.8473,
+      "step": 13320
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7056e-05,
+      "loss": 0.8377,
+      "step": 13325
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.70624e-05,
+      "loss": 0.8472,
+      "step": 13330
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.70688e-05,
+      "loss": 0.8496,
+      "step": 13335
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7075200000000002e-05,
+      "loss": 0.8464,
+      "step": 13340
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.70816e-05,
+      "loss": 0.8389,
+      "step": 13345
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7088000000000003e-05,
+      "loss": 0.8449,
+      "step": 13350
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7094400000000004e-05,
+      "loss": 0.8227,
+      "step": 13355
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.71008e-05,
+      "loss": 0.8428,
+      "step": 13360
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.71072e-05,
+      "loss": 0.8433,
+      "step": 13365
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.71136e-05,
+      "loss": 0.8234,
+      "step": 13370
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7120000000000002e-05,
+      "loss": 0.8491,
+      "step": 13375
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.71264e-05,
+      "loss": 0.8404,
+      "step": 13380
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.71328e-05,
+      "loss": 0.8367,
+      "step": 13385
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7139200000000004e-05,
+      "loss": 0.8479,
+      "step": 13390
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.71456e-05,
+      "loss": 0.8498,
+      "step": 13395
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7152e-05,
+      "loss": 0.8428,
+      "step": 13400
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.71584e-05,
+      "loss": 0.8454,
+      "step": 13405
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7164800000000002e-05,
+      "loss": 0.8406,
+      "step": 13410
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7171200000000003e-05,
+      "loss": 0.8114,
+      "step": 13415
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.71776e-05,
+      "loss": 0.8524,
+      "step": 13420
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7184e-05,
+      "loss": 0.8417,
+      "step": 13425
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.71904e-05,
+      "loss": 0.841,
+      "step": 13430
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.71968e-05,
+      "loss": 0.8215,
+      "step": 13435
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.72032e-05,
+      "loss": 0.8316,
+      "step": 13440
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7209600000000002e-05,
+      "loss": 0.8345,
+      "step": 13445
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7216000000000003e-05,
+      "loss": 0.8411,
+      "step": 13450
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7222400000000003e-05,
+      "loss": 0.8375,
+      "step": 13455
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.72288e-05,
+      "loss": 0.8322,
+      "step": 13460
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.72352e-05,
+      "loss": 0.8426,
+      "step": 13465
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.72416e-05,
+      "loss": 0.8496,
+      "step": 13470
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7248e-05,
+      "loss": 0.8344,
+      "step": 13475
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7254400000000002e-05,
+      "loss": 0.8341,
+      "step": 13480
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7260800000000003e-05,
+      "loss": 0.84,
+      "step": 13485
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7267200000000003e-05,
+      "loss": 0.8412,
+      "step": 13490
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.72736e-05,
+      "loss": 0.8325,
+      "step": 13495
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.728e-05,
+      "loss": 0.823,
+      "step": 13500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.72864e-05,
+      "loss": 0.8372,
+      "step": 13505
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.72928e-05,
+      "loss": 0.8447,
+      "step": 13510
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7299200000000002e-05,
+      "loss": 0.8273,
+      "step": 13515
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7305600000000002e-05,
+      "loss": 0.8418,
+      "step": 13520
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7312000000000003e-05,
+      "loss": 0.8366,
+      "step": 13525
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7318400000000003e-05,
+      "loss": 0.823,
+      "step": 13530
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.73248e-05,
+      "loss": 0.8209,
+      "step": 13535
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.73312e-05,
+      "loss": 0.8375,
+      "step": 13540
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.73376e-05,
+      "loss": 0.8351,
+      "step": 13545
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7344000000000002e-05,
+      "loss": 0.8305,
+      "step": 13550
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.73504e-05,
+      "loss": 0.8408,
+      "step": 13555
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.73568e-05,
+      "loss": 0.833,
+      "step": 13560
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7363200000000003e-05,
+      "loss": 0.8321,
+      "step": 13565
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.73696e-05,
+      "loss": 0.8381,
+      "step": 13570
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7376e-05,
+      "loss": 0.8467,
+      "step": 13575
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.73824e-05,
+      "loss": 0.8482,
+      "step": 13580
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7388800000000002e-05,
+      "loss": 0.8389,
+      "step": 13585
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7395200000000002e-05,
+      "loss": 0.8333,
+      "step": 13590
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.74016e-05,
+      "loss": 0.8334,
+      "step": 13595
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7408e-05,
+      "loss": 0.8202,
+      "step": 13600
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7414400000000004e-05,
+      "loss": 0.8289,
+      "step": 13605
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.74208e-05,
+      "loss": 0.8391,
+      "step": 13610
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.74272e-05,
+      "loss": 0.8322,
+      "step": 13615
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7433600000000002e-05,
+      "loss": 0.8326,
+      "step": 13620
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7440000000000002e-05,
+      "loss": 0.8305,
+      "step": 13625
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.74464e-05,
+      "loss": 0.8288,
+      "step": 13630
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.74528e-05,
+      "loss": 0.8473,
+      "step": 13635
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.74592e-05,
+      "loss": 0.8329,
+      "step": 13640
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.74656e-05,
+      "loss": 0.8264,
+      "step": 13645
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7472e-05,
+      "loss": 0.8282,
+      "step": 13650
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7478400000000002e-05,
+      "loss": 0.8288,
+      "step": 13655
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7484800000000002e-05,
+      "loss": 0.8347,
+      "step": 13660
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7491200000000003e-05,
+      "loss": 0.8384,
+      "step": 13665
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.74976e-05,
+      "loss": 0.84,
+      "step": 13670
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7504e-05,
+      "loss": 0.8378,
+      "step": 13675
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.75104e-05,
+      "loss": 0.8314,
+      "step": 13680
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.75168e-05,
+      "loss": 0.8138,
+      "step": 13685
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7523200000000002e-05,
+      "loss": 0.844,
+      "step": 13690
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7529600000000002e-05,
+      "loss": 0.8374,
+      "step": 13695
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7536000000000003e-05,
+      "loss": 0.8283,
+      "step": 13700
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.75424e-05,
+      "loss": 0.8344,
+      "step": 13705
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.75488e-05,
+      "loss": 0.827,
+      "step": 13710
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.75552e-05,
+      "loss": 0.8313,
+      "step": 13715
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.75616e-05,
+      "loss": 0.8432,
+      "step": 13720
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7568000000000002e-05,
+      "loss": 0.8406,
+      "step": 13725
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.75744e-05,
+      "loss": 0.8435,
+      "step": 13730
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7580800000000003e-05,
+      "loss": 0.8322,
+      "step": 13735
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7587200000000003e-05,
+      "loss": 0.8353,
+      "step": 13740
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.75936e-05,
+      "loss": 0.8315,
+      "step": 13745
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.76e-05,
+      "loss": 0.8392,
+      "step": 13750
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.76064e-05,
+      "loss": 0.8329,
+      "step": 13755
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7612800000000002e-05,
+      "loss": 0.8362,
+      "step": 13760
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7619200000000002e-05,
+      "loss": 0.809,
+      "step": 13765
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.76256e-05,
+      "loss": 0.8346,
+      "step": 13770
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7632000000000003e-05,
+      "loss": 0.8344,
+      "step": 13775
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7638400000000004e-05,
+      "loss": 0.8157,
+      "step": 13780
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.76448e-05,
+      "loss": 0.8401,
+      "step": 13785
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.76512e-05,
+      "loss": 0.8476,
+      "step": 13790
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7657600000000002e-05,
+      "loss": 0.8206,
+      "step": 13795
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7664000000000002e-05,
+      "loss": 0.8315,
+      "step": 13800
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.76704e-05,
+      "loss": 0.8308,
+      "step": 13805
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.76768e-05,
+      "loss": 0.8367,
+      "step": 13810
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.76832e-05,
+      "loss": 0.8209,
+      "step": 13815
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.76896e-05,
+      "loss": 0.8276,
+      "step": 13820
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7696e-05,
+      "loss": 0.8253,
+      "step": 13825
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7702400000000002e-05,
+      "loss": 0.8285,
+      "step": 13830
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7708800000000002e-05,
+      "loss": 0.8152,
+      "step": 13835
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7715200000000003e-05,
+      "loss": 0.8225,
+      "step": 13840
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.77216e-05,
+      "loss": 0.8245,
+      "step": 13845
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7728e-05,
+      "loss": 0.8366,
+      "step": 13850
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.77344e-05,
+      "loss": 0.8314,
+      "step": 13855
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.77408e-05,
+      "loss": 0.8294,
+      "step": 13860
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7747200000000002e-05,
+      "loss": 0.8377,
+      "step": 13865
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7753600000000002e-05,
+      "loss": 0.8382,
+      "step": 13870
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7760000000000003e-05,
+      "loss": 0.8231,
+      "step": 13875
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.77664e-05,
+      "loss": 0.8425,
+      "step": 13880
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.77728e-05,
+      "loss": 0.8297,
+      "step": 13885
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.77792e-05,
+      "loss": 0.8363,
+      "step": 13890
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.77856e-05,
+      "loss": 0.8457,
+      "step": 13895
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7792000000000002e-05,
+      "loss": 0.8423,
+      "step": 13900
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7798400000000002e-05,
+      "loss": 0.8165,
+      "step": 13905
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7804800000000003e-05,
+      "loss": 0.831,
+      "step": 13910
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7811200000000003e-05,
+      "loss": 0.8165,
+      "step": 13915
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.78176e-05,
+      "loss": 0.8331,
+      "step": 13920
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7824e-05,
+      "loss": 0.8325,
+      "step": 13925
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.78304e-05,
+      "loss": 0.8252,
+      "step": 13930
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7836800000000002e-05,
+      "loss": 0.8264,
+      "step": 13935
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.78432e-05,
+      "loss": 0.8399,
+      "step": 13940
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7849600000000003e-05,
+      "loss": 0.8421,
+      "step": 13945
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7856000000000003e-05,
+      "loss": 0.8385,
+      "step": 13950
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.78624e-05,
+      "loss": 0.8377,
+      "step": 13955
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.78688e-05,
+      "loss": 0.8317,
+      "step": 13960
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.78752e-05,
+      "loss": 0.8353,
+      "step": 13965
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7881600000000002e-05,
+      "loss": 0.8298,
+      "step": 13970
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7888000000000002e-05,
+      "loss": 0.8283,
+      "step": 13975
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.78944e-05,
+      "loss": 0.8425,
+      "step": 13980
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.79008e-05,
+      "loss": 0.8288,
+      "step": 13985
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7907200000000004e-05,
+      "loss": 0.8473,
+      "step": 13990
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.79136e-05,
+      "loss": 0.8394,
+      "step": 13995
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.792e-05,
+      "loss": 0.8262,
+      "step": 14000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7926400000000002e-05,
+      "loss": 0.837,
+      "step": 14005
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7932800000000002e-05,
+      "loss": 0.8411,
+      "step": 14010
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7939200000000003e-05,
+      "loss": 0.8244,
+      "step": 14015
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.79456e-05,
+      "loss": 0.8164,
+      "step": 14020
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7952e-05,
+      "loss": 0.8266,
+      "step": 14025
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7958400000000004e-05,
+      "loss": 0.8267,
+      "step": 14030
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.79648e-05,
+      "loss": 0.8274,
+      "step": 14035
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.79712e-05,
+      "loss": 0.8357,
+      "step": 14040
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7977600000000002e-05,
+      "loss": 0.8345,
+      "step": 14045
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7984000000000003e-05,
+      "loss": 0.8204,
+      "step": 14050
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.79904e-05,
+      "loss": 0.8326,
+      "step": 14055
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.79968e-05,
+      "loss": 0.8247,
+      "step": 14060
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.80032e-05,
+      "loss": 0.824,
+      "step": 14065
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.80096e-05,
+      "loss": 0.8242,
+      "step": 14070
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8016e-05,
+      "loss": 0.845,
+      "step": 14075
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8022400000000002e-05,
+      "loss": 0.8278,
+      "step": 14080
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8028800000000003e-05,
+      "loss": 0.8275,
+      "step": 14085
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8035200000000003e-05,
+      "loss": 0.8284,
+      "step": 14090
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.80416e-05,
+      "loss": 0.8216,
+      "step": 14095
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8048e-05,
+      "loss": 0.8265,
+      "step": 14100
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.80544e-05,
+      "loss": 0.8289,
+      "step": 14105
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.80608e-05,
+      "loss": 0.8176,
+      "step": 14110
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8067200000000002e-05,
+      "loss": 0.8307,
+      "step": 14115
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8073600000000003e-05,
+      "loss": 0.8193,
+      "step": 14120
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8080000000000003e-05,
+      "loss": 0.81,
+      "step": 14125
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.80864e-05,
+      "loss": 0.8343,
+      "step": 14130
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.80928e-05,
+      "loss": 0.8357,
+      "step": 14135
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.80992e-05,
+      "loss": 0.8303,
+      "step": 14140
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.81056e-05,
+      "loss": 0.8301,
+      "step": 14145
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8112000000000002e-05,
+      "loss": 0.8367,
+      "step": 14150
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.81184e-05,
+      "loss": 0.8317,
+      "step": 14155
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8124800000000003e-05,
+      "loss": 0.8385,
+      "step": 14160
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8131200000000004e-05,
+      "loss": 0.826,
+      "step": 14165
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.81376e-05,
+      "loss": 0.8383,
+      "step": 14170
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8144e-05,
+      "loss": 0.832,
+      "step": 14175
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.81504e-05,
+      "loss": 0.8251,
+      "step": 14180
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8156800000000002e-05,
+      "loss": 0.822,
+      "step": 14185
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.81632e-05,
+      "loss": 0.8238,
+      "step": 14190
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.81696e-05,
+      "loss": 0.8259,
+      "step": 14195
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8176000000000004e-05,
+      "loss": 0.8203,
+      "step": 14200
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.81824e-05,
+      "loss": 0.8335,
+      "step": 14205
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.81888e-05,
+      "loss": 0.8195,
+      "step": 14210
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.81952e-05,
+      "loss": 0.8292,
+      "step": 14215
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8201600000000002e-05,
+      "loss": 0.8102,
+      "step": 14220
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8208000000000003e-05,
+      "loss": 0.8244,
+      "step": 14225
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.82144e-05,
+      "loss": 0.8269,
+      "step": 14230
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.82208e-05,
+      "loss": 0.8148,
+      "step": 14235
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.82272e-05,
+      "loss": 0.8124,
+      "step": 14240
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.82336e-05,
+      "loss": 0.8362,
+      "step": 14245
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.824e-05,
+      "loss": 0.8217,
+      "step": 14250
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8246400000000002e-05,
+      "loss": 0.8276,
+      "step": 14255
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8252800000000003e-05,
+      "loss": 0.832,
+      "step": 14260
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8259200000000003e-05,
+      "loss": 0.82,
+      "step": 14265
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.82656e-05,
+      "loss": 0.8157,
+      "step": 14270
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8272e-05,
+      "loss": 0.8223,
+      "step": 14275
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.82784e-05,
+      "loss": 0.8264,
+      "step": 14280
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.82848e-05,
+      "loss": 0.8374,
+      "step": 14285
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8291200000000002e-05,
+      "loss": 0.8347,
+      "step": 14290
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8297600000000002e-05,
+      "loss": 0.834,
+      "step": 14295
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8304000000000003e-05,
+      "loss": 0.8229,
+      "step": 14300
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.83104e-05,
+      "loss": 0.8201,
+      "step": 14305
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.83168e-05,
+      "loss": 0.8185,
+      "step": 14310
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.83232e-05,
+      "loss": 0.8217,
+      "step": 14315
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.83296e-05,
+      "loss": 0.8258,
+      "step": 14320
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8336000000000002e-05,
+      "loss": 0.8282,
+      "step": 14325
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8342400000000002e-05,
+      "loss": 0.8146,
+      "step": 14330
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8348800000000003e-05,
+      "loss": 0.8165,
+      "step": 14335
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8355200000000003e-05,
+      "loss": 0.8282,
+      "step": 14340
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.83616e-05,
+      "loss": 0.8401,
+      "step": 14345
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8368e-05,
+      "loss": 0.8248,
+      "step": 14350
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.83744e-05,
+      "loss": 0.8189,
+      "step": 14355
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8380800000000002e-05,
+      "loss": 0.821,
+      "step": 14360
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.83872e-05,
+      "loss": 0.8285,
+      "step": 14365
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8393600000000003e-05,
+      "loss": 0.833,
+      "step": 14370
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8400000000000003e-05,
+      "loss": 0.8169,
+      "step": 14375
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.84064e-05,
+      "loss": 0.8324,
+      "step": 14380
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.84128e-05,
+      "loss": 0.8258,
+      "step": 14385
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.84192e-05,
+      "loss": 0.8041,
+      "step": 14390
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8425600000000002e-05,
+      "loss": 0.8109,
+      "step": 14395
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8432000000000002e-05,
+      "loss": 0.8242,
+      "step": 14400
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.84384e-05,
+      "loss": 0.8168,
+      "step": 14405
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.84448e-05,
+      "loss": 0.8191,
+      "step": 14410
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8451200000000004e-05,
+      "loss": 0.832,
+      "step": 14415
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.84576e-05,
+      "loss": 0.8185,
+      "step": 14420
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8464e-05,
+      "loss": 0.8364,
+      "step": 14425
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8470400000000002e-05,
+      "loss": 0.8067,
+      "step": 14430
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8476800000000002e-05,
+      "loss": 0.83,
+      "step": 14435
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.84832e-05,
+      "loss": 0.8221,
+      "step": 14440
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.84896e-05,
+      "loss": 0.8158,
+      "step": 14445
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8496e-05,
+      "loss": 0.816,
+      "step": 14450
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.85024e-05,
+      "loss": 0.8217,
+      "step": 14455
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.85088e-05,
+      "loss": 0.8166,
+      "step": 14460
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8515200000000002e-05,
+      "loss": 0.8211,
+      "step": 14465
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8521600000000002e-05,
+      "loss": 0.8156,
+      "step": 14470
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8528000000000003e-05,
+      "loss": 0.8404,
+      "step": 14475
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.85344e-05,
+      "loss": 0.8224,
+      "step": 14480
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.85408e-05,
+      "loss": 0.8208,
+      "step": 14485
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.85472e-05,
+      "loss": 0.8342,
+      "step": 14490
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.85536e-05,
+      "loss": 0.8169,
+      "step": 14495
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8560000000000002e-05,
+      "loss": 0.8162,
+      "step": 14500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8566400000000002e-05,
+      "loss": 0.8226,
+      "step": 14505
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8572800000000003e-05,
+      "loss": 0.8195,
+      "step": 14510
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8579200000000003e-05,
+      "loss": 0.823,
+      "step": 14515
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.85856e-05,
+      "loss": 0.8294,
+      "step": 14520
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8592e-05,
+      "loss": 0.8119,
+      "step": 14525
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.85984e-05,
+      "loss": 0.8194,
+      "step": 14530
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8604800000000002e-05,
+      "loss": 0.8173,
+      "step": 14535
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8611200000000002e-05,
+      "loss": 0.8379,
+      "step": 14540
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8617600000000003e-05,
+      "loss": 0.8357,
+      "step": 14545
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8624000000000003e-05,
+      "loss": 0.8291,
+      "step": 14550
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.86304e-05,
+      "loss": 0.8102,
+      "step": 14555
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.86368e-05,
+      "loss": 0.8192,
+      "step": 14560
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.86432e-05,
+      "loss": 0.8203,
+      "step": 14565
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8649600000000002e-05,
+      "loss": 0.8047,
+      "step": 14570
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8656000000000002e-05,
+      "loss": 0.824,
+      "step": 14575
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.86624e-05,
+      "loss": 0.8291,
+      "step": 14580
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8668800000000003e-05,
+      "loss": 0.8011,
+      "step": 14585
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8675200000000004e-05,
+      "loss": 0.8259,
+      "step": 14590
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.86816e-05,
+      "loss": 0.8334,
+      "step": 14595
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8688e-05,
+      "loss": 0.8105,
+      "step": 14600
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8694400000000002e-05,
+      "loss": 0.8151,
+      "step": 14605
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8700800000000002e-05,
+      "loss": 0.8265,
+      "step": 14610
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.87072e-05,
+      "loss": 0.8152,
+      "step": 14615
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.87136e-05,
+      "loss": 0.8142,
+      "step": 14620
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8720000000000004e-05,
+      "loss": 0.8174,
+      "step": 14625
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.87264e-05,
+      "loss": 0.8227,
+      "step": 14630
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.87328e-05,
+      "loss": 0.8242,
+      "step": 14635
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8739200000000002e-05,
+      "loss": 0.813,
+      "step": 14640
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8745600000000002e-05,
+      "loss": 0.8191,
+      "step": 14645
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8752000000000003e-05,
+      "loss": 0.8163,
+      "step": 14650
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.87584e-05,
+      "loss": 0.8222,
+      "step": 14655
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.87648e-05,
+      "loss": 0.8249,
+      "step": 14660
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.87712e-05,
+      "loss": 0.8192,
+      "step": 14665
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.87776e-05,
+      "loss": 0.8249,
+      "step": 14670
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8784000000000002e-05,
+      "loss": 0.8122,
+      "step": 14675
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8790400000000002e-05,
+      "loss": 0.8147,
+      "step": 14680
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8796800000000003e-05,
+      "loss": 0.8354,
+      "step": 14685
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.88032e-05,
+      "loss": 0.8265,
+      "step": 14690
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.88096e-05,
+      "loss": 0.8261,
+      "step": 14695
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8816e-05,
+      "loss": 0.8129,
+      "step": 14700
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.88224e-05,
+      "loss": 0.8254,
+      "step": 14705
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8828800000000002e-05,
+      "loss": 0.8266,
+      "step": 14710
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8835200000000002e-05,
+      "loss": 0.8103,
+      "step": 14715
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8841600000000003e-05,
+      "loss": 0.8131,
+      "step": 14720
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8848000000000003e-05,
+      "loss": 0.8187,
+      "step": 14725
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.88544e-05,
+      "loss": 0.8266,
+      "step": 14730
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.88608e-05,
+      "loss": 0.8159,
+      "step": 14735
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.88672e-05,
+      "loss": 0.8238,
+      "step": 14740
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8873600000000002e-05,
+      "loss": 0.8262,
+      "step": 14745
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8880000000000002e-05,
+      "loss": 0.8193,
+      "step": 14750
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8886400000000003e-05,
+      "loss": 0.8217,
+      "step": 14755
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8892800000000003e-05,
+      "loss": 0.8167,
+      "step": 14760
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8899200000000004e-05,
+      "loss": 0.8317,
+      "step": 14765
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.89056e-05,
+      "loss": 0.8183,
+      "step": 14770
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8912e-05,
+      "loss": 0.814,
+      "step": 14775
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.89184e-05,
+      "loss": 0.8142,
+      "step": 14780
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8924800000000002e-05,
+      "loss": 0.8178,
+      "step": 14785
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.89312e-05,
+      "loss": 0.8218,
+      "step": 14790
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8937600000000003e-05,
+      "loss": 0.8264,
+      "step": 14795
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8944000000000004e-05,
+      "loss": 0.8231,
+      "step": 14800
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.89504e-05,
+      "loss": 0.8124,
+      "step": 14805
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.89568e-05,
+      "loss": 0.821,
+      "step": 14810
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.89632e-05,
+      "loss": 0.8452,
+      "step": 14815
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8969600000000002e-05,
+      "loss": 0.7973,
+      "step": 14820
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8976000000000003e-05,
+      "loss": 0.8171,
+      "step": 14825
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.89824e-05,
+      "loss": 0.8168,
+      "step": 14830
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.89888e-05,
+      "loss": 0.8293,
+      "step": 14835
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8995200000000004e-05,
+      "loss": 0.8081,
+      "step": 14840
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.90016e-05,
+      "loss": 0.8286,
+      "step": 14845
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9008e-05,
+      "loss": 0.8109,
+      "step": 14850
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9014400000000002e-05,
+      "loss": 0.8159,
+      "step": 14855
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9020800000000003e-05,
+      "loss": 0.8288,
+      "step": 14860
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.90272e-05,
+      "loss": 0.8245,
+      "step": 14865
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.90336e-05,
+      "loss": 0.8048,
+      "step": 14870
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.904e-05,
+      "loss": 0.8285,
+      "step": 14875
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.90464e-05,
+      "loss": 0.8262,
+      "step": 14880
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.90528e-05,
+      "loss": 0.8195,
+      "step": 14885
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9059200000000002e-05,
+      "loss": 0.82,
+      "step": 14890
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9065600000000003e-05,
+      "loss": 0.808,
+      "step": 14895
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9072000000000003e-05,
+      "loss": 0.8213,
+      "step": 14900
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.90784e-05,
+      "loss": 0.8193,
+      "step": 14905
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.90848e-05,
+      "loss": 0.8236,
+      "step": 14910
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.90912e-05,
+      "loss": 0.8113,
+      "step": 14915
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.90976e-05,
+      "loss": 0.8313,
+      "step": 14920
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9104000000000002e-05,
+      "loss": 0.814,
+      "step": 14925
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9110400000000003e-05,
+      "loss": 0.8084,
+      "step": 14930
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9116800000000003e-05,
+      "loss": 0.8003,
+      "step": 14935
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.91232e-05,
+      "loss": 0.816,
+      "step": 14940
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.91296e-05,
+      "loss": 0.8159,
+      "step": 14945
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9136e-05,
+      "loss": 0.8166,
+      "step": 14950
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.91424e-05,
+      "loss": 0.8202,
+      "step": 14955
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9148800000000002e-05,
+      "loss": 0.8206,
+      "step": 14960
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.91552e-05,
+      "loss": 0.8203,
+      "step": 14965
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9161600000000003e-05,
+      "loss": 0.8355,
+      "step": 14970
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9168000000000004e-05,
+      "loss": 0.8197,
+      "step": 14975
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.91744e-05,
+      "loss": 0.8188,
+      "step": 14980
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.91808e-05,
+      "loss": 0.8269,
+      "step": 14985
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.91872e-05,
+      "loss": 0.8194,
+      "step": 14990
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9193600000000002e-05,
+      "loss": 0.823,
+      "step": 14995
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9200000000000003e-05,
+      "loss": 0.8193,
+      "step": 15000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.92064e-05,
+      "loss": 0.8409,
+      "step": 15005
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9212800000000003e-05,
+      "loss": 0.8122,
+      "step": 15010
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9219200000000004e-05,
+      "loss": 0.8158,
+      "step": 15015
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.92256e-05,
+      "loss": 0.8234,
+      "step": 15020
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9232e-05,
+      "loss": 0.8141,
+      "step": 15025
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9238400000000002e-05,
+      "loss": 0.8137,
+      "step": 15030
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9244800000000002e-05,
+      "loss": 0.8033,
+      "step": 15035
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.92512e-05,
+      "loss": 0.8187,
+      "step": 15040
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.92576e-05,
+      "loss": 0.8124,
+      "step": 15045
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9264e-05,
+      "loss": 0.8189,
+      "step": 15050
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.92704e-05,
+      "loss": 0.7961,
+      "step": 15055
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.92768e-05,
+      "loss": 0.8208,
+      "step": 15060
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9283200000000002e-05,
+      "loss": 0.8329,
+      "step": 15065
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9289600000000002e-05,
+      "loss": 0.825,
+      "step": 15070
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9296000000000003e-05,
+      "loss": 0.8111,
+      "step": 15075
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.93024e-05,
+      "loss": 0.8215,
+      "step": 15080
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.93088e-05,
+      "loss": 0.8053,
+      "step": 15085
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.93152e-05,
+      "loss": 0.8189,
+      "step": 15090
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.93216e-05,
+      "loss": 0.8149,
+      "step": 15095
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9328000000000002e-05,
+      "loss": 0.8291,
+      "step": 15100
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9334400000000002e-05,
+      "loss": 0.8115,
+      "step": 15105
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9340800000000003e-05,
+      "loss": 0.8223,
+      "step": 15110
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.93472e-05,
+      "loss": 0.8199,
+      "step": 15115
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.93536e-05,
+      "loss": 0.8097,
+      "step": 15120
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.936e-05,
+      "loss": 0.8141,
+      "step": 15125
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.93664e-05,
+      "loss": 0.822,
+      "step": 15130
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9372800000000002e-05,
+      "loss": 0.813,
+      "step": 15135
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9379200000000002e-05,
+      "loss": 0.8085,
+      "step": 15140
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9385600000000003e-05,
+      "loss": 0.8092,
+      "step": 15145
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9392000000000003e-05,
+      "loss": 0.8128,
+      "step": 15150
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.93984e-05,
+      "loss": 0.8244,
+      "step": 15155
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.94048e-05,
+      "loss": 0.8262,
+      "step": 15160
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.94112e-05,
+      "loss": 0.815,
+      "step": 15165
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9417600000000002e-05,
+      "loss": 0.819,
+      "step": 15170
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9424e-05,
+      "loss": 0.8138,
+      "step": 15175
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9430400000000003e-05,
+      "loss": 0.8208,
+      "step": 15180
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9436800000000003e-05,
+      "loss": 0.8094,
+      "step": 15185
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.94432e-05,
+      "loss": 0.8128,
+      "step": 15190
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.94496e-05,
+      "loss": 0.8044,
+      "step": 15195
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9456e-05,
+      "loss": 0.8114,
+      "step": 15200
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9462400000000002e-05,
+      "loss": 0.7994,
+      "step": 15205
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9468800000000002e-05,
+      "loss": 0.8119,
+      "step": 15210
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.94752e-05,
+      "loss": 0.8175,
+      "step": 15215
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.94816e-05,
+      "loss": 0.8117,
+      "step": 15220
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9488000000000004e-05,
+      "loss": 0.8078,
+      "step": 15225
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.94944e-05,
+      "loss": 0.8044,
+      "step": 15230
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.95008e-05,
+      "loss": 0.8112,
+      "step": 15235
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9507200000000002e-05,
+      "loss": 0.8014,
+      "step": 15240
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9513600000000002e-05,
+      "loss": 0.8117,
+      "step": 15245
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9520000000000003e-05,
+      "loss": 0.8116,
+      "step": 15250
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.95264e-05,
+      "loss": 0.8038,
+      "step": 15255
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.95328e-05,
+      "loss": 0.8122,
+      "step": 15260
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9539200000000004e-05,
+      "loss": 0.821,
+      "step": 15265
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.95456e-05,
+      "loss": 0.8156,
+      "step": 15270
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9552000000000002e-05,
+      "loss": 0.8174,
+      "step": 15275
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9558400000000002e-05,
+      "loss": 0.8201,
+      "step": 15280
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9564800000000003e-05,
+      "loss": 0.8094,
+      "step": 15285
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.95712e-05,
+      "loss": 0.8115,
+      "step": 15290
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.95776e-05,
+      "loss": 0.8136,
+      "step": 15295
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9584e-05,
+      "loss": 0.8165,
+      "step": 15300
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.95904e-05,
+      "loss": 0.8034,
+      "step": 15305
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9596800000000002e-05,
+      "loss": 0.8141,
+      "step": 15310
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9603200000000002e-05,
+      "loss": 0.8227,
+      "step": 15315
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9609600000000003e-05,
+      "loss": 0.8107,
+      "step": 15320
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9616000000000003e-05,
+      "loss": 0.8058,
+      "step": 15325
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.96224e-05,
+      "loss": 0.8095,
+      "step": 15330
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.96288e-05,
+      "loss": 0.8046,
+      "step": 15335
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.96352e-05,
+      "loss": 0.804,
+      "step": 15340
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9641600000000002e-05,
+      "loss": 0.8001,
+      "step": 15345
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9648000000000002e-05,
+      "loss": 0.8179,
+      "step": 15350
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9654400000000003e-05,
+      "loss": 0.8088,
+      "step": 15355
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9660800000000003e-05,
+      "loss": 0.8192,
+      "step": 15360
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.96672e-05,
+      "loss": 0.8096,
+      "step": 15365
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.96736e-05,
+      "loss": 0.8134,
+      "step": 15370
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.968e-05,
+      "loss": 0.816,
+      "step": 15375
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9686400000000002e-05,
+      "loss": 0.8206,
+      "step": 15380
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9692800000000002e-05,
+      "loss": 0.8196,
+      "step": 15385
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.96992e-05,
+      "loss": 0.8118,
+      "step": 15390
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9705600000000003e-05,
+      "loss": 0.8125,
+      "step": 15395
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9712000000000004e-05,
+      "loss": 0.81,
+      "step": 15400
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.97184e-05,
+      "loss": 0.8187,
+      "step": 15405
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.97248e-05,
+      "loss": 0.8041,
+      "step": 15410
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9731200000000002e-05,
+      "loss": 0.8132,
+      "step": 15415
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9737600000000002e-05,
+      "loss": 0.8071,
+      "step": 15420
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9744e-05,
+      "loss": 0.8211,
+      "step": 15425
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.97504e-05,
+      "loss": 0.8091,
+      "step": 15430
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9756800000000004e-05,
+      "loss": 0.802,
+      "step": 15435
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.97632e-05,
+      "loss": 0.7965,
+      "step": 15440
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.97696e-05,
+      "loss": 0.8086,
+      "step": 15445
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9776000000000002e-05,
+      "loss": 0.8325,
+      "step": 15450
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9782400000000002e-05,
+      "loss": 0.8133,
+      "step": 15455
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9788800000000003e-05,
+      "loss": 0.8049,
+      "step": 15460
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.97952e-05,
+      "loss": 0.8062,
+      "step": 15465
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.98016e-05,
+      "loss": 0.8196,
+      "step": 15470
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9808e-05,
+      "loss": 0.8209,
+      "step": 15475
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.98144e-05,
+      "loss": 0.8192,
+      "step": 15480
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9820800000000002e-05,
+      "loss": 0.8127,
+      "step": 15485
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9827200000000002e-05,
+      "loss": 0.8146,
+      "step": 15490
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9833600000000003e-05,
+      "loss": 0.8263,
+      "step": 15495
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9840000000000003e-05,
+      "loss": 0.8107,
+      "step": 15500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.98464e-05,
+      "loss": 0.8219,
+      "step": 15505
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.98528e-05,
+      "loss": 0.8127,
+      "step": 15510
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.98592e-05,
+      "loss": 0.8284,
+      "step": 15515
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9865600000000002e-05,
+      "loss": 0.806,
+      "step": 15520
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9872000000000002e-05,
+      "loss": 0.8063,
+      "step": 15525
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9878400000000003e-05,
+      "loss": 0.8135,
+      "step": 15530
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9884800000000003e-05,
+      "loss": 0.8099,
+      "step": 15535
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.98912e-05,
+      "loss": 0.7989,
+      "step": 15540
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.98976e-05,
+      "loss": 0.8064,
+      "step": 15545
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9904e-05,
+      "loss": 0.7978,
+      "step": 15550
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.99104e-05,
+      "loss": 0.8143,
+      "step": 15555
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9916800000000002e-05,
+      "loss": 0.832,
+      "step": 15560
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9923200000000003e-05,
+      "loss": 0.821,
+      "step": 15565
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9929600000000003e-05,
+      "loss": 0.8074,
+      "step": 15570
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9936000000000004e-05,
+      "loss": 0.8159,
+      "step": 15575
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.99424e-05,
+      "loss": 0.8129,
+      "step": 15580
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.99488e-05,
+      "loss": 0.8072,
+      "step": 15585
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.99552e-05,
+      "loss": 0.8191,
+      "step": 15590
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9961600000000002e-05,
+      "loss": 0.8145,
+      "step": 15595
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9968e-05,
+      "loss": 0.8192,
+      "step": 15600
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9974400000000003e-05,
+      "loss": 0.8088,
+      "step": 15605
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9980800000000004e-05,
+      "loss": 0.8024,
+      "step": 15610
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.99872e-05,
+      "loss": 0.8031,
+      "step": 15615
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.99936e-05,
+      "loss": 0.8081,
+      "step": 15620
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2e-05,
+      "loss": 0.812,
+      "step": 15625
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9999999937614355e-05,
+      "loss": 0.8161,
+      "step": 15630
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9999999750457414e-05,
+      "loss": 0.8073,
+      "step": 15635
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.999999943852918e-05,
+      "loss": 0.8099,
+      "step": 15640
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9999999001829658e-05,
+      "loss": 0.7946,
+      "step": 15645
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9999998440358852e-05,
+      "loss": 0.8048,
+      "step": 15650
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9999997754116774e-05,
+      "loss": 0.7974,
+      "step": 15655
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9999996943103425e-05,
+      "loss": 0.8168,
+      "step": 15660
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9999996007318823e-05,
+      "loss": 0.8126,
+      "step": 15665
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9999994946762974e-05,
+      "loss": 0.8016,
+      "step": 15670
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9999993761435893e-05,
+      "loss": 0.8118,
+      "step": 15675
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9999992451337596e-05,
+      "loss": 0.8062,
+      "step": 15680
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9999991016468096e-05,
+      "loss": 0.8151,
+      "step": 15685
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9999989456827415e-05,
+      "loss": 0.8158,
+      "step": 15690
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9999987772415568e-05,
+      "loss": 0.8143,
+      "step": 15695
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9999985963232583e-05,
+      "loss": 0.7976,
+      "step": 15700
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9999984029278474e-05,
+      "loss": 0.8105,
+      "step": 15705
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.999998197055327e-05,
+      "loss": 0.796,
+      "step": 15710
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9999979787056998e-05,
+      "loss": 0.8051,
+      "step": 15715
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9999977478789682e-05,
+      "loss": 0.804,
+      "step": 15720
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9999975045751353e-05,
+      "loss": 0.8058,
+      "step": 15725
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9999972487942035e-05,
+      "loss": 0.7975,
+      "step": 15730
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9999969805361772e-05,
+      "loss": 0.8125,
+      "step": 15735
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.999996699801059e-05,
+      "loss": 0.8174,
+      "step": 15740
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.999996406588852e-05,
+      "loss": 0.8169,
+      "step": 15745
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9999961008995607e-05,
+      "loss": 0.8072,
+      "step": 15750
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9999957827331883e-05,
+      "loss": 0.8119,
+      "step": 15755
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9999954520897394e-05,
+      "loss": 0.8049,
+      "step": 15760
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9999951089692174e-05,
+      "loss": 0.7939,
+      "step": 15765
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.999994753371627e-05,
+      "loss": 0.8087,
+      "step": 15770
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.999994385296973e-05,
+      "loss": 0.818,
+      "step": 15775
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.999994004745259e-05,
+      "loss": 0.8049,
+      "step": 15780
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9999936117164904e-05,
+      "loss": 0.8091,
+      "step": 15785
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.999993206210672e-05,
+      "loss": 0.7908,
+      "step": 15790
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9999927882278092e-05,
+      "loss": 0.8107,
+      "step": 15795
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9999923577679066e-05,
+      "loss": 0.8138,
+      "step": 15800
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.99999191483097e-05,
+      "loss": 0.8172,
+      "step": 15805
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9999914594170044e-05,
+      "loss": 0.8231,
+      "step": 15810
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.999990991526016e-05,
+      "loss": 0.804,
+      "step": 15815
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9999905111580106e-05,
+      "loss": 0.7954,
+      "step": 15820
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9999900183129944e-05,
+      "loss": 0.805,
+      "step": 15825
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.999989512990973e-05,
+      "loss": 0.8058,
+      "step": 15830
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9999889951919527e-05,
+      "loss": 0.7999,
+      "step": 15835
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9999884649159404e-05,
+      "loss": 0.8072,
+      "step": 15840
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9999879221629427e-05,
+      "loss": 0.7836,
+      "step": 15845
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.999987366932966e-05,
+      "loss": 0.8084,
+      "step": 15850
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9999867992260173e-05,
+      "loss": 0.7962,
+      "step": 15855
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.999986219042104e-05,
+      "loss": 0.8081,
+      "step": 15860
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9999856263812333e-05,
+      "loss": 0.8021,
+      "step": 15865
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9999850212434123e-05,
+      "loss": 0.8116,
+      "step": 15870
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9999844036286483e-05,
+      "loss": 0.7887,
+      "step": 15875
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9999837735369497e-05,
+      "loss": 0.8014,
+      "step": 15880
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.999983130968324e-05,
+      "loss": 0.8128,
+      "step": 15885
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9999824759227794e-05,
+      "loss": 0.7902,
+      "step": 15890
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9999818084003243e-05,
+      "loss": 0.8068,
+      "step": 15895
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.999981128400966e-05,
+      "loss": 0.8007,
+      "step": 15900
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9999804359247143e-05,
+      "loss": 0.8065,
+      "step": 15905
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9999797309715764e-05,
+      "loss": 0.8123,
+      "step": 15910
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9999790135415625e-05,
+      "loss": 0.8061,
+      "step": 15915
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.999978283634681e-05,
+      "loss": 0.8218,
+      "step": 15920
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9999775412509405e-05,
+      "loss": 0.8129,
+      "step": 15925
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.999976786390351e-05,
+      "loss": 0.7972,
+      "step": 15930
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9999760190529214e-05,
+      "loss": 0.8042,
+      "step": 15935
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.999975239238662e-05,
+      "loss": 0.8006,
+      "step": 15940
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.999974446947582e-05,
+      "loss": 0.8065,
+      "step": 15945
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9999736421796908e-05,
+      "loss": 0.8079,
+      "step": 15950
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9999728249349994e-05,
+      "loss": 0.8058,
+      "step": 15955
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9999719952135173e-05,
+      "loss": 0.8074,
+      "step": 15960
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9999711530152553e-05,
+      "loss": 0.8038,
+      "step": 15965
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9999702983402236e-05,
+      "loss": 0.8037,
+      "step": 15970
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.999969431188433e-05,
+      "loss": 0.8069,
+      "step": 15975
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9999685515598943e-05,
+      "loss": 0.7963,
+      "step": 15980
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9999676594546187e-05,
+      "loss": 0.798,
+      "step": 15985
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.999966754872617e-05,
+      "loss": 0.7953,
+      "step": 15990
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9999658378139007e-05,
+      "loss": 0.7928,
+      "step": 15995
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9999649082784807e-05,
+      "loss": 0.812,
+      "step": 16000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9999639662663694e-05,
+      "loss": 0.801,
+      "step": 16005
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9999630117775784e-05,
+      "loss": 0.8012,
+      "step": 16010
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.999962044812119e-05,
+      "loss": 0.8092,
+      "step": 16015
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9999610653700037e-05,
+      "loss": 0.7979,
+      "step": 16020
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.999960073451245e-05,
+      "loss": 0.813,
+      "step": 16025
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9999590690558545e-05,
+      "loss": 0.8182,
+      "step": 16030
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.999958052183846e-05,
+      "loss": 0.804,
+      "step": 16035
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9999570228352305e-05,
+      "loss": 0.7902,
+      "step": 16040
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9999559810100222e-05,
+      "loss": 0.8077,
+      "step": 16045
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9999549267082337e-05,
+      "loss": 0.7958,
+      "step": 16050
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.999953859929878e-05,
+      "loss": 0.8167,
+      "step": 16055
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.999952780674968e-05,
+      "loss": 0.8166,
+      "step": 16060
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9999516889435184e-05,
+      "loss": 0.7958,
+      "step": 16065
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9999505847355417e-05,
+      "loss": 0.7823,
+      "step": 16070
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.999949468051052e-05,
+      "loss": 0.7901,
+      "step": 16075
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9999483388900633e-05,
+      "loss": 0.8043,
+      "step": 16080
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9999471972525897e-05,
+      "loss": 0.7981,
+      "step": 16085
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9999460431386458e-05,
+      "loss": 0.806,
+      "step": 16090
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9999448765482453e-05,
+      "loss": 0.8109,
+      "step": 16095
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.999943697481403e-05,
+      "loss": 0.8082,
+      "step": 16100
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9999425059381335e-05,
+      "loss": 0.8006,
+      "step": 16105
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9999413019184523e-05,
+      "loss": 0.8063,
+      "step": 16110
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.999940085422374e-05,
+      "loss": 0.8109,
+      "step": 16115
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9999388564499135e-05,
+      "loss": 0.7764,
+      "step": 16120
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9999376150010868e-05,
+      "loss": 0.7937,
+      "step": 16125
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9999363610759082e-05,
+      "loss": 0.7925,
+      "step": 16130
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9999350946743948e-05,
+      "loss": 0.7932,
+      "step": 16135
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9999338157965618e-05,
+      "loss": 0.7925,
+      "step": 16140
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9999325244424247e-05,
+      "loss": 0.7932,
+      "step": 16145
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9999312206120002e-05,
+      "loss": 0.8033,
+      "step": 16150
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9999299043053046e-05,
+      "loss": 0.8026,
+      "step": 16155
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9999285755223536e-05,
+      "loss": 0.8036,
+      "step": 16160
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9999272342631644e-05,
+      "loss": 0.805,
+      "step": 16165
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.999925880527754e-05,
+      "loss": 0.8061,
+      "step": 16170
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.999924514316138e-05,
+      "loss": 0.8002,
+      "step": 16175
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.999923135628335e-05,
+      "loss": 0.8003,
+      "step": 16180
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9999217444643616e-05,
+      "loss": 0.8034,
+      "step": 16185
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.999920340824235e-05,
+      "loss": 0.8001,
+      "step": 16190
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9999189247079727e-05,
+      "loss": 0.8051,
+      "step": 16195
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9999174961155928e-05,
+      "loss": 0.81,
+      "step": 16200
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.999916055047112e-05,
+      "loss": 0.8035,
+      "step": 16205
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9999146015025503e-05,
+      "loss": 0.8204,
+      "step": 16210
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9999131354819236e-05,
+      "loss": 0.7889,
+      "step": 16215
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9999116569852518e-05,
+      "loss": 0.808,
+      "step": 16220
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9999101660125525e-05,
+      "loss": 0.8148,
+      "step": 16225
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9999086625638445e-05,
+      "loss": 0.8018,
+      "step": 16230
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.999907146639147e-05,
+      "loss": 0.8026,
+      "step": 16235
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.999905618238478e-05,
+      "loss": 0.8034,
+      "step": 16240
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9999040773618578e-05,
+      "loss": 0.7854,
+      "step": 16245
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9999025240093045e-05,
+      "loss": 0.7926,
+      "step": 16250
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.999900958180838e-05,
+      "loss": 0.8031,
+      "step": 16255
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9998993798764775e-05,
+      "loss": 0.8155,
+      "step": 16260
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9998977890962434e-05,
+      "loss": 0.7911,
+      "step": 16265
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9998961858401545e-05,
+      "loss": 0.8107,
+      "step": 16270
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9998945701082322e-05,
+      "loss": 0.7884,
+      "step": 16275
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9998929419004954e-05,
+      "loss": 0.7931,
+      "step": 16280
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9998913012169646e-05,
+      "loss": 0.8023,
+      "step": 16285
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.999889648057661e-05,
+      "loss": 0.8034,
+      "step": 16290
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9998879824226048e-05,
+      "loss": 0.795,
+      "step": 16295
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9998863043118163e-05,
+      "loss": 0.8051,
+      "step": 16300
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9998846137253173e-05,
+      "loss": 0.8022,
+      "step": 16305
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9998829106631285e-05,
+      "loss": 0.795,
+      "step": 16310
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.999881195125271e-05,
+      "loss": 0.8107,
+      "step": 16315
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9998794671117664e-05,
+      "loss": 0.8018,
+      "step": 16320
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.999877726622636e-05,
+      "loss": 0.8064,
+      "step": 16325
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9998759736579018e-05,
+      "loss": 0.8018,
+      "step": 16330
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.999874208217586e-05,
+      "loss": 0.8011,
+      "step": 16335
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.99987243030171e-05,
+      "loss": 0.8013,
+      "step": 16340
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.999870639910296e-05,
+      "loss": 0.8062,
+      "step": 16345
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9998688370433667e-05,
+      "loss": 0.7988,
+      "step": 16350
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.999867021700944e-05,
+      "loss": 0.7937,
+      "step": 16355
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9998651938830517e-05,
+      "loss": 0.8087,
+      "step": 16360
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9998633535897114e-05,
+      "loss": 0.8028,
+      "step": 16365
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9998615008209467e-05,
+      "loss": 0.7911,
+      "step": 16370
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9998596355767805e-05,
+      "loss": 0.7958,
+      "step": 16375
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9998577578572362e-05,
+      "loss": 0.7933,
+      "step": 16380
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9998558676623373e-05,
+      "loss": 0.8002,
+      "step": 16385
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.999853964992107e-05,
+      "loss": 0.7892,
+      "step": 16390
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9998520498465696e-05,
+      "loss": 0.7978,
+      "step": 16395
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9998501222257486e-05,
+      "loss": 0.7932,
+      "step": 16400
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.999848182129668e-05,
+      "loss": 0.7964,
+      "step": 16405
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9998462295583522e-05,
+      "loss": 0.7962,
+      "step": 16410
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9998442645118253e-05,
+      "loss": 0.7919,
+      "step": 16415
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9998422869901124e-05,
+      "loss": 0.7992,
+      "step": 16420
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9998402969932376e-05,
+      "loss": 0.7978,
+      "step": 16425
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9998382945212264e-05,
+      "loss": 0.8122,
+      "step": 16430
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9998362795741027e-05,
+      "loss": 0.8033,
+      "step": 16435
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9998342521518924e-05,
+      "loss": 0.8104,
+      "step": 16440
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9998322122546208e-05,
+      "loss": 0.8012,
+      "step": 16445
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.999830159882313e-05,
+      "loss": 0.7953,
+      "step": 16450
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.999828095034995e-05,
+      "loss": 0.7918,
+      "step": 16455
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9998260177126924e-05,
+      "loss": 0.7934,
+      "step": 16460
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9998239279154312e-05,
+      "loss": 0.8083,
+      "step": 16465
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.999821825643237e-05,
+      "loss": 0.7969,
+      "step": 16470
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.999819710896137e-05,
+      "loss": 0.7886,
+      "step": 16475
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9998175836741564e-05,
+      "loss": 0.7856,
+      "step": 16480
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9998154439773227e-05,
+      "loss": 0.7983,
+      "step": 16485
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9998132918056618e-05,
+      "loss": 0.7917,
+      "step": 16490
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9998111271592015e-05,
+      "loss": 0.7987,
+      "step": 16495
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.999808950037968e-05,
+      "loss": 0.7993,
+      "step": 16500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9998067604419886e-05,
+      "loss": 0.7989,
+      "step": 16505
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9998045583712912e-05,
+      "loss": 0.7924,
+      "step": 16510
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9998023438259025e-05,
+      "loss": 0.8027,
+      "step": 16515
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9998001168058506e-05,
+      "loss": 0.8022,
+      "step": 16520
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.999797877311163e-05,
+      "loss": 0.8021,
+      "step": 16525
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.999795625341868e-05,
+      "loss": 0.7978,
+      "step": 16530
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9997933608979935e-05,
+      "loss": 0.7946,
+      "step": 16535
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.999791083979568e-05,
+      "loss": 0.791,
+      "step": 16540
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9997887945866193e-05,
+      "loss": 0.7827,
+      "step": 16545
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9997864927191763e-05,
+      "loss": 0.8033,
+      "step": 16550
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9997841783772677e-05,
+      "loss": 0.8018,
+      "step": 16555
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.999781851560923e-05,
+      "loss": 0.7866,
+      "step": 16560
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.99977951227017e-05,
+      "loss": 0.791,
+      "step": 16565
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.999777160505039e-05,
+      "loss": 0.7873,
+      "step": 16570
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.999774796265559e-05,
+      "loss": 0.7936,
+      "step": 16575
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.999772419551759e-05,
+      "loss": 0.7932,
+      "step": 16580
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.999770030363669e-05,
+      "loss": 0.79,
+      "step": 16585
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9997676287013192e-05,
+      "loss": 0.8021,
+      "step": 16590
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9997652145647392e-05,
+      "loss": 0.7878,
+      "step": 16595
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.999762787953959e-05,
+      "loss": 0.777,
+      "step": 16600
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.999760348869009e-05,
+      "loss": 0.799,
+      "step": 16605
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9997578973099198e-05,
+      "loss": 0.7981,
+      "step": 16610
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9997554332767214e-05,
+      "loss": 0.7854,
+      "step": 16615
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9997529567694457e-05,
+      "loss": 0.7926,
+      "step": 16620
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9997504677881224e-05,
+      "loss": 0.7918,
+      "step": 16625
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.999747966332783e-05,
+      "loss": 0.7897,
+      "step": 16630
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.999745452403459e-05,
+      "loss": 0.8022,
+      "step": 16635
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9997429260001814e-05,
+      "loss": 0.7988,
+      "step": 16640
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.999740387122982e-05,
+      "loss": 0.7887,
+      "step": 16645
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9997378357718923e-05,
+      "loss": 0.7911,
+      "step": 16650
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.999735271946944e-05,
+      "loss": 0.7977,
+      "step": 16655
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9997326956481693e-05,
+      "loss": 0.7926,
+      "step": 16660
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9997301068756e-05,
+      "loss": 0.7915,
+      "step": 16665
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9997275056292692e-05,
+      "loss": 0.7975,
+      "step": 16670
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9997248919092087e-05,
+      "loss": 0.7834,
+      "step": 16675
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9997222657154512e-05,
+      "loss": 0.7915,
+      "step": 16680
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9997196270480295e-05,
+      "loss": 0.7949,
+      "step": 16685
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9997169759069766e-05,
+      "loss": 0.8013,
+      "step": 16690
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9997143122923254e-05,
+      "loss": 0.7742,
+      "step": 16695
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9997116362041095e-05,
+      "loss": 0.7995,
+      "step": 16700
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9997089476423617e-05,
+      "loss": 0.7932,
+      "step": 16705
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9997062466071163e-05,
+      "loss": 0.7954,
+      "step": 16710
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.999703533098406e-05,
+      "loss": 0.8034,
+      "step": 16715
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.999700807116266e-05,
+      "loss": 0.8046,
+      "step": 16720
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9996980686607293e-05,
+      "loss": 0.7854,
+      "step": 16725
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.99969531773183e-05,
+      "loss": 0.7747,
+      "step": 16730
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9996925543296033e-05,
+      "loss": 0.7984,
+      "step": 16735
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.999689778454083e-05,
+      "loss": 0.8052,
+      "step": 16740
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9996869901053033e-05,
+      "loss": 0.7937,
+      "step": 16745
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9996841892833e-05,
+      "loss": 0.7886,
+      "step": 16750
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9996813759881077e-05,
+      "loss": 0.7856,
+      "step": 16755
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9996785502197612e-05,
+      "loss": 0.7923,
+      "step": 16760
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9996757119782965e-05,
+      "loss": 0.8013,
+      "step": 16765
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9996728612637478e-05,
+      "loss": 0.7776,
+      "step": 16770
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9996699980761515e-05,
+      "loss": 0.7908,
+      "step": 16775
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9996671224155433e-05,
+      "loss": 0.7996,
+      "step": 16780
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.999664234281959e-05,
+      "loss": 0.792,
+      "step": 16785
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9996613336754347e-05,
+      "loss": 0.7879,
+      "step": 16790
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9996584205960063e-05,
+      "loss": 0.7953,
+      "step": 16795
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9996554950437105e-05,
+      "loss": 0.7925,
+      "step": 16800
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.999652557018583e-05,
+      "loss": 0.7896,
+      "step": 16805
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9996496065206615e-05,
+      "loss": 0.7841,
+      "step": 16810
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.999646643549983e-05,
+      "loss": 0.7986,
+      "step": 16815
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9996436681065832e-05,
+      "loss": 0.7807,
+      "step": 16820
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9996406801905e-05,
+      "loss": 0.7994,
+      "step": 16825
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9996376798017703e-05,
+      "loss": 0.7908,
+      "step": 16830
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9996346669404324e-05,
+      "loss": 0.7914,
+      "step": 16835
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.999631641606523e-05,
+      "loss": 0.7935,
+      "step": 16840
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9996286038000804e-05,
+      "loss": 0.8017,
+      "step": 16845
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.999625553521142e-05,
+      "loss": 0.7787,
+      "step": 16850
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9996224907697463e-05,
+      "loss": 0.7809,
+      "step": 16855
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.999619415545931e-05,
+      "loss": 0.7966,
+      "step": 16860
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.999616327849735e-05,
+      "loss": 0.7997,
+      "step": 16865
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9996132276811967e-05,
+      "loss": 0.7927,
+      "step": 16870
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9996101150403543e-05,
+      "loss": 0.8082,
+      "step": 16875
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9996069899272473e-05,
+      "loss": 0.7971,
+      "step": 16880
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9996038523419148e-05,
+      "loss": 0.8072,
+      "step": 16885
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9996007022843952e-05,
+      "loss": 0.795,
+      "step": 16890
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.999597539754728e-05,
+      "loss": 0.7845,
+      "step": 16895
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9995943647529533e-05,
+      "loss": 0.8029,
+      "step": 16900
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9995911772791102e-05,
+      "loss": 0.7906,
+      "step": 16905
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9995879773332386e-05,
+      "loss": 0.7852,
+      "step": 16910
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.999584764915378e-05,
+      "loss": 0.7807,
+      "step": 16915
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9995815400255692e-05,
+      "loss": 0.8018,
+      "step": 16920
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.999578302663852e-05,
+      "loss": 0.7903,
+      "step": 16925
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9995750528302668e-05,
+      "loss": 0.8071,
+      "step": 16930
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.999571790524854e-05,
+      "loss": 0.7847,
+      "step": 16935
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.999568515747655e-05,
+      "loss": 0.7888,
+      "step": 16940
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9995652284987096e-05,
+      "loss": 0.8046,
+      "step": 16945
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.99956192877806e-05,
+      "loss": 0.7917,
+      "step": 16950
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.999558616585746e-05,
+      "loss": 0.8026,
+      "step": 16955
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9995552919218106e-05,
+      "loss": 0.7967,
+      "step": 16960
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9995519547862937e-05,
+      "loss": 0.797,
+      "step": 16965
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.999548605179238e-05,
+      "loss": 0.7935,
+      "step": 16970
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9995452431006844e-05,
+      "loss": 0.7889,
+      "step": 16975
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.999541868550676e-05,
+      "loss": 0.7921,
+      "step": 16980
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9995384815292537e-05,
+      "loss": 0.8021,
+      "step": 16985
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9995350820364608e-05,
+      "loss": 0.7878,
+      "step": 16990
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9995316700723393e-05,
+      "loss": 0.7911,
+      "step": 16995
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9995282456369313e-05,
+      "loss": 0.7908,
+      "step": 17000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.99952480873028e-05,
+      "loss": 0.7852,
+      "step": 17005
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9995213593524284e-05,
+      "loss": 0.7886,
+      "step": 17010
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9995178975034192e-05,
+      "loss": 0.8128,
+      "step": 17015
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.999514423183296e-05,
+      "loss": 0.7891,
+      "step": 17020
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9995109363921017e-05,
+      "loss": 0.7979,
+      "step": 17025
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.99950743712988e-05,
+      "loss": 0.7899,
+      "step": 17030
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9995039253966747e-05,
+      "loss": 0.7863,
+      "step": 17035
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9995004011925294e-05,
+      "loss": 0.7946,
+      "step": 17040
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9994968645174882e-05,
+      "loss": 0.8047,
+      "step": 17045
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.999493315371595e-05,
+      "loss": 0.7986,
+      "step": 17050
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9994897537548944e-05,
+      "loss": 0.7912,
+      "step": 17055
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9994861796674308e-05,
+      "loss": 0.7776,
+      "step": 17060
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9994825931092486e-05,
+      "loss": 0.7861,
+      "step": 17065
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9994789940803926e-05,
+      "loss": 0.7856,
+      "step": 17070
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.999475382580908e-05,
+      "loss": 0.7942,
+      "step": 17075
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9994717586108395e-05,
+      "loss": 0.783,
+      "step": 17080
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9994681221702324e-05,
+      "loss": 0.7999,
+      "step": 17085
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9994644732591322e-05,
+      "loss": 0.7919,
+      "step": 17090
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.999460811877584e-05,
+      "loss": 0.7913,
+      "step": 17095
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9994571380256343e-05,
+      "loss": 0.7873,
+      "step": 17100
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.999453451703328e-05,
+      "loss": 0.781,
+      "step": 17105
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9994497529107118e-05,
+      "loss": 0.7837,
+      "step": 17110
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9994460416478315e-05,
+      "loss": 0.7863,
+      "step": 17115
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9994423179147335e-05,
+      "loss": 0.7935,
+      "step": 17120
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9994385817114644e-05,
+      "loss": 0.7906,
+      "step": 17125
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9994348330380706e-05,
+      "loss": 0.7967,
+      "step": 17130
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.999431071894599e-05,
+      "loss": 0.7842,
+      "step": 17135
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9994272982810966e-05,
+      "loss": 0.8055,
+      "step": 17140
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9994235121976104e-05,
+      "loss": 0.7899,
+      "step": 17145
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9994197136441874e-05,
+      "loss": 0.8048,
+      "step": 17150
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.999415902620875e-05,
+      "loss": 0.7852,
+      "step": 17155
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9994120791277214e-05,
+      "loss": 0.7933,
+      "step": 17160
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9994082431647735e-05,
+      "loss": 0.7959,
+      "step": 17165
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9994043947320797e-05,
+      "loss": 0.7917,
+      "step": 17170
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.999400533829688e-05,
+      "loss": 0.7864,
+      "step": 17175
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9993966604576463e-05,
+      "loss": 0.7963,
+      "step": 17180
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.999392774616003e-05,
+      "loss": 0.7757,
+      "step": 17185
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9993888763048068e-05,
+      "loss": 0.7946,
+      "step": 17190
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9993849655241058e-05,
+      "loss": 0.7849,
+      "step": 17195
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9993810422739496e-05,
+      "loss": 0.788,
+      "step": 17200
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9993771065543862e-05,
+      "loss": 0.8009,
+      "step": 17205
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9993731583654655e-05,
+      "loss": 0.8047,
+      "step": 17210
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9993691977072366e-05,
+      "loss": 0.7931,
+      "step": 17215
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.999365224579749e-05,
+      "loss": 0.7869,
+      "step": 17220
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9993612389830516e-05,
+      "loss": 0.8,
+      "step": 17225
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9993572409171945e-05,
+      "loss": 0.7887,
+      "step": 17230
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.999353230382228e-05,
+      "loss": 0.8025,
+      "step": 17235
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.999349207378202e-05,
+      "loss": 0.7926,
+      "step": 17240
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9993451719051663e-05,
+      "loss": 0.7856,
+      "step": 17245
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9993411239631713e-05,
+      "loss": 0.7885,
+      "step": 17250
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9993370635522676e-05,
+      "loss": 0.7891,
+      "step": 17255
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9993329906725066e-05,
+      "loss": 0.7792,
+      "step": 17260
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.999328905323938e-05,
+      "loss": 0.7941,
+      "step": 17265
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9993248075066135e-05,
+      "loss": 0.7877,
+      "step": 17270
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9993206972205836e-05,
+      "loss": 0.7872,
+      "step": 17275
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9993165744659e-05,
+      "loss": 0.7769,
+      "step": 17280
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9993124392426142e-05,
+      "loss": 0.7763,
+      "step": 17285
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9993082915507776e-05,
+      "loss": 0.7893,
+      "step": 17290
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9993041313904422e-05,
+      "loss": 0.792,
+      "step": 17295
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.99929995876166e-05,
+      "loss": 0.7964,
+      "step": 17300
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.999295773664482e-05,
+      "loss": 0.8055,
+      "step": 17305
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9992915760989617e-05,
+      "loss": 0.7838,
+      "step": 17310
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9992873660651512e-05,
+      "loss": 0.7795,
+      "step": 17315
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9992831435631024e-05,
+      "loss": 0.7923,
+      "step": 17320
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9992789085928686e-05,
+      "loss": 0.7888,
+      "step": 17325
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9992746611545027e-05,
+      "loss": 0.7942,
+      "step": 17330
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.999270401248057e-05,
+      "loss": 0.7993,
+      "step": 17335
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.999266128873585e-05,
+      "loss": 0.7932,
+      "step": 17340
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9992618440311407e-05,
+      "loss": 0.7697,
+      "step": 17345
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9992575467207765e-05,
+      "loss": 0.785,
+      "step": 17350
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9992532369425468e-05,
+      "loss": 0.7809,
+      "step": 17355
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.999248914696505e-05,
+      "loss": 0.7794,
+      "step": 17360
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.999244579982705e-05,
+      "loss": 0.7813,
+      "step": 17365
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.999240232801201e-05,
+      "loss": 0.7972,
+      "step": 17370
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.999235873152047e-05,
+      "loss": 0.7865,
+      "step": 17375
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9992315010352978e-05,
+      "loss": 0.7781,
+      "step": 17380
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9992271164510077e-05,
+      "loss": 0.7833,
+      "step": 17385
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9992227193992316e-05,
+      "loss": 0.7826,
+      "step": 17390
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.999218309880024e-05,
+      "loss": 0.7971,
+      "step": 17395
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9992138878934405e-05,
+      "loss": 0.8041,
+      "step": 17400
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9992094534395358e-05,
+      "loss": 0.7839,
+      "step": 17405
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.999205006518365e-05,
+      "loss": 0.7784,
+      "step": 17410
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9992005471299843e-05,
+      "loss": 0.7738,
+      "step": 17415
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9991960752744487e-05,
+      "loss": 0.7821,
+      "step": 17420
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9991915909518146e-05,
+      "loss": 0.8009,
+      "step": 17425
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9991870941621373e-05,
+      "loss": 0.7956,
+      "step": 17430
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9991825849054736e-05,
+      "loss": 0.7858,
+      "step": 17435
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.999178063181879e-05,
+      "loss": 0.7937,
+      "step": 17440
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.999173528991411e-05,
+      "loss": 0.7911,
+      "step": 17445
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9991689823341246e-05,
+      "loss": 0.7851,
+      "step": 17450
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9991644232100777e-05,
+      "loss": 0.7952,
+      "step": 17455
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.999159851619327e-05,
+      "loss": 0.7841,
+      "step": 17460
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9991552675619295e-05,
+      "loss": 0.7903,
+      "step": 17465
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9991506710379424e-05,
+      "loss": 0.7887,
+      "step": 17470
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9991460620474226e-05,
+      "loss": 0.789,
+      "step": 17475
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9991414405904285e-05,
+      "loss": 0.7943,
+      "step": 17480
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.999136806667017e-05,
+      "loss": 0.7921,
+      "step": 17485
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.999132160277246e-05,
+      "loss": 0.7916,
+      "step": 17490
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.999127501421174e-05,
+      "loss": 0.7846,
+      "step": 17495
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9991228300988586e-05,
+      "loss": 0.7849,
+      "step": 17500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9991181463103582e-05,
+      "loss": 0.7762,
+      "step": 17505
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9991134500557314e-05,
+      "loss": 0.7766,
+      "step": 17510
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9991087413350367e-05,
+      "loss": 0.7789,
+      "step": 17515
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9991040201483326e-05,
+      "loss": 0.7817,
+      "step": 17520
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9990992864956785e-05,
+      "loss": 0.7826,
+      "step": 17525
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9990945403771334e-05,
+      "loss": 0.7854,
+      "step": 17530
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.999089781792756e-05,
+      "loss": 0.7736,
+      "step": 17535
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.999085010742606e-05,
+      "loss": 0.7784,
+      "step": 17540
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9990802272267432e-05,
+      "loss": 0.7852,
+      "step": 17545
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9990754312452266e-05,
+      "loss": 0.793,
+      "step": 17550
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.999070622798117e-05,
+      "loss": 0.793,
+      "step": 17555
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9990658018854737e-05,
+      "loss": 0.7967,
+      "step": 17560
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.999060968507357e-05,
+      "loss": 0.7935,
+      "step": 17565
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.999056122663827e-05,
+      "loss": 0.7738,
+      "step": 17570
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9990512643549444e-05,
+      "loss": 0.7943,
+      "step": 17575
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.99904639358077e-05,
+      "loss": 0.7765,
+      "step": 17580
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.999041510341364e-05,
+      "loss": 0.7782,
+      "step": 17585
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.999036614636788e-05,
+      "loss": 0.7819,
+      "step": 17590
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.999031706467103e-05,
+      "loss": 0.7797,
+      "step": 17595
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9990267858323697e-05,
+      "loss": 0.7859,
+      "step": 17600
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.99902185273265e-05,
+      "loss": 0.7792,
+      "step": 17605
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.999016907168005e-05,
+      "loss": 0.7817,
+      "step": 17610
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.999011949138497e-05,
+      "loss": 0.7855,
+      "step": 17615
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9990069786441874e-05,
+      "loss": 0.7691,
+      "step": 17620
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9990019956851384e-05,
+      "loss": 0.7852,
+      "step": 17625
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9989970002614118e-05,
+      "loss": 0.7847,
+      "step": 17630
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.998991992373071e-05,
+      "loss": 0.7722,
+      "step": 17635
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.998986972020177e-05,
+      "loss": 0.7866,
+      "step": 17640
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.998981939202793e-05,
+      "loss": 0.7912,
+      "step": 17645
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9989768939209826e-05,
+      "loss": 0.7878,
+      "step": 17650
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9989718361748082e-05,
+      "loss": 0.7918,
+      "step": 17655
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9989667659643322e-05,
+      "loss": 0.7751,
+      "step": 17660
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.998961683289619e-05,
+      "loss": 0.7902,
+      "step": 17665
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9989565881507317e-05,
+      "loss": 0.7958,
+      "step": 17670
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9989514805477332e-05,
+      "loss": 0.7786,
+      "step": 17675
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.998946360480688e-05,
+      "loss": 0.7785,
+      "step": 17680
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.998941227949659e-05,
+      "loss": 0.7763,
+      "step": 17685
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9989360829547117e-05,
+      "loss": 0.7854,
+      "step": 17690
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9989309254959096e-05,
+      "loss": 0.7791,
+      "step": 17695
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9989257555733164e-05,
+      "loss": 0.7901,
+      "step": 17700
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9989205731869977e-05,
+      "loss": 0.7867,
+      "step": 17705
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.998915378337017e-05,
+      "loss": 0.7873,
+      "step": 17710
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9989101710234403e-05,
+      "loss": 0.8055,
+      "step": 17715
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.998904951246332e-05,
+      "loss": 0.791,
+      "step": 17720
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9988997190057574e-05,
+      "loss": 0.7741,
+      "step": 17725
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9988944743017814e-05,
+      "loss": 0.7945,
+      "step": 17730
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9988892171344698e-05,
+      "loss": 0.7847,
+      "step": 17735
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.998883947503888e-05,
+      "loss": 0.7916,
+      "step": 17740
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9988786654101016e-05,
+      "loss": 0.7865,
+      "step": 17745
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9988733708531772e-05,
+      "loss": 0.7945,
+      "step": 17750
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.99886806383318e-05,
+      "loss": 0.7809,
+      "step": 17755
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.998862744350177e-05,
+      "loss": 0.7709,
+      "step": 17760
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.998857412404234e-05,
+      "loss": 0.7882,
+      "step": 17765
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9988520679954175e-05,
+      "loss": 0.7893,
+      "step": 17770
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9988467111237947e-05,
+      "loss": 0.7949,
+      "step": 17775
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9988413417894322e-05,
+      "loss": 0.7723,
+      "step": 17780
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9988359599923964e-05,
+      "loss": 0.7767,
+      "step": 17785
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9988305657327552e-05,
+      "loss": 0.7874,
+      "step": 17790
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.998825159010576e-05,
+      "loss": 0.7961,
+      "step": 17795
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9988197398259257e-05,
+      "loss": 0.7845,
+      "step": 17800
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9988143081788722e-05,
+      "loss": 0.787,
+      "step": 17805
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.998808864069483e-05,
+      "loss": 0.7909,
+      "step": 17810
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9988034074978268e-05,
+      "loss": 0.7986,
+      "step": 17815
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.998797938463971e-05,
+      "loss": 0.7742,
+      "step": 17820
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9987924569679836e-05,
+      "loss": 0.7799,
+      "step": 17825
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9987869630099333e-05,
+      "loss": 0.7805,
+      "step": 17830
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9987814565898894e-05,
+      "loss": 0.7886,
+      "step": 17835
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9987759377079193e-05,
+      "loss": 0.7778,
+      "step": 17840
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9987704063640925e-05,
+      "loss": 0.7831,
+      "step": 17845
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9987648625584785e-05,
+      "loss": 0.7849,
+      "step": 17850
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9987593062911457e-05,
+      "loss": 0.7729,
+      "step": 17855
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9987537375621635e-05,
+      "loss": 0.7932,
+      "step": 17860
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9987481563716015e-05,
+      "loss": 0.789,
+      "step": 17865
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.99874256271953e-05,
+      "loss": 0.7778,
+      "step": 17870
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.998736956606018e-05,
+      "loss": 0.7953,
+      "step": 17875
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9987313380311352e-05,
+      "loss": 0.7918,
+      "step": 17880
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9987257069949522e-05,
+      "loss": 0.7737,
+      "step": 17885
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9987200634975397e-05,
+      "loss": 0.7828,
+      "step": 17890
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9987144075389674e-05,
+      "loss": 0.7836,
+      "step": 17895
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.998708739119306e-05,
+      "loss": 0.7735,
+      "step": 17900
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9987030582386264e-05,
+      "loss": 0.7686,
+      "step": 17905
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9986973648969996e-05,
+      "loss": 0.7814,
+      "step": 17910
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9986916590944964e-05,
+      "loss": 0.7792,
+      "step": 17915
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9986859408311878e-05,
+      "loss": 0.783,
+      "step": 17920
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9986802101071453e-05,
+      "loss": 0.7787,
+      "step": 17925
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.998674466922441e-05,
+      "loss": 0.7809,
+      "step": 17930
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9986687112771457e-05,
+      "loss": 0.7825,
+      "step": 17935
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9986629431713315e-05,
+      "loss": 0.7737,
+      "step": 17940
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9986571626050705e-05,
+      "loss": 0.7896,
+      "step": 17945
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.998651369578435e-05,
+      "loss": 0.7813,
+      "step": 17950
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9986455640914967e-05,
+      "loss": 0.7751,
+      "step": 17955
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9986397461443286e-05,
+      "loss": 0.7832,
+      "step": 17960
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9986339157370026e-05,
+      "loss": 0.7842,
+      "step": 17965
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9986280728695923e-05,
+      "loss": 0.7829,
+      "step": 17970
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9986222175421704e-05,
+      "loss": 0.7832,
+      "step": 17975
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9986163497548096e-05,
+      "loss": 0.7913,
+      "step": 17980
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9986104695075832e-05,
+      "loss": 0.7809,
+      "step": 17985
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9986045768005645e-05,
+      "loss": 0.7859,
+      "step": 17990
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9985986716338274e-05,
+      "loss": 0.7821,
+      "step": 17995
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9985927540074453e-05,
+      "loss": 0.7691,
+      "step": 18000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9985868239214922e-05,
+      "loss": 0.7877,
+      "step": 18005
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.998580881376042e-05,
+      "loss": 0.781,
+      "step": 18010
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9985749263711685e-05,
+      "loss": 0.7697,
+      "step": 18015
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9985689589069467e-05,
+      "loss": 0.7816,
+      "step": 18020
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9985629789834503e-05,
+      "loss": 0.7736,
+      "step": 18025
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9985569866007545e-05,
+      "loss": 0.781,
+      "step": 18030
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9985509817589338e-05,
+      "loss": 0.7733,
+      "step": 18035
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9985449644580635e-05,
+      "loss": 0.7856,
+      "step": 18040
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.998538934698218e-05,
+      "loss": 0.781,
+      "step": 18045
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9985328924794732e-05,
+      "loss": 0.7725,
+      "step": 18050
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.998526837801904e-05,
+      "loss": 0.7625,
+      "step": 18055
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.998520770665586e-05,
+      "loss": 0.7889,
+      "step": 18060
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9985146910705954e-05,
+      "loss": 0.783,
+      "step": 18065
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9985085990170072e-05,
+      "loss": 0.7721,
+      "step": 18070
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9985024945048982e-05,
+      "loss": 0.7708,
+      "step": 18075
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.998496377534344e-05,
+      "loss": 0.7721,
+      "step": 18080
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9984902481054217e-05,
+      "loss": 0.7922,
+      "step": 18085
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.998484106218207e-05,
+      "loss": 0.7676,
+      "step": 18090
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.998477951872777e-05,
+      "loss": 0.7756,
+      "step": 18095
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.998471785069208e-05,
+      "loss": 0.7706,
+      "step": 18100
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9984656058075772e-05,
+      "loss": 0.7932,
+      "step": 18105
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.998459414087962e-05,
+      "loss": 0.783,
+      "step": 18110
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.998453209910439e-05,
+      "loss": 0.766,
+      "step": 18115
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9984469932750862e-05,
+      "loss": 0.7776,
+      "step": 18120
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9984407641819812e-05,
+      "loss": 0.7832,
+      "step": 18125
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9984345226312012e-05,
+      "loss": 0.7855,
+      "step": 18130
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9984282686228244e-05,
+      "loss": 0.7837,
+      "step": 18135
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.998422002156929e-05,
+      "loss": 0.7945,
+      "step": 18140
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9984157232335926e-05,
+      "loss": 0.7823,
+      "step": 18145
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9984094318528943e-05,
+      "loss": 0.7739,
+      "step": 18150
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9984031280149122e-05,
+      "loss": 0.784,
+      "step": 18155
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9983968117197247e-05,
+      "loss": 0.7687,
+      "step": 18160
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9983904829674113e-05,
+      "loss": 0.7749,
+      "step": 18165
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.99838414175805e-05,
+      "loss": 0.7871,
+      "step": 18170
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.998377788091721e-05,
+      "loss": 0.785,
+      "step": 18175
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9983714219685027e-05,
+      "loss": 0.7903,
+      "step": 18180
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.998365043388475e-05,
+      "loss": 0.7881,
+      "step": 18185
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9983586523517175e-05,
+      "loss": 0.7811,
+      "step": 18190
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9983522488583095e-05,
+      "loss": 0.7674,
+      "step": 18195
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9983458329083313e-05,
+      "loss": 0.7784,
+      "step": 18200
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9983394045018628e-05,
+      "loss": 0.7572,
+      "step": 18205
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9983329636389844e-05,
+      "loss": 0.772,
+      "step": 18210
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9983265103197764e-05,
+      "loss": 0.7824,
+      "step": 18215
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.998320044544319e-05,
+      "loss": 0.7781,
+      "step": 18220
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9983135663126937e-05,
+      "loss": 0.7825,
+      "step": 18225
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9983070756249802e-05,
+      "loss": 0.7669,
+      "step": 18230
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.99830057248126e-05,
+      "loss": 0.7682,
+      "step": 18235
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9982940568816147e-05,
+      "loss": 0.7862,
+      "step": 18240
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9982875288261246e-05,
+      "loss": 0.7574,
+      "step": 18245
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.998280988314872e-05,
+      "loss": 0.7726,
+      "step": 18250
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9982744353479383e-05,
+      "loss": 0.7774,
+      "step": 18255
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9982678699254054e-05,
+      "loss": 0.7779,
+      "step": 18260
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9982612920473547e-05,
+      "loss": 0.7687,
+      "step": 18265
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9982547017138688e-05,
+      "loss": 0.7738,
+      "step": 18270
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9982480989250293e-05,
+      "loss": 0.7755,
+      "step": 18275
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.998241483680919e-05,
+      "loss": 0.7638,
+      "step": 18280
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.998234855981621e-05,
+      "loss": 0.7847,
+      "step": 18285
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9982282158272173e-05,
+      "loss": 0.7699,
+      "step": 18290
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9982215632177905e-05,
+      "loss": 0.7734,
+      "step": 18295
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.998214898153424e-05,
+      "loss": 0.7749,
+      "step": 18300
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9982082206342014e-05,
+      "loss": 0.771,
+      "step": 18305
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.998201530660205e-05,
+      "loss": 0.784,
+      "step": 18310
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9981948282315194e-05,
+      "loss": 0.7818,
+      "step": 18315
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9981881133482273e-05,
+      "loss": 0.7738,
+      "step": 18320
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.998181386010413e-05,
+      "loss": 0.773,
+      "step": 18325
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.99817464621816e-05,
+      "loss": 0.7871,
+      "step": 18330
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9981678939715528e-05,
+      "loss": 0.7758,
+      "step": 18335
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.998161129270676e-05,
+      "loss": 0.7792,
+      "step": 18340
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.998154352115613e-05,
+      "loss": 0.7832,
+      "step": 18345
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.998147562506449e-05,
+      "loss": 0.7741,
+      "step": 18350
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9981407604432686e-05,
+      "loss": 0.7716,
+      "step": 18355
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9981339459261567e-05,
+      "loss": 0.7788,
+      "step": 18360
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9981271189551983e-05,
+      "loss": 0.776,
+      "step": 18365
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9981202795304787e-05,
+      "loss": 0.7827,
+      "step": 18370
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9981134276520828e-05,
+      "loss": 0.7732,
+      "step": 18375
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9981065633200966e-05,
+      "loss": 0.7686,
+      "step": 18380
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9980996865346057e-05,
+      "loss": 0.7664,
+      "step": 18385
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9980927972956952e-05,
+      "loss": 0.7757,
+      "step": 18390
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9980858956034523e-05,
+      "loss": 0.782,
+      "step": 18395
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9980789814579622e-05,
+      "loss": 0.7716,
+      "step": 18400
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.998072054859311e-05,
+      "loss": 0.7846,
+      "step": 18405
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.998065115807586e-05,
+      "loss": 0.7947,
+      "step": 18410
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9980581643028732e-05,
+      "loss": 0.7732,
+      "step": 18415
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9980512003452593e-05,
+      "loss": 0.7669,
+      "step": 18420
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9980442239348313e-05,
+      "loss": 0.7637,
+      "step": 18425
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9980372350716766e-05,
+      "loss": 0.7702,
+      "step": 18430
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.998030233755882e-05,
+      "loss": 0.7664,
+      "step": 18435
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9980232199875343e-05,
+      "loss": 0.779,
+      "step": 18440
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9980161937667223e-05,
+      "loss": 0.7699,
+      "step": 18445
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.998009155093533e-05,
+      "loss": 0.7792,
+      "step": 18450
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.998002103968054e-05,
+      "loss": 0.7805,
+      "step": 18455
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9979950403903732e-05,
+      "loss": 0.7716,
+      "step": 18460
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.99798796436058e-05,
+      "loss": 0.783,
+      "step": 18465
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9979808758787608e-05,
+      "loss": 0.7766,
+      "step": 18470
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.997973774945005e-05,
+      "loss": 0.7756,
+      "step": 18475
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9979666615594017e-05,
+      "loss": 0.7746,
+      "step": 18480
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9979595357220388e-05,
+      "loss": 0.7724,
+      "step": 18485
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9979523974330056e-05,
+      "loss": 0.7666,
+      "step": 18490
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9979452466923913e-05,
+      "loss": 0.782,
+      "step": 18495
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9979380835002846e-05,
+      "loss": 0.7774,
+      "step": 18500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9979309078567756e-05,
+      "loss": 0.7661,
+      "step": 18505
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.997923719761953e-05,
+      "loss": 0.7799,
+      "step": 18510
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9979165192159068e-05,
+      "loss": 0.7618,
+      "step": 18515
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9979093062187273e-05,
+      "loss": 0.7753,
+      "step": 18520
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9979020807705043e-05,
+      "loss": 0.7724,
+      "step": 18525
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9978948428713276e-05,
+      "loss": 0.7846,
+      "step": 18530
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9978875925212876e-05,
+      "loss": 0.7761,
+      "step": 18535
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.997880329720475e-05,
+      "loss": 0.7746,
+      "step": 18540
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9978730544689803e-05,
+      "loss": 0.7699,
+      "step": 18545
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9978657667668945e-05,
+      "loss": 0.7656,
+      "step": 18550
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.997858466614308e-05,
+      "loss": 0.7729,
+      "step": 18555
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9978511540113122e-05,
+      "loss": 0.783,
+      "step": 18560
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9978438289579984e-05,
+      "loss": 0.7778,
+      "step": 18565
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.997836491454458e-05,
+      "loss": 0.7675,
+      "step": 18570
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.997829141500782e-05,
+      "loss": 0.7762,
+      "step": 18575
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9978217790970633e-05,
+      "loss": 0.7723,
+      "step": 18580
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9978144042433926e-05,
+      "loss": 0.7802,
+      "step": 18585
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9978070169398626e-05,
+      "loss": 0.7696,
+      "step": 18590
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.997799617186565e-05,
+      "loss": 0.7659,
+      "step": 18595
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9977922049835926e-05,
+      "loss": 0.7762,
+      "step": 18600
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9977847803310376e-05,
+      "loss": 0.777,
+      "step": 18605
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9977773432289926e-05,
+      "loss": 0.7876,
+      "step": 18610
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9977698936775503e-05,
+      "loss": 0.7681,
+      "step": 18615
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.997762431676804e-05,
+      "loss": 0.7774,
+      "step": 18620
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.997754957226847e-05,
+      "loss": 0.7636,
+      "step": 18625
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9977474703277716e-05,
+      "loss": 0.7647,
+      "step": 18630
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.997739970979672e-05,
+      "loss": 0.7675,
+      "step": 18635
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9977324591826415e-05,
+      "loss": 0.764,
+      "step": 18640
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.997724934936774e-05,
+      "loss": 0.7801,
+      "step": 18645
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.997717398242163e-05,
+      "loss": 0.7612,
+      "step": 18650
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9977098490989035e-05,
+      "loss": 0.7747,
+      "step": 18655
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9977022875070885e-05,
+      "loss": 0.7594,
+      "step": 18660
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.997694713466813e-05,
+      "loss": 0.7751,
+      "step": 18665
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9976871269781714e-05,
+      "loss": 0.7744,
+      "step": 18670
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.997679528041258e-05,
+      "loss": 0.7688,
+      "step": 18675
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.997671916656168e-05,
+      "loss": 0.7768,
+      "step": 18680
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9976642928229965e-05,
+      "loss": 0.7793,
+      "step": 18685
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9976566565418383e-05,
+      "loss": 0.7845,
+      "step": 18690
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.997649007812789e-05,
+      "loss": 0.7827,
+      "step": 18695
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9976413466359437e-05,
+      "loss": 0.7606,
+      "step": 18700
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.997633673011398e-05,
+      "loss": 0.7717,
+      "step": 18705
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.997625986939248e-05,
+      "loss": 0.7697,
+      "step": 18710
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9976182884195892e-05,
+      "loss": 0.7663,
+      "step": 18715
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.997610577452518e-05,
+      "loss": 0.7781,
+      "step": 18720
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9976028540381304e-05,
+      "loss": 0.7785,
+      "step": 18725
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9975951181765226e-05,
+      "loss": 0.7779,
+      "step": 18730
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9975873698677914e-05,
+      "loss": 0.7792,
+      "step": 18735
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9975796091120333e-05,
+      "loss": 0.7655,
+      "step": 18740
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9975718359093453e-05,
+      "loss": 0.7638,
+      "step": 18745
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9975640502598243e-05,
+      "loss": 0.7833,
+      "step": 18750
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9975562521635678e-05,
+      "loss": 0.7668,
+      "step": 18755
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9975484416206723e-05,
+      "loss": 0.7619,
+      "step": 18760
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.997540618631236e-05,
+      "loss": 0.7721,
+      "step": 18765
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.997532783195356e-05,
+      "loss": 0.762,
+      "step": 18770
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9975249353131304e-05,
+      "loss": 0.7724,
+      "step": 18775
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.997517074984657e-05,
+      "loss": 0.7809,
+      "step": 18780
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.997509202210034e-05,
+      "loss": 0.7667,
+      "step": 18785
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9975013169893593e-05,
+      "loss": 0.7661,
+      "step": 18790
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9974934193227315e-05,
+      "loss": 0.778,
+      "step": 18795
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.997485509210249e-05,
+      "loss": 0.7694,
+      "step": 18800
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.997477586652011e-05,
+      "loss": 0.7692,
+      "step": 18805
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9974696516481158e-05,
+      "loss": 0.7676,
+      "step": 18810
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9974617041986627e-05,
+      "loss": 0.7791,
+      "step": 18815
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9974537443037504e-05,
+      "loss": 0.7748,
+      "step": 18820
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.997445771963479e-05,
+      "loss": 0.7713,
+      "step": 18825
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9974377871779472e-05,
+      "loss": 0.7651,
+      "step": 18830
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.997429789947255e-05,
+      "loss": 0.7716,
+      "step": 18835
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9974217802715024e-05,
+      "loss": 0.7727,
+      "step": 18840
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.997413758150789e-05,
+      "loss": 0.7516,
+      "step": 18845
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9974057235852148e-05,
+      "loss": 0.7719,
+      "step": 18850
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9973976765748803e-05,
+      "loss": 0.7647,
+      "step": 18855
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9973896171198858e-05,
+      "loss": 0.764,
+      "step": 18860
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9973815452203314e-05,
+      "loss": 0.767,
+      "step": 18865
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.997373460876319e-05,
+      "loss": 0.7818,
+      "step": 18870
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9973653640879486e-05,
+      "loss": 0.7703,
+      "step": 18875
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9973572548553215e-05,
+      "loss": 0.7616,
+      "step": 18880
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9973491331785384e-05,
+      "loss": 0.7724,
+      "step": 18885
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9973409990577014e-05,
+      "loss": 0.7678,
+      "step": 18890
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9973328524929113e-05,
+      "loss": 0.7691,
+      "step": 18895
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.99732469348427e-05,
+      "loss": 0.7775,
+      "step": 18900
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9973165220318797e-05,
+      "loss": 0.7597,
+      "step": 18905
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.997308338135842e-05,
+      "loss": 0.7806,
+      "step": 18910
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9973001417962586e-05,
+      "loss": 0.7739,
+      "step": 18915
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9972919330132326e-05,
+      "loss": 0.7714,
+      "step": 18920
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9972837117868657e-05,
+      "loss": 0.7745,
+      "step": 18925
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.997275478117261e-05,
+      "loss": 0.7742,
+      "step": 18930
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.997267232004521e-05,
+      "loss": 0.7666,
+      "step": 18935
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9972589734487485e-05,
+      "loss": 0.794,
+      "step": 18940
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9972507024500463e-05,
+      "loss": 0.7598,
+      "step": 18945
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9972424190085186e-05,
+      "loss": 0.7586,
+      "step": 18950
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9972341231242675e-05,
+      "loss": 0.7678,
+      "step": 18955
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9972258147973976e-05,
+      "loss": 0.7576,
+      "step": 18960
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9972174940280118e-05,
+      "loss": 0.7733,
+      "step": 18965
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.997209160816214e-05,
+      "loss": 0.7643,
+      "step": 18970
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9972008151621086e-05,
+      "loss": 0.7718,
+      "step": 18975
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9971924570657996e-05,
+      "loss": 0.7775,
+      "step": 18980
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.997184086527391e-05,
+      "loss": 0.7647,
+      "step": 18985
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9971757035469874e-05,
+      "loss": 0.7547,
+      "step": 18990
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9971673081246936e-05,
+      "loss": 0.7677,
+      "step": 18995
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.997158900260614e-05,
+      "loss": 0.7569,
+      "step": 19000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.997150479954854e-05,
+      "loss": 0.7683,
+      "step": 19005
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.997142047207518e-05,
+      "loss": 0.7662,
+      "step": 19010
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9971336020187114e-05,
+      "loss": 0.7727,
+      "step": 19015
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.99712514438854e-05,
+      "loss": 0.7742,
+      "step": 19020
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.997116674317109e-05,
+      "loss": 0.7698,
+      "step": 19025
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.997108191804524e-05,
+      "loss": 0.7806,
+      "step": 19030
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9970996968508908e-05,
+      "loss": 0.7785,
+      "step": 19035
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.997091189456316e-05,
+      "loss": 0.7665,
+      "step": 19040
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.997082669620905e-05,
+      "loss": 0.7701,
+      "step": 19045
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9970741373447645e-05,
+      "loss": 0.7717,
+      "step": 19050
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.997065592628001e-05,
+      "loss": 0.7688,
+      "step": 19055
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.997057035470721e-05,
+      "loss": 0.7412,
+      "step": 19060
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.997048465873031e-05,
+      "loss": 0.7745,
+      "step": 19065
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.997039883835038e-05,
+      "loss": 0.7613,
+      "step": 19070
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9970312893568497e-05,
+      "loss": 0.7666,
+      "step": 19075
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9970226824385728e-05,
+      "loss": 0.7644,
+      "step": 19080
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9970140630803145e-05,
+      "loss": 0.762,
+      "step": 19085
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.997005431282183e-05,
+      "loss": 0.7598,
+      "step": 19090
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.996996787044285e-05,
+      "loss": 0.7699,
+      "step": 19095
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9969881303667296e-05,
+      "loss": 0.7787,
+      "step": 19100
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.996979461249624e-05,
+      "loss": 0.778,
+      "step": 19105
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9969707796930768e-05,
+      "loss": 0.7649,
+      "step": 19110
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9969620856971955e-05,
+      "loss": 0.7654,
+      "step": 19115
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9969533792620895e-05,
+      "loss": 0.7714,
+      "step": 19120
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9969446603878673e-05,
+      "loss": 0.7586,
+      "step": 19125
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.996935929074637e-05,
+      "loss": 0.7638,
+      "step": 19130
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9969271853225083e-05,
+      "loss": 0.76,
+      "step": 19135
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.99691842913159e-05,
+      "loss": 0.7905,
+      "step": 19140
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.996909660501991e-05,
+      "loss": 0.7799,
+      "step": 19145
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9969008794338214e-05,
+      "loss": 0.7528,
+      "step": 19150
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9968920859271905e-05,
+      "loss": 0.7562,
+      "step": 19155
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9968832799822077e-05,
+      "loss": 0.7776,
+      "step": 19160
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.996874461598983e-05,
+      "loss": 0.7703,
+      "step": 19165
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.996865630777627e-05,
+      "loss": 0.7593,
+      "step": 19170
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9968567875182492e-05,
+      "loss": 0.7627,
+      "step": 19175
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9968479318209603e-05,
+      "loss": 0.7513,
+      "step": 19180
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9968390636858706e-05,
+      "loss": 0.7595,
+      "step": 19185
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9968301831130906e-05,
+      "loss": 0.7761,
+      "step": 19190
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9968212901027316e-05,
+      "loss": 0.7635,
+      "step": 19195
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.996812384654904e-05,
+      "loss": 0.7626,
+      "step": 19200
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9968034667697193e-05,
+      "loss": 0.7652,
+      "step": 19205
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9967945364472885e-05,
+      "loss": 0.7585,
+      "step": 19210
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9967855936877237e-05,
+      "loss": 0.7631,
+      "step": 19215
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9967766384911357e-05,
+      "loss": 0.7661,
+      "step": 19220
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9967676708576362e-05,
+      "loss": 0.7701,
+      "step": 19225
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9967586907873376e-05,
+      "loss": 0.7597,
+      "step": 19230
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9967496982803517e-05,
+      "loss": 0.7587,
+      "step": 19235
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9967406933367906e-05,
+      "loss": 0.7568,
+      "step": 19240
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9967316759567673e-05,
+      "loss": 0.781,
+      "step": 19245
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9967226461403934e-05,
+      "loss": 0.759,
+      "step": 19250
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.996713603887782e-05,
+      "loss": 0.776,
+      "step": 19255
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.996704549199046e-05,
+      "loss": 0.7502,
+      "step": 19260
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.996695482074298e-05,
+      "loss": 0.7634,
+      "step": 19265
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9966864025136518e-05,
+      "loss": 0.7659,
+      "step": 19270
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.99667731051722e-05,
+      "loss": 0.7609,
+      "step": 19275
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9966682060851163e-05,
+      "loss": 0.7731,
+      "step": 19280
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9966590892174544e-05,
+      "loss": 0.7647,
+      "step": 19285
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.996649959914348e-05,
+      "loss": 0.7616,
+      "step": 19290
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.996640818175911e-05,
+      "loss": 0.76,
+      "step": 19295
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9966316640022577e-05,
+      "loss": 0.7579,
+      "step": 19300
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9966224973935014e-05,
+      "loss": 0.7702,
+      "step": 19305
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9966133183497578e-05,
+      "loss": 0.7618,
+      "step": 19310
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9966041268711404e-05,
+      "loss": 0.7619,
+      "step": 19315
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.996594922957764e-05,
+      "loss": 0.7632,
+      "step": 19320
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9965857066097444e-05,
+      "loss": 0.761,
+      "step": 19325
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9965764778271954e-05,
+      "loss": 0.751,
+      "step": 19330
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9965672366102325e-05,
+      "loss": 0.7748,
+      "step": 19335
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9965579829589713e-05,
+      "loss": 0.7642,
+      "step": 19340
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9965487168735273e-05,
+      "loss": 0.7597,
+      "step": 19345
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9965394383540158e-05,
+      "loss": 0.7553,
+      "step": 19350
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.996530147400553e-05,
+      "loss": 0.7664,
+      "step": 19355
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9965208440132538e-05,
+      "loss": 0.764,
+      "step": 19360
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9965115281922354e-05,
+      "loss": 0.7648,
+      "step": 19365
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9965021999376133e-05,
+      "loss": 0.7651,
+      "step": 19370
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9964928592495046e-05,
+      "loss": 0.764,
+      "step": 19375
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9964835061280253e-05,
+      "loss": 0.7552,
+      "step": 19380
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.996474140573292e-05,
+      "loss": 0.758,
+      "step": 19385
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.996464762585422e-05,
+      "loss": 0.764,
+      "step": 19390
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9964553721645322e-05,
+      "loss": 0.7608,
+      "step": 19395
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9964459693107396e-05,
+      "loss": 0.7669,
+      "step": 19400
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9964365540241614e-05,
+      "loss": 0.7527,
+      "step": 19405
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9964271263049157e-05,
+      "loss": 0.7619,
+      "step": 19410
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9964176861531196e-05,
+      "loss": 0.7653,
+      "step": 19415
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.996408233568891e-05,
+      "loss": 0.7707,
+      "step": 19420
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9963987685523475e-05,
+      "loss": 0.7721,
+      "step": 19425
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9963892911036077e-05,
+      "loss": 0.7563,
+      "step": 19430
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9963798012227898e-05,
+      "loss": 0.7595,
+      "step": 19435
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.996370298910012e-05,
+      "loss": 0.7696,
+      "step": 19440
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9963607841653928e-05,
+      "loss": 0.7586,
+      "step": 19445
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9963512569890512e-05,
+      "loss": 0.7654,
+      "step": 19450
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.996341717381106e-05,
+      "loss": 0.7665,
+      "step": 19455
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9963321653416764e-05,
+      "loss": 0.7559,
+      "step": 19460
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.996322600870881e-05,
+      "loss": 0.758,
+      "step": 19465
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9963130239688397e-05,
+      "loss": 0.7677,
+      "step": 19470
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9963034346356714e-05,
+      "loss": 0.7574,
+      "step": 19475
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9962938328714966e-05,
+      "loss": 0.7735,
+      "step": 19480
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.996284218676434e-05,
+      "loss": 0.7566,
+      "step": 19485
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.996274592050605e-05,
+      "loss": 0.7588,
+      "step": 19490
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9962649529941283e-05,
+      "loss": 0.7699,
+      "step": 19495
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.996255301507125e-05,
+      "loss": 0.7653,
+      "step": 19500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.996245637589715e-05,
+      "loss": 0.7713,
+      "step": 19505
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9962359612420195e-05,
+      "loss": 0.7727,
+      "step": 19510
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9962262724641585e-05,
+      "loss": 0.7612,
+      "step": 19515
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9962165712562538e-05,
+      "loss": 0.7773,
+      "step": 19520
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9962068576184258e-05,
+      "loss": 0.7574,
+      "step": 19525
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9961971315507955e-05,
+      "loss": 0.7599,
+      "step": 19530
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.996187393053485e-05,
+      "loss": 0.7653,
+      "step": 19535
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.996177642126615e-05,
+      "loss": 0.7613,
+      "step": 19540
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.996167878770308e-05,
+      "loss": 0.7711,
+      "step": 19545
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.996158102984685e-05,
+      "loss": 0.741,
+      "step": 19550
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9961483147698682e-05,
+      "loss": 0.7531,
+      "step": 19555
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9961385141259804e-05,
+      "loss": 0.7742,
+      "step": 19560
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.996128701053143e-05,
+      "loss": 0.7611,
+      "step": 19565
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.996118875551479e-05,
+      "loss": 0.7453,
+      "step": 19570
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9961090376211107e-05,
+      "loss": 0.7636,
+      "step": 19575
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9960991872621605e-05,
+      "loss": 0.7659,
+      "step": 19580
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9960893244747525e-05,
+      "loss": 0.7558,
+      "step": 19585
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9960794492590087e-05,
+      "loss": 0.7626,
+      "step": 19590
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9960695616150526e-05,
+      "loss": 0.761,
+      "step": 19595
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9960596615430076e-05,
+      "loss": 0.7618,
+      "step": 19600
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9960497490429967e-05,
+      "loss": 0.7638,
+      "step": 19605
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.996039824115145e-05,
+      "loss": 0.7757,
+      "step": 19610
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.996029886759575e-05,
+      "loss": 0.7667,
+      "step": 19615
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9960199369764108e-05,
+      "loss": 0.7751,
+      "step": 19620
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9960099747657774e-05,
+      "loss": 0.7624,
+      "step": 19625
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9960000001277985e-05,
+      "loss": 0.767,
+      "step": 19630
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.995990013062598e-05,
+      "loss": 0.7533,
+      "step": 19635
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9959800135703018e-05,
+      "loss": 0.7736,
+      "step": 19640
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9959700016510337e-05,
+      "loss": 0.7558,
+      "step": 19645
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9959599773049188e-05,
+      "loss": 0.781,
+      "step": 19650
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9959499405320825e-05,
+      "loss": 0.7573,
+      "step": 19655
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.99593989133265e-05,
+      "loss": 0.7641,
+      "step": 19660
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9959298297067463e-05,
+      "loss": 0.7696,
+      "step": 19665
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9959197556544972e-05,
+      "loss": 0.7607,
+      "step": 19670
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9959096691760284e-05,
+      "loss": 0.7722,
+      "step": 19675
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9958995702714653e-05,
+      "loss": 0.7746,
+      "step": 19680
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9958894589409343e-05,
+      "loss": 0.771,
+      "step": 19685
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.995879335184562e-05,
+      "loss": 0.7464,
+      "step": 19690
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.995869199002474e-05,
+      "loss": 0.7708,
+      "step": 19695
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9958590503947973e-05,
+      "loss": 0.7592,
+      "step": 19700
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9958488893616577e-05,
+      "loss": 0.7622,
+      "step": 19705
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.995838715903183e-05,
+      "loss": 0.7576,
+      "step": 19710
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9958285300194996e-05,
+      "loss": 0.7676,
+      "step": 19715
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9958183317107345e-05,
+      "loss": 0.7792,
+      "step": 19720
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9958081209770155e-05,
+      "loss": 0.7621,
+      "step": 19725
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.995797897818469e-05,
+      "loss": 0.7542,
+      "step": 19730
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9957876622352235e-05,
+      "loss": 0.7638,
+      "step": 19735
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9957774142274067e-05,
+      "loss": 0.7735,
+      "step": 19740
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.995767153795146e-05,
+      "loss": 0.7545,
+      "step": 19745
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9957568809385693e-05,
+      "loss": 0.7682,
+      "step": 19750
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.995746595657805e-05,
+      "loss": 0.7553,
+      "step": 19755
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.995736297952982e-05,
+      "loss": 0.7529,
+      "step": 19760
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9957259878242277e-05,
+      "loss": 0.7565,
+      "step": 19765
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9957156652716717e-05,
+      "loss": 0.7673,
+      "step": 19770
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9957053302954422e-05,
+      "loss": 0.7601,
+      "step": 19775
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.995694982895668e-05,
+      "loss": 0.7601,
+      "step": 19780
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9956846230724793e-05,
+      "loss": 0.7525,
+      "step": 19785
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.995674250826004e-05,
+      "loss": 0.7469,
+      "step": 19790
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9956638661563725e-05,
+      "loss": 0.7623,
+      "step": 19795
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9956534690637137e-05,
+      "loss": 0.7706,
+      "step": 19800
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.995643059548158e-05,
+      "loss": 0.7583,
+      "step": 19805
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9956326376098345e-05,
+      "loss": 0.7609,
+      "step": 19810
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.995622203248874e-05,
+      "loss": 0.7644,
+      "step": 19815
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9956117564654064e-05,
+      "loss": 0.7574,
+      "step": 19820
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9956012972595617e-05,
+      "loss": 0.7598,
+      "step": 19825
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9955908256314706e-05,
+      "loss": 0.7519,
+      "step": 19830
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.995580341581264e-05,
+      "loss": 0.7533,
+      "step": 19835
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9955698451090727e-05,
+      "loss": 0.7571,
+      "step": 19840
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.995559336215027e-05,
+      "loss": 0.7532,
+      "step": 19845
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9955488148992593e-05,
+      "loss": 0.7702,
+      "step": 19850
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9955382811618994e-05,
+      "loss": 0.7661,
+      "step": 19855
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.99552773500308e-05,
+      "loss": 0.7627,
+      "step": 19860
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9955171764229316e-05,
+      "loss": 0.7629,
+      "step": 19865
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9955066054215867e-05,
+      "loss": 0.7553,
+      "step": 19870
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.995496021999177e-05,
+      "loss": 0.755,
+      "step": 19875
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9954854261558348e-05,
+      "loss": 0.7572,
+      "step": 19880
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9954748178916915e-05,
+      "loss": 0.7503,
+      "step": 19885
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9954641972068806e-05,
+      "loss": 0.764,
+      "step": 19890
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9954535641015333e-05,
+      "loss": 0.7498,
+      "step": 19895
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9954429185757835e-05,
+      "loss": 0.7619,
+      "step": 19900
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9954322606297632e-05,
+      "loss": 0.773,
+      "step": 19905
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9954215902636056e-05,
+      "loss": 0.766,
+      "step": 19910
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9954109074774443e-05,
+      "loss": 0.7509,
+      "step": 19915
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9954002122714116e-05,
+      "loss": 0.7628,
+      "step": 19920
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.995389504645642e-05,
+      "loss": 0.7527,
+      "step": 19925
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9953787846002687e-05,
+      "loss": 0.7692,
+      "step": 19930
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.995368052135425e-05,
+      "loss": 0.7639,
+      "step": 19935
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.995357307251246e-05,
+      "loss": 0.7639,
+      "step": 19940
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.995346549947864e-05,
+      "loss": 0.7587,
+      "step": 19945
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9953357802254147e-05,
+      "loss": 0.7507,
+      "step": 19950
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9953249980840322e-05,
+      "loss": 0.7694,
+      "step": 19955
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9953142035238504e-05,
+      "loss": 0.7696,
+      "step": 19960
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9953033965450044e-05,
+      "loss": 0.7663,
+      "step": 19965
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.995292577147629e-05,
+      "loss": 0.7558,
+      "step": 19970
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9952817453318592e-05,
+      "loss": 0.7504,
+      "step": 19975
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9952709010978307e-05,
+      "loss": 0.7558,
+      "step": 19980
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.995260044445678e-05,
+      "loss": 0.7484,
+      "step": 19985
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9952491753755366e-05,
+      "loss": 0.7685,
+      "step": 19990
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9952382938875425e-05,
+      "loss": 0.7633,
+      "step": 19995
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9952273999818312e-05,
+      "loss": 0.7555,
+      "step": 20000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.995216493658539e-05,
+      "loss": 0.7522,
+      "step": 20005
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9952055749178018e-05,
+      "loss": 0.7594,
+      "step": 20010
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9951946437597553e-05,
+      "loss": 0.7537,
+      "step": 20015
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9951837001845365e-05,
+      "loss": 0.7442,
+      "step": 20020
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9951727441922823e-05,
+      "loss": 0.754,
+      "step": 20025
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9951617757831285e-05,
+      "loss": 0.7598,
+      "step": 20030
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.995150794957212e-05,
+      "loss": 0.751,
+      "step": 20035
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9951398017146706e-05,
+      "loss": 0.7587,
+      "step": 20040
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.995128796055641e-05,
+      "loss": 0.7514,
+      "step": 20045
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9951177779802604e-05,
+      "loss": 0.7731,
+      "step": 20050
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9951067474886666e-05,
+      "loss": 0.7627,
+      "step": 20055
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.995095704580997e-05,
+      "loss": 0.758,
+      "step": 20060
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9950846492573892e-05,
+      "loss": 0.7678,
+      "step": 20065
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9950735815179817e-05,
+      "loss": 0.7498,
+      "step": 20070
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.995062501362912e-05,
+      "loss": 0.7654,
+      "step": 20075
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9950514087923185e-05,
+      "loss": 0.7546,
+      "step": 20080
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.99504030380634e-05,
+      "loss": 0.7527,
+      "step": 20085
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9950291864051147e-05,
+      "loss": 0.7616,
+      "step": 20090
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9950180565887814e-05,
+      "loss": 0.7514,
+      "step": 20095
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9950069143574787e-05,
+      "loss": 0.7529,
+      "step": 20100
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9949957597113463e-05,
+      "loss": 0.7537,
+      "step": 20105
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9949845926505227e-05,
+      "loss": 0.7503,
+      "step": 20110
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9949734131751475e-05,
+      "loss": 0.7693,
+      "step": 20115
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9949622212853603e-05,
+      "loss": 0.7645,
+      "step": 20120
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9949510169813006e-05,
+      "loss": 0.7689,
+      "step": 20125
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.994939800263108e-05,
+      "loss": 0.7709,
+      "step": 20130
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9949285711309224e-05,
+      "loss": 0.7488,
+      "step": 20135
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.994917329584885e-05,
+      "loss": 0.7458,
+      "step": 20140
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9949060756251347e-05,
+      "loss": 0.7625,
+      "step": 20145
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.994894809251812e-05,
+      "loss": 0.7495,
+      "step": 20150
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.994883530465059e-05,
+      "loss": 0.7622,
+      "step": 20155
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9948722392650147e-05,
+      "loss": 0.7713,
+      "step": 20160
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9948609356518206e-05,
+      "loss": 0.7567,
+      "step": 20165
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9948496196256178e-05,
+      "loss": 0.7527,
+      "step": 20170
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.994838291186548e-05,
+      "loss": 0.7649,
+      "step": 20175
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9948269503347515e-05,
+      "loss": 0.7619,
+      "step": 20180
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9948155970703705e-05,
+      "loss": 0.7586,
+      "step": 20185
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9948042313935463e-05,
+      "loss": 0.7546,
+      "step": 20190
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.994792853304421e-05,
+      "loss": 0.7613,
+      "step": 20195
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9947814628031363e-05,
+      "loss": 0.7662,
+      "step": 20200
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9947700598898344e-05,
+      "loss": 0.7497,
+      "step": 20205
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.994758644564658e-05,
+      "loss": 0.736,
+      "step": 20210
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.994747216827749e-05,
+      "loss": 0.7649,
+      "step": 20215
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.99473577667925e-05,
+      "loss": 0.7403,
+      "step": 20220
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.994724324119304e-05,
+      "loss": 0.7585,
+      "step": 20225
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9947128591480536e-05,
+      "loss": 0.7609,
+      "step": 20230
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9947013817656425e-05,
+      "loss": 0.7493,
+      "step": 20235
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.994689891972213e-05,
+      "loss": 0.7581,
+      "step": 20240
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.994678389767909e-05,
+      "loss": 0.7328,
+      "step": 20245
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9946668751528745e-05,
+      "loss": 0.754,
+      "step": 20250
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.994655348127252e-05,
+      "loss": 0.7686,
+      "step": 20255
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9946438086911855e-05,
+      "loss": 0.757,
+      "step": 20260
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.99463225684482e-05,
+      "loss": 0.7637,
+      "step": 20265
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9946206925882992e-05,
+      "loss": 0.751,
+      "step": 20270
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9946091159217668e-05,
+      "loss": 0.7393,
+      "step": 20275
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.994597526845368e-05,
+      "loss": 0.7449,
+      "step": 20280
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.994585925359247e-05,
+      "loss": 0.759,
+      "step": 20285
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.994574311463548e-05,
+      "loss": 0.7524,
+      "step": 20290
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.994562685158417e-05,
+      "loss": 0.7746,
+      "step": 20295
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9945510464439984e-05,
+      "loss": 0.7499,
+      "step": 20300
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9945393953204378e-05,
+      "loss": 0.753,
+      "step": 20305
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9945277317878808e-05,
+      "loss": 0.7546,
+      "step": 20310
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9945160558464717e-05,
+      "loss": 0.7579,
+      "step": 20315
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9945043674963575e-05,
+      "loss": 0.7449,
+      "step": 20320
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9944926667376833e-05,
+      "loss": 0.7658,
+      "step": 20325
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9944809535705954e-05,
+      "loss": 0.7615,
+      "step": 20330
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9944692279952396e-05,
+      "loss": 0.7602,
+      "step": 20335
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9944574900117625e-05,
+      "loss": 0.7592,
+      "step": 20340
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9944457396203108e-05,
+      "loss": 0.7551,
+      "step": 20345
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.994433976821031e-05,
+      "loss": 0.7421,
+      "step": 20350
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9944222016140692e-05,
+      "loss": 0.757,
+      "step": 20355
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9944104139995727e-05,
+      "loss": 0.7556,
+      "step": 20360
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.994398613977689e-05,
+      "loss": 0.7651,
+      "step": 20365
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.994386801548565e-05,
+      "loss": 0.7561,
+      "step": 20370
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.994374976712348e-05,
+      "loss": 0.755,
+      "step": 20375
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.994363139469186e-05,
+      "loss": 0.7451,
+      "step": 20380
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9943512898192263e-05,
+      "loss": 0.7576,
+      "step": 20385
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.994339427762616e-05,
+      "loss": 0.7399,
+      "step": 20390
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.994327553299505e-05,
+      "loss": 0.7573,
+      "step": 20395
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9943156664300394e-05,
+      "loss": 0.7508,
+      "step": 20400
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.994303767154369e-05,
+      "loss": 0.7579,
+      "step": 20405
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9942918554726415e-05,
+      "loss": 0.7491,
+      "step": 20410
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.994279931385006e-05,
+      "loss": 0.7433,
+      "step": 20415
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9942679948916107e-05,
+      "loss": 0.7433,
+      "step": 20420
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.994256045992605e-05,
+      "loss": 0.7557,
+      "step": 20425
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9942440846881383e-05,
+      "loss": 0.7545,
+      "step": 20430
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9942321109783586e-05,
+      "loss": 0.7474,
+      "step": 20435
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9942201248634165e-05,
+      "loss": 0.7594,
+      "step": 20440
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.994208126343461e-05,
+      "loss": 0.753,
+      "step": 20445
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9941961154186424e-05,
+      "loss": 0.7359,
+      "step": 20450
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.99418409208911e-05,
+      "loss": 0.7487,
+      "step": 20455
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9941720563550136e-05,
+      "loss": 0.7452,
+      "step": 20460
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9941600082165037e-05,
+      "loss": 0.7582,
+      "step": 20465
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.994147947673731e-05,
+      "loss": 0.7478,
+      "step": 20470
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9941358747268455e-05,
+      "loss": 0.7505,
+      "step": 20475
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9941237893759976e-05,
+      "loss": 0.7588,
+      "step": 20480
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.994111691621339e-05,
+      "loss": 0.7559,
+      "step": 20485
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9940995814630194e-05,
+      "loss": 0.765,
+      "step": 20490
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9940874589011912e-05,
+      "loss": 0.7553,
+      "step": 20495
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9940753239360047e-05,
+      "loss": 0.754,
+      "step": 20500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.994063176567612e-05,
+      "loss": 0.7531,
+      "step": 20505
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.994051016796164e-05,
+      "loss": 0.754,
+      "step": 20510
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9940388446218126e-05,
+      "loss": 0.7445,
+      "step": 20515
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.99402666004471e-05,
+      "loss": 0.7673,
+      "step": 20520
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9940144630650083e-05,
+      "loss": 0.767,
+      "step": 20525
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.994002253682859e-05,
+      "loss": 0.7436,
+      "step": 20530
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.993990031898415e-05,
+      "loss": 0.7479,
+      "step": 20535
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9939777977118287e-05,
+      "loss": 0.7516,
+      "step": 20540
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9939655511232527e-05,
+      "loss": 0.7559,
+      "step": 20545
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9939532921328398e-05,
+      "loss": 0.7493,
+      "step": 20550
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9939410207407427e-05,
+      "loss": 0.7585,
+      "step": 20555
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.993928736947115e-05,
+      "loss": 0.7563,
+      "step": 20560
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.99391644075211e-05,
+      "loss": 0.7461,
+      "step": 20565
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.99390413215588e-05,
+      "loss": 0.7544,
+      "step": 20570
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9938918111585805e-05,
+      "loss": 0.7483,
+      "step": 20575
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9938794777603635e-05,
+      "loss": 0.7547,
+      "step": 20580
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.993867131961384e-05,
+      "loss": 0.7632,
+      "step": 20585
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.993854773761795e-05,
+      "loss": 0.7554,
+      "step": 20590
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9938424031617515e-05,
+      "loss": 0.7448,
+      "step": 20595
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9938300201614077e-05,
+      "loss": 0.7525,
+      "step": 20600
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9938176247609183e-05,
+      "loss": 0.7465,
+      "step": 20605
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9938052169604375e-05,
+      "loss": 0.7537,
+      "step": 20610
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9937927967601204e-05,
+      "loss": 0.7499,
+      "step": 20615
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9937803641601215e-05,
+      "loss": 0.7677,
+      "step": 20620
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9937679191605964e-05,
+      "loss": 0.7475,
+      "step": 20625
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9937554617617005e-05,
+      "loss": 0.762,
+      "step": 20630
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9937429919635888e-05,
+      "loss": 0.7583,
+      "step": 20635
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9937305097664176e-05,
+      "loss": 0.7431,
+      "step": 20640
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9937180151703416e-05,
+      "loss": 0.7521,
+      "step": 20645
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9937055081755172e-05,
+      "loss": 0.742,
+      "step": 20650
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9936929887821007e-05,
+      "loss": 0.746,
+      "step": 20655
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9936804569902482e-05,
+      "loss": 0.7737,
+      "step": 20660
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.993667912800116e-05,
+      "loss": 0.7451,
+      "step": 20665
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9936553562118602e-05,
+      "loss": 0.7507,
+      "step": 20670
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.993642787225638e-05,
+      "loss": 0.7498,
+      "step": 20675
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.993630205841606e-05,
+      "loss": 0.7428,
+      "step": 20680
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9936176120599214e-05,
+      "loss": 0.7458,
+      "step": 20685
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9936050058807414e-05,
+      "loss": 0.7563,
+      "step": 20690
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9935923873042225e-05,
+      "loss": 0.7516,
+      "step": 20695
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9935797563305233e-05,
+      "loss": 0.7459,
+      "step": 20700
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9935671129598003e-05,
+      "loss": 0.752,
+      "step": 20705
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9935544571922124e-05,
+      "loss": 0.7541,
+      "step": 20710
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9935417890279162e-05,
+      "loss": 0.7419,
+      "step": 20715
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.993529108467071e-05,
+      "loss": 0.7506,
+      "step": 20720
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.993516415509834e-05,
+      "loss": 0.7525,
+      "step": 20725
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.993503710156364e-05,
+      "loss": 0.7559,
+      "step": 20730
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9934909924068198e-05,
+      "loss": 0.7533,
+      "step": 20735
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9934782622613596e-05,
+      "loss": 0.7562,
+      "step": 20740
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9934655197201428e-05,
+      "loss": 0.7544,
+      "step": 20745
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9934527647833276e-05,
+      "loss": 0.7553,
+      "step": 20750
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.993439997451074e-05,
+      "loss": 0.7656,
+      "step": 20755
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9934272177235407e-05,
+      "loss": 0.7323,
+      "step": 20760
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9934144256008878e-05,
+      "loss": 0.7614,
+      "step": 20765
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9934016210832738e-05,
+      "loss": 0.7423,
+      "step": 20770
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9933888041708593e-05,
+      "loss": 0.7552,
+      "step": 20775
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9933759748638043e-05,
+      "loss": 0.7572,
+      "step": 20780
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9933631331622686e-05,
+      "loss": 0.7553,
+      "step": 20785
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.993350279066412e-05,
+      "loss": 0.7397,
+      "step": 20790
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.993337412576396e-05,
+      "loss": 0.7449,
+      "step": 20795
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9933245336923798e-05,
+      "loss": 0.7519,
+      "step": 20800
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.993311642414525e-05,
+      "loss": 0.752,
+      "step": 20805
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9932987387429924e-05,
+      "loss": 0.7341,
+      "step": 20810
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9932858226779426e-05,
+      "loss": 0.75,
+      "step": 20815
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.993272894219537e-05,
+      "loss": 0.7529,
+      "step": 20820
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.993259953367937e-05,
+      "loss": 0.7608,
+      "step": 20825
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.993247000123304e-05,
+      "loss": 0.7517,
+      "step": 20830
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.993234034485799e-05,
+      "loss": 0.7501,
+      "step": 20835
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.993221056455585e-05,
+      "loss": 0.7485,
+      "step": 20840
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.993208066032823e-05,
+      "loss": 0.7538,
+      "step": 20845
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9931950632176753e-05,
+      "loss": 0.7591,
+      "step": 20850
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9931820480103037e-05,
+      "loss": 0.7675,
+      "step": 20855
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9931690204108716e-05,
+      "loss": 0.7677,
+      "step": 20860
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9931559804195413e-05,
+      "loss": 0.7464,
+      "step": 20865
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9931429280364747e-05,
+      "loss": 0.7474,
+      "step": 20870
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9931298632618355e-05,
+      "loss": 0.7398,
+      "step": 20875
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9931167860957864e-05,
+      "loss": 0.7535,
+      "step": 20880
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.99310369653849e-05,
+      "loss": 0.7442,
+      "step": 20885
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9930905945901106e-05,
+      "loss": 0.7478,
+      "step": 20890
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9930774802508113e-05,
+      "loss": 0.7484,
+      "step": 20895
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9930643535207556e-05,
+      "loss": 0.7555,
+      "step": 20900
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9930512144001074e-05,
+      "loss": 0.7462,
+      "step": 20905
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9930380628890305e-05,
+      "loss": 0.7631,
+      "step": 20910
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.993024898987689e-05,
+      "loss": 0.7579,
+      "step": 20915
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9930117226962475e-05,
+      "loss": 0.7509,
+      "step": 20920
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.99299853401487e-05,
+      "loss": 0.7524,
+      "step": 20925
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.992985332943721e-05,
+      "loss": 0.748,
+      "step": 20930
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9929721194829657e-05,
+      "loss": 0.7531,
+      "step": 20935
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9929588936327686e-05,
+      "loss": 0.746,
+      "step": 20940
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.992945655393295e-05,
+      "loss": 0.7464,
+      "step": 20945
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9929324047647095e-05,
+      "loss": 0.7463,
+      "step": 20950
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.992919141747178e-05,
+      "loss": 0.7512,
+      "step": 20955
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9929058663408656e-05,
+      "loss": 0.7466,
+      "step": 20960
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9928925785459385e-05,
+      "loss": 0.7504,
+      "step": 20965
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9928792783625615e-05,
+      "loss": 0.7537,
+      "step": 20970
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.992865965790902e-05,
+      "loss": 0.7646,
+      "step": 20975
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9928526408311245e-05,
+      "loss": 0.7401,
+      "step": 20980
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9928393034833964e-05,
+      "loss": 0.7527,
+      "step": 20985
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9928259537478832e-05,
+      "loss": 0.7539,
+      "step": 20990
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9928125916247528e-05,
+      "loss": 0.7515,
+      "step": 20995
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9927992171141707e-05,
+      "loss": 0.7422,
+      "step": 21000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9927858302163044e-05,
+      "loss": 0.7408,
+      "step": 21005
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9927724309313205e-05,
+      "loss": 0.756,
+      "step": 21010
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9927590192593867e-05,
+      "loss": 0.7471,
+      "step": 21015
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9927455952006695e-05,
+      "loss": 0.756,
+      "step": 21020
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9927321587553378e-05,
+      "loss": 0.7339,
+      "step": 21025
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9927187099235577e-05,
+      "loss": 0.7436,
+      "step": 21030
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.992705248705498e-05,
+      "loss": 0.743,
+      "step": 21035
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9926917751013262e-05,
+      "loss": 0.7502,
+      "step": 21040
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9926782891112103e-05,
+      "loss": 0.7494,
+      "step": 21045
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9926647907353198e-05,
+      "loss": 0.7562,
+      "step": 21050
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.992651279973821e-05,
+      "loss": 0.7415,
+      "step": 21055
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.992637756826884e-05,
+      "loss": 0.7481,
+      "step": 21060
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9926242212946777e-05,
+      "loss": 0.7561,
+      "step": 21065
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.99261067337737e-05,
+      "loss": 0.7392,
+      "step": 21070
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.99259711307513e-05,
+      "loss": 0.7388,
+      "step": 21075
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.992583540388128e-05,
+      "loss": 0.7475,
+      "step": 21080
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9925699553165324e-05,
+      "loss": 0.7446,
+      "step": 21085
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.992556357860513e-05,
+      "loss": 0.7507,
+      "step": 21090
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.992542748020239e-05,
+      "loss": 0.7611,
+      "step": 21095
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.992529125795881e-05,
+      "loss": 0.7403,
+      "step": 21100
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9925154911876085e-05,
+      "loss": 0.7435,
+      "step": 21105
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9925018441955916e-05,
+      "loss": 0.758,
+      "step": 21110
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9924881848200007e-05,
+      "loss": 0.7523,
+      "step": 21115
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.992474513061006e-05,
+      "loss": 0.7495,
+      "step": 21120
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9924608289187786e-05,
+      "loss": 0.7477,
+      "step": 21125
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9924471323934885e-05,
+      "loss": 0.75,
+      "step": 21130
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9924334234853076e-05,
+      "loss": 0.7493,
+      "step": 21135
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.992419702194406e-05,
+      "loss": 0.7628,
+      "step": 21140
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9924059685209548e-05,
+      "loss": 0.7462,
+      "step": 21145
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9923922224651265e-05,
+      "loss": 0.7555,
+      "step": 21150
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9923784640270914e-05,
+      "loss": 0.7566,
+      "step": 21155
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9923646932070223e-05,
+      "loss": 0.7514,
+      "step": 21160
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9923509100050903e-05,
+      "loss": 0.7528,
+      "step": 21165
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.992337114421467e-05,
+      "loss": 0.7454,
+      "step": 21170
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9923233064563253e-05,
+      "loss": 0.7561,
+      "step": 21175
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9923094861098372e-05,
+      "loss": 0.7446,
+      "step": 21180
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.992295653382175e-05,
+      "loss": 0.7527,
+      "step": 21185
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9922818082735116e-05,
+      "loss": 0.7444,
+      "step": 21190
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9922679507840197e-05,
+      "loss": 0.7492,
+      "step": 21195
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9922540809138716e-05,
+      "loss": 0.7507,
+      "step": 21200
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9922401986632414e-05,
+      "loss": 0.7476,
+      "step": 21205
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9922263040323013e-05,
+      "loss": 0.7543,
+      "step": 21210
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9922123970212256e-05,
+      "loss": 0.7479,
+      "step": 21215
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9921984776301867e-05,
+      "loss": 0.7459,
+      "step": 21220
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9921845458593595e-05,
+      "loss": 0.7333,
+      "step": 21225
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.992170601708917e-05,
+      "loss": 0.7551,
+      "step": 21230
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.992156645179033e-05,
+      "loss": 0.7441,
+      "step": 21235
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9921426762698827e-05,
+      "loss": 0.7508,
+      "step": 21240
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9921286949816394e-05,
+      "loss": 0.7423,
+      "step": 21245
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9921147013144782e-05,
+      "loss": 0.7407,
+      "step": 21250
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.992100695268573e-05,
+      "loss": 0.7472,
+      "step": 21255
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9920866768440993e-05,
+      "loss": 0.7344,
+      "step": 21260
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9920726460412312e-05,
+      "loss": 0.7512,
+      "step": 21265
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9920586028601445e-05,
+      "loss": 0.754,
+      "step": 21270
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.992044547301014e-05,
+      "loss": 0.7401,
+      "step": 21275
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9920304793640155e-05,
+      "loss": 0.7535,
+      "step": 21280
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.992016399049324e-05,
+      "loss": 0.7521,
+      "step": 21285
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9920023063571153e-05,
+      "loss": 0.7449,
+      "step": 21290
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9919882012875656e-05,
+      "loss": 0.755,
+      "step": 21295
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9919740838408506e-05,
+      "loss": 0.7436,
+      "step": 21300
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9919599540171466e-05,
+      "loss": 0.7507,
+      "step": 21305
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9919458118166297e-05,
+      "loss": 0.7493,
+      "step": 21310
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9919316572394764e-05,
+      "loss": 0.747,
+      "step": 21315
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9919174902858632e-05,
+      "loss": 0.7504,
+      "step": 21320
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9919033109559677e-05,
+      "loss": 0.7523,
+      "step": 21325
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.991889119249965e-05,
+      "loss": 0.7427,
+      "step": 21330
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.991874915168034e-05,
+      "loss": 0.7415,
+      "step": 21335
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9918606987103515e-05,
+      "loss": 0.7398,
+      "step": 21340
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9918464698770944e-05,
+      "loss": 0.7481,
+      "step": 21345
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9918322286684402e-05,
+      "loss": 0.7391,
+      "step": 21350
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.991817975084567e-05,
+      "loss": 0.7516,
+      "step": 21355
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9918037091256527e-05,
+      "loss": 0.7411,
+      "step": 21360
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9917894307918745e-05,
+      "loss": 0.7472,
+      "step": 21365
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9917751400834114e-05,
+      "loss": 0.7569,
+      "step": 21370
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9917608370004417e-05,
+      "loss": 0.7466,
+      "step": 21375
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9917465215431434e-05,
+      "loss": 0.7383,
+      "step": 21380
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9917321937116955e-05,
+      "loss": 0.7354,
+      "step": 21385
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9917178535062764e-05,
+      "loss": 0.7418,
+      "step": 21390
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9917035009270652e-05,
+      "loss": 0.7549,
+      "step": 21395
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.991689135974241e-05,
+      "loss": 0.7441,
+      "step": 21400
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.991674758647983e-05,
+      "loss": 0.7462,
+      "step": 21405
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9916603689484707e-05,
+      "loss": 0.7466,
+      "step": 21410
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9916459668758837e-05,
+      "loss": 0.758,
+      "step": 21415
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.991631552430401e-05,
+      "loss": 0.7412,
+      "step": 21420
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9916171256122036e-05,
+      "loss": 0.7408,
+      "step": 21425
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9916026864214707e-05,
+      "loss": 0.7451,
+      "step": 21430
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9915882348583827e-05,
+      "loss": 0.7312,
+      "step": 21435
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.99157377092312e-05,
+      "loss": 0.7475,
+      "step": 21440
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9915592946158626e-05,
+      "loss": 0.7532,
+      "step": 21445
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9915448059367916e-05,
+      "loss": 0.744,
+      "step": 21450
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9915303048860876e-05,
+      "loss": 0.7455,
+      "step": 21455
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.991515791463932e-05,
+      "loss": 0.7555,
+      "step": 21460
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.991501265670505e-05,
+      "loss": 0.7522,
+      "step": 21465
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9914867275059887e-05,
+      "loss": 0.7358,
+      "step": 21470
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9914721769705637e-05,
+      "loss": 0.7435,
+      "step": 21475
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.991457614064412e-05,
+      "loss": 0.7526,
+      "step": 21480
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9914430387877155e-05,
+      "loss": 0.7473,
+      "step": 21485
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9914284511406557e-05,
+      "loss": 0.7292,
+      "step": 21490
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9914138511234145e-05,
+      "loss": 0.7544,
+      "step": 21495
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9913992387361747e-05,
+      "loss": 0.7402,
+      "step": 21500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.991384613979118e-05,
+      "loss": 0.7463,
+      "step": 21505
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.991369976852427e-05,
+      "loss": 0.7579,
+      "step": 21510
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9913553273562846e-05,
+      "loss": 0.7417,
+      "step": 21515
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9913406654908732e-05,
+      "loss": 0.7432,
+      "step": 21520
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9913259912563762e-05,
+      "loss": 0.7408,
+      "step": 21525
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9913113046529762e-05,
+      "loss": 0.7525,
+      "step": 21530
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.991296605680857e-05,
+      "loss": 0.7534,
+      "step": 21535
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9912818943402013e-05,
+      "loss": 0.7328,
+      "step": 21540
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.991267170631193e-05,
+      "loss": 0.7535,
+      "step": 21545
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9912524345540164e-05,
+      "loss": 0.7405,
+      "step": 21550
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.991237686108854e-05,
+      "loss": 0.7383,
+      "step": 21555
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9912229252958914e-05,
+      "loss": 0.7365,
+      "step": 21560
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9912081521153114e-05,
+      "loss": 0.7566,
+      "step": 21565
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9911933665672993e-05,
+      "loss": 0.7426,
+      "step": 21570
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.991178568652039e-05,
+      "loss": 0.7391,
+      "step": 21575
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9911637583697155e-05,
+      "loss": 0.7313,
+      "step": 21580
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.991148935720513e-05,
+      "loss": 0.7476,
+      "step": 21585
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9911341007046175e-05,
+      "loss": 0.7496,
+      "step": 21590
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.991119253322213e-05,
+      "loss": 0.7421,
+      "step": 21595
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9911043935734855e-05,
+      "loss": 0.744,
+      "step": 21600
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9910895214586198e-05,
+      "loss": 0.7467,
+      "step": 21605
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.991074636977802e-05,
+      "loss": 0.756,
+      "step": 21610
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9910597401312176e-05,
+      "loss": 0.7465,
+      "step": 21615
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9910448309190525e-05,
+      "loss": 0.7468,
+      "step": 21620
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.991029909341493e-05,
+      "loss": 0.7523,
+      "step": 21625
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9910149753987244e-05,
+      "loss": 0.7619,
+      "step": 21630
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9910000290909343e-05,
+      "loss": 0.7455,
+      "step": 21635
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.990985070418308e-05,
+      "loss": 0.7539,
+      "step": 21640
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9909700993810327e-05,
+      "loss": 0.7573,
+      "step": 21645
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.990955115979295e-05,
+      "loss": 0.7484,
+      "step": 21650
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9909401202132824e-05,
+      "loss": 0.7373,
+      "step": 21655
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9909251120831813e-05,
+      "loss": 0.7402,
+      "step": 21660
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9909100915891795e-05,
+      "loss": 0.7406,
+      "step": 21665
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.990895058731464e-05,
+      "loss": 0.7625,
+      "step": 21670
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.990880013510222e-05,
+      "loss": 0.7369,
+      "step": 21675
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9908649559256426e-05,
+      "loss": 0.7432,
+      "step": 21680
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9908498859779125e-05,
+      "loss": 0.7486,
+      "step": 21685
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9908348036672202e-05,
+      "loss": 0.7473,
+      "step": 21690
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9908197089937534e-05,
+      "loss": 0.7395,
+      "step": 21695
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.990804601957701e-05,
+      "loss": 0.7524,
+      "step": 21700
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9907894825592514e-05,
+      "loss": 0.749,
+      "step": 21705
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9907743507985926e-05,
+      "loss": 0.7343,
+      "step": 21710
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9907592066759146e-05,
+      "loss": 0.745,
+      "step": 21715
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9907440501914053e-05,
+      "loss": 0.7319,
+      "step": 21720
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.990728881345254e-05,
+      "loss": 0.7519,
+      "step": 21725
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9907137001376506e-05,
+      "loss": 0.7428,
+      "step": 21730
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9906985065687835e-05,
+      "loss": 0.7515,
+      "step": 21735
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9906833006388435e-05,
+      "loss": 0.7365,
+      "step": 21740
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9906680823480192e-05,
+      "loss": 0.7501,
+      "step": 21745
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.990652851696501e-05,
+      "loss": 0.7409,
+      "step": 21750
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.990637608684479e-05,
+      "loss": 0.7494,
+      "step": 21755
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9906223533121435e-05,
+      "loss": 0.7548,
+      "step": 21760
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9906070855796842e-05,
+      "loss": 0.7475,
+      "step": 21765
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.990591805487292e-05,
+      "loss": 0.7504,
+      "step": 21770
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9905765130351577e-05,
+      "loss": 0.7441,
+      "step": 21775
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.990561208223472e-05,
+      "loss": 0.7542,
+      "step": 21780
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9905458910524256e-05,
+      "loss": 0.7541,
+      "step": 21785
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9905305615222104e-05,
+      "loss": 0.7489,
+      "step": 21790
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9905152196330168e-05,
+      "loss": 0.7349,
+      "step": 21795
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.990499865385036e-05,
+      "loss": 0.7535,
+      "step": 21800
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.990484498778461e-05,
+      "loss": 0.7445,
+      "step": 21805
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.990469119813482e-05,
+      "loss": 0.7434,
+      "step": 21810
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9904537284902917e-05,
+      "loss": 0.749,
+      "step": 21815
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9904383248090823e-05,
+      "loss": 0.7475,
+      "step": 21820
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9904229087700454e-05,
+      "loss": 0.7442,
+      "step": 21825
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9904074803733735e-05,
+      "loss": 0.7376,
+      "step": 21830
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9903920396192592e-05,
+      "loss": 0.741,
+      "step": 21835
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9903765865078952e-05,
+      "loss": 0.7407,
+      "step": 21840
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.990361121039474e-05,
+      "loss": 0.7428,
+      "step": 21845
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9903456432141898e-05,
+      "loss": 0.7466,
+      "step": 21850
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.990330153032234e-05,
+      "loss": 0.7467,
+      "step": 21855
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9903146504938005e-05,
+      "loss": 0.7546,
+      "step": 21860
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9902991355990832e-05,
+      "loss": 0.7433,
+      "step": 21865
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.990283608348275e-05,
+      "loss": 0.7391,
+      "step": 21870
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9902680687415704e-05,
+      "loss": 0.7386,
+      "step": 21875
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9902525167791626e-05,
+      "loss": 0.7452,
+      "step": 21880
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.990236952461246e-05,
+      "loss": 0.7432,
+      "step": 21885
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9902213757880147e-05,
+      "loss": 0.7407,
+      "step": 21890
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9902057867596633e-05,
+      "loss": 0.7528,
+      "step": 21895
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9901901853763857e-05,
+      "loss": 0.7354,
+      "step": 21900
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9901745716383773e-05,
+      "loss": 0.7481,
+      "step": 21905
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9901589455458322e-05,
+      "loss": 0.7438,
+      "step": 21910
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9901433070989458e-05,
+      "loss": 0.734,
+      "step": 21915
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.990127656297913e-05,
+      "loss": 0.7508,
+      "step": 21920
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9901119931429294e-05,
+      "loss": 0.7374,
+      "step": 21925
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9900963176341902e-05,
+      "loss": 0.7486,
+      "step": 21930
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9900806297718912e-05,
+      "loss": 0.751,
+      "step": 21935
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9900649295562275e-05,
+      "loss": 0.7366,
+      "step": 21940
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9900492169873957e-05,
+      "loss": 0.7401,
+      "step": 21945
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.990033492065592e-05,
+      "loss": 0.7289,
+      "step": 21950
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9900177547910116e-05,
+      "loss": 0.7459,
+      "step": 21955
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9900020051638517e-05,
+      "loss": 0.7398,
+      "step": 21960
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9899862431843087e-05,
+      "loss": 0.7425,
+      "step": 21965
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.989970468852579e-05,
+      "loss": 0.7455,
+      "step": 21970
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9899546821688597e-05,
+      "loss": 0.7408,
+      "step": 21975
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9899388831333475e-05,
+      "loss": 0.7512,
+      "step": 21980
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9899230717462398e-05,
+      "loss": 0.744,
+      "step": 21985
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9899072480077337e-05,
+      "loss": 0.7425,
+      "step": 21990
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9898914119180265e-05,
+      "loss": 0.7343,
+      "step": 21995
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.989875563477316e-05,
+      "loss": 0.737,
+      "step": 22000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9898597026858e-05,
+      "loss": 0.7428,
+      "step": 22005
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.989843829543676e-05,
+      "loss": 0.7307,
+      "step": 22010
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.989827944051143e-05,
+      "loss": 0.7353,
+      "step": 22015
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9898120462083977e-05,
+      "loss": 0.7449,
+      "step": 22020
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9897961360156398e-05,
+      "loss": 0.7391,
+      "step": 22025
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9897802134730673e-05,
+      "loss": 0.7261,
+      "step": 22030
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.989764278580879e-05,
+      "loss": 0.7459,
+      "step": 22035
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9897483313392735e-05,
+      "loss": 0.7344,
+      "step": 22040
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.98973237174845e-05,
+      "loss": 0.7396,
+      "step": 22045
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9897163998086076e-05,
+      "loss": 0.7376,
+      "step": 22050
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9897004155199452e-05,
+      "loss": 0.7331,
+      "step": 22055
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.989684418882663e-05,
+      "loss": 0.7517,
+      "step": 22060
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9896684098969598e-05,
+      "loss": 0.7413,
+      "step": 22065
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9896523885630356e-05,
+      "loss": 0.7488,
+      "step": 22070
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9896363548810905e-05,
+      "loss": 0.7415,
+      "step": 22075
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9896203088513245e-05,
+      "loss": 0.7442,
+      "step": 22080
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9896042504739377e-05,
+      "loss": 0.7391,
+      "step": 22085
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9895881797491307e-05,
+      "loss": 0.7602,
+      "step": 22090
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9895720966771034e-05,
+      "loss": 0.7378,
+      "step": 22095
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9895560012580574e-05,
+      "loss": 0.7416,
+      "step": 22100
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9895398934921927e-05,
+      "loss": 0.7429,
+      "step": 22105
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9895237733797107e-05,
+      "loss": 0.7442,
+      "step": 22110
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.989507640920812e-05,
+      "loss": 0.745,
+      "step": 22115
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.989491496115699e-05,
+      "loss": 0.7458,
+      "step": 22120
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9894753389645723e-05,
+      "loss": 0.7415,
+      "step": 22125
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9894591694676334e-05,
+      "loss": 0.7358,
+      "step": 22130
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9894429876250844e-05,
+      "loss": 0.7338,
+      "step": 22135
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9894267934371273e-05,
+      "loss": 0.7483,
+      "step": 22140
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9894105869039636e-05,
+      "loss": 0.7427,
+      "step": 22145
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9893943680257964e-05,
+      "loss": 0.7357,
+      "step": 22150
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9893781368028274e-05,
+      "loss": 0.7382,
+      "step": 22155
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.989361893235259e-05,
+      "loss": 0.7373,
+      "step": 22160
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9893456373232944e-05,
+      "loss": 0.7276,
+      "step": 22165
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.989329369067136e-05,
+      "loss": 0.7338,
+      "step": 22170
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.989313088466987e-05,
+      "loss": 0.7416,
+      "step": 22175
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9892967955230504e-05,
+      "loss": 0.733,
+      "step": 22180
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9892804902355293e-05,
+      "loss": 0.7419,
+      "step": 22185
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9892641726046277e-05,
+      "loss": 0.7412,
+      "step": 22190
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9892478426305492e-05,
+      "loss": 0.7458,
+      "step": 22195
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9892315003134968e-05,
+      "loss": 0.726,
+      "step": 22200
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.989215145653675e-05,
+      "loss": 0.7427,
+      "step": 22205
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.989198778651288e-05,
+      "loss": 0.7497,
+      "step": 22210
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9891823993065393e-05,
+      "loss": 0.7308,
+      "step": 22215
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9891660076196338e-05,
+      "loss": 0.7517,
+      "step": 22220
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.989149603590776e-05,
+      "loss": 0.7437,
+      "step": 22225
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9891331872201707e-05,
+      "loss": 0.7504,
+      "step": 22230
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9891167585080226e-05,
+      "loss": 0.7558,
+      "step": 22235
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.989100317454536e-05,
+      "loss": 0.7384,
+      "step": 22240
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.989083864059917e-05,
+      "loss": 0.7518,
+      "step": 22245
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9890673983243708e-05,
+      "loss": 0.7517,
+      "step": 22250
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.989050920248102e-05,
+      "loss": 0.7406,
+      "step": 22255
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9890344298313175e-05,
+      "loss": 0.7296,
+      "step": 22260
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9890179270742217e-05,
+      "loss": 0.7422,
+      "step": 22265
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9890014119770214e-05,
+      "loss": 0.7461,
+      "step": 22270
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9889848845399226e-05,
+      "loss": 0.7561,
+      "step": 22275
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.988968344763131e-05,
+      "loss": 0.7494,
+      "step": 22280
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9889517926468534e-05,
+      "loss": 0.7431,
+      "step": 22285
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9889352281912963e-05,
+      "loss": 0.7364,
+      "step": 22290
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.988918651396666e-05,
+      "loss": 0.7455,
+      "step": 22295
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.98890206226317e-05,
+      "loss": 0.7318,
+      "step": 22300
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.988885460791015e-05,
+      "loss": 0.7369,
+      "step": 22305
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9888688469804075e-05,
+      "loss": 0.7461,
+      "step": 22310
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9888522208315557e-05,
+      "loss": 0.743,
+      "step": 22315
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.988835582344667e-05,
+      "loss": 0.7438,
+      "step": 22320
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.988818931519948e-05,
+      "loss": 0.7326,
+      "step": 22325
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9888022683576073e-05,
+      "loss": 0.7356,
+      "step": 22330
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9887855928578527e-05,
+      "loss": 0.7334,
+      "step": 22335
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9887689050208924e-05,
+      "loss": 0.742,
+      "step": 22340
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9887522048469342e-05,
+      "loss": 0.743,
+      "step": 22345
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.988735492336187e-05,
+      "loss": 0.7406,
+      "step": 22350
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9887187674888587e-05,
+      "loss": 0.7334,
+      "step": 22355
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9887020303051586e-05,
+      "loss": 0.7201,
+      "step": 22360
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.988685280785295e-05,
+      "loss": 0.7406,
+      "step": 22365
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.988668518929477e-05,
+      "loss": 0.7312,
+      "step": 22370
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.988651744737914e-05,
+      "loss": 0.7424,
+      "step": 22375
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9886349582108154e-05,
+      "loss": 0.7396,
+      "step": 22380
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.98861815934839e-05,
+      "loss": 0.7373,
+      "step": 22385
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.988601348150848e-05,
+      "loss": 0.7389,
+      "step": 22390
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.988584524618399e-05,
+      "loss": 0.7361,
+      "step": 22395
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.988567688751253e-05,
+      "loss": 0.7269,
+      "step": 22400
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9885508405496193e-05,
+      "loss": 0.7258,
+      "step": 22405
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9885339800137094e-05,
+      "loss": 0.757,
+      "step": 22410
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9885171071437327e-05,
+      "loss": 0.7379,
+      "step": 22415
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9885002219399e-05,
+      "loss": 0.7412,
+      "step": 22420
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9884833244024226e-05,
+      "loss": 0.7325,
+      "step": 22425
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.98846641453151e-05,
+      "loss": 0.7347,
+      "step": 22430
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9884494923273742e-05,
+      "loss": 0.7194,
+      "step": 22435
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9884325577902262e-05,
+      "loss": 0.7506,
+      "step": 22440
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.988415610920277e-05,
+      "loss": 0.731,
+      "step": 22445
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9883986517177382e-05,
+      "loss": 0.7321,
+      "step": 22450
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9883816801828216e-05,
+      "loss": 0.7339,
+      "step": 22455
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.988364696315739e-05,
+      "loss": 0.7262,
+      "step": 22460
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9883477001167012e-05,
+      "loss": 0.7422,
+      "step": 22465
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.988330691585922e-05,
+      "loss": 0.732,
+      "step": 22470
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9883136707236127e-05,
+      "loss": 0.7321,
+      "step": 22475
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9882966375299854e-05,
+      "loss": 0.7416,
+      "step": 22480
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9882795920052532e-05,
+      "loss": 0.749,
+      "step": 22485
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9882625341496285e-05,
+      "loss": 0.7347,
+      "step": 22490
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.988245463963324e-05,
+      "loss": 0.7553,
+      "step": 22495
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.988228381446553e-05,
+      "loss": 0.7413,
+      "step": 22500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9882112865995284e-05,
+      "loss": 0.7365,
+      "step": 22505
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9881941794224636e-05,
+      "loss": 0.7411,
+      "step": 22510
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9881770599155726e-05,
+      "loss": 0.737,
+      "step": 22515
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.988159928079068e-05,
+      "loss": 0.752,
+      "step": 22520
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9881427839131637e-05,
+      "loss": 0.7335,
+      "step": 22525
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9881256274180745e-05,
+      "loss": 0.7475,
+      "step": 22530
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9881084585940135e-05,
+      "loss": 0.7385,
+      "step": 22535
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9880912774411955e-05,
+      "loss": 0.7372,
+      "step": 22540
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9880740839598346e-05,
+      "loss": 0.7352,
+      "step": 22545
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9880568781501454e-05,
+      "loss": 0.7248,
+      "step": 22550
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9880396600123425e-05,
+      "loss": 0.7408,
+      "step": 22555
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.988022429546641e-05,
+      "loss": 0.7553,
+      "step": 22560
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9880051867532556e-05,
+      "loss": 0.7317,
+      "step": 22565
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9879879316324015e-05,
+      "loss": 0.7316,
+      "step": 22570
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.987970664184294e-05,
+      "loss": 0.7351,
+      "step": 22575
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.987953384409149e-05,
+      "loss": 0.7291,
+      "step": 22580
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9879360923071812e-05,
+      "loss": 0.7223,
+      "step": 22585
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9879187878786067e-05,
+      "loss": 0.7425,
+      "step": 22590
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.987901471123642e-05,
+      "loss": 0.7431,
+      "step": 22595
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9878841420425023e-05,
+      "loss": 0.7328,
+      "step": 22600
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9878668006354044e-05,
+      "loss": 0.7363,
+      "step": 22605
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9878494469025646e-05,
+      "loss": 0.7335,
+      "step": 22610
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9878320808441994e-05,
+      "loss": 0.7388,
+      "step": 22615
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.987814702460525e-05,
+      "loss": 0.7355,
+      "step": 22620
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.987797311751759e-05,
+      "loss": 0.7352,
+      "step": 22625
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.987779908718118e-05,
+      "loss": 0.7416,
+      "step": 22630
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.987762493359819e-05,
+      "loss": 0.7338,
+      "step": 22635
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9877450656770794e-05,
+      "loss": 0.7339,
+      "step": 22640
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.987727625670117e-05,
+      "loss": 0.7248,
+      "step": 22645
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.987710173339149e-05,
+      "loss": 0.7397,
+      "step": 22650
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.987692708684393e-05,
+      "loss": 0.7482,
+      "step": 22655
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9876752317060676e-05,
+      "loss": 0.7455,
+      "step": 22660
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.98765774240439e-05,
+      "loss": 0.762,
+      "step": 22665
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.987640240779579e-05,
+      "loss": 0.7369,
+      "step": 22670
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.987622726831853e-05,
+      "loss": 0.7299,
+      "step": 22675
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.98760520056143e-05,
+      "loss": 0.7314,
+      "step": 22680
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9875876619685294e-05,
+      "loss": 0.7276,
+      "step": 22685
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9875701110533693e-05,
+      "loss": 0.7434,
+      "step": 22690
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.987552547816169e-05,
+      "loss": 0.7272,
+      "step": 22695
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.987534972257148e-05,
+      "loss": 0.7213,
+      "step": 22700
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9875173843765252e-05,
+      "loss": 0.7402,
+      "step": 22705
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9874997841745202e-05,
+      "loss": 0.7382,
+      "step": 22710
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.987482171651352e-05,
+      "loss": 0.7446,
+      "step": 22715
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9874645468072417e-05,
+      "loss": 0.7345,
+      "step": 22720
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9874469096424075e-05,
+      "loss": 0.7389,
+      "step": 22725
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.987429260157071e-05,
+      "loss": 0.7371,
+      "step": 22730
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9874115983514513e-05,
+      "loss": 0.7424,
+      "step": 22735
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9873939242257694e-05,
+      "loss": 0.7268,
+      "step": 22740
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9873762377802453e-05,
+      "loss": 0.7375,
+      "step": 22745
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9873585390151003e-05,
+      "loss": 0.7337,
+      "step": 22750
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.987340827930555e-05,
+      "loss": 0.7388,
+      "step": 22755
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9873231045268303e-05,
+      "loss": 0.7341,
+      "step": 22760
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9873053688041474e-05,
+      "loss": 0.7263,
+      "step": 22765
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9872876207627276e-05,
+      "loss": 0.7549,
+      "step": 22770
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9872698604027924e-05,
+      "loss": 0.7168,
+      "step": 22775
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9872520877245628e-05,
+      "loss": 0.7477,
+      "step": 22780
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9872343027282617e-05,
+      "loss": 0.7286,
+      "step": 22785
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.98721650541411e-05,
+      "loss": 0.7367,
+      "step": 22790
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.98719869578233e-05,
+      "loss": 0.7403,
+      "step": 22795
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.987180873833144e-05,
+      "loss": 0.7308,
+      "step": 22800
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9871630395667748e-05,
+      "loss": 0.733,
+      "step": 22805
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9871451929834443e-05,
+      "loss": 0.7312,
+      "step": 22810
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9871273340833752e-05,
+      "loss": 0.7434,
+      "step": 22815
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9871094628667905e-05,
+      "loss": 0.7418,
+      "step": 22820
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9870915793339137e-05,
+      "loss": 0.7237,
+      "step": 22825
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.987073683484967e-05,
+      "loss": 0.7344,
+      "step": 22830
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9870557753201742e-05,
+      "loss": 0.74,
+      "step": 22835
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9870378548397585e-05,
+      "loss": 0.7391,
+      "step": 22840
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9870199220439442e-05,
+      "loss": 0.7389,
+      "step": 22845
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.987001976932954e-05,
+      "loss": 0.7268,
+      "step": 22850
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9869840195070126e-05,
+      "loss": 0.7376,
+      "step": 22855
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9869660497663435e-05,
+      "loss": 0.7411,
+      "step": 22860
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9869480677111713e-05,
+      "loss": 0.7225,
+      "step": 22865
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9869300733417206e-05,
+      "loss": 0.7406,
+      "step": 22870
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9869120666582153e-05,
+      "loss": 0.7297,
+      "step": 22875
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9868940476608802e-05,
+      "loss": 0.7318,
+      "step": 22880
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9868760163499405e-05,
+      "loss": 0.7396,
+      "step": 22885
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9868579727256208e-05,
+      "loss": 0.7269,
+      "step": 22890
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.986839916788146e-05,
+      "loss": 0.7407,
+      "step": 22895
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9868218485377424e-05,
+      "loss": 0.7422,
+      "step": 22900
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9868037679746347e-05,
+      "loss": 0.7348,
+      "step": 22905
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9867856750990483e-05,
+      "loss": 0.7299,
+      "step": 22910
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9867675699112095e-05,
+      "loss": 0.733,
+      "step": 22915
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9867494524113436e-05,
+      "loss": 0.7428,
+      "step": 22920
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9867313225996773e-05,
+      "loss": 0.7427,
+      "step": 22925
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9867131804764366e-05,
+      "loss": 0.7314,
+      "step": 22930
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9866950260418475e-05,
+      "loss": 0.7192,
+      "step": 22935
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9866768592961366e-05,
+      "loss": 0.7274,
+      "step": 22940
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9866586802395313e-05,
+      "loss": 0.7452,
+      "step": 22945
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9866404888722575e-05,
+      "loss": 0.7245,
+      "step": 22950
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.986622285194543e-05,
+      "loss": 0.7313,
+      "step": 22955
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9866040692066137e-05,
+      "loss": 0.7285,
+      "step": 22960
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9865858409086984e-05,
+      "loss": 0.7407,
+      "step": 22965
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9865676003010233e-05,
+      "loss": 0.7259,
+      "step": 22970
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9865493473838168e-05,
+      "loss": 0.7279,
+      "step": 22975
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9865310821573063e-05,
+      "loss": 0.7427,
+      "step": 22980
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9865128046217195e-05,
+      "loss": 0.7392,
+      "step": 22985
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.986494514777285e-05,
+      "loss": 0.7402,
+      "step": 22990
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.98647621262423e-05,
+      "loss": 0.7414,
+      "step": 22995
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9864578981627844e-05,
+      "loss": 0.7378,
+      "step": 23000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9864395713931754e-05,
+      "loss": 0.7275,
+      "step": 23005
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9864212323156322e-05,
+      "loss": 0.7377,
+      "step": 23010
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9864028809303838e-05,
+      "loss": 0.7419,
+      "step": 23015
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9863845172376586e-05,
+      "loss": 0.7546,
+      "step": 23020
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.986366141237686e-05,
+      "loss": 0.7394,
+      "step": 23025
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9863477529306956e-05,
+      "loss": 0.7151,
+      "step": 23030
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9863293523169162e-05,
+      "loss": 0.725,
+      "step": 23035
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9863109393965782e-05,
+      "loss": 0.7434,
+      "step": 23040
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9862925141699106e-05,
+      "loss": 0.7545,
+      "step": 23045
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9862740766371434e-05,
+      "loss": 0.7446,
+      "step": 23050
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.986255626798507e-05,
+      "loss": 0.7247,
+      "step": 23055
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9862371646542314e-05,
+      "loss": 0.7348,
+      "step": 23060
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.986218690204547e-05,
+      "loss": 0.7357,
+      "step": 23065
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9862002034496846e-05,
+      "loss": 0.7409,
+      "step": 23070
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9861817043898743e-05,
+      "loss": 0.7297,
+      "step": 23075
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9861631930253468e-05,
+      "loss": 0.7367,
+      "step": 23080
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.986144669356334e-05,
+      "loss": 0.7444,
+      "step": 23085
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9861261333830662e-05,
+      "loss": 0.7532,
+      "step": 23090
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9861075851057747e-05,
+      "loss": 0.7471,
+      "step": 23095
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9860890245246915e-05,
+      "loss": 0.7361,
+      "step": 23100
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.986070451640048e-05,
+      "loss": 0.7359,
+      "step": 23105
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9860518664520752e-05,
+      "loss": 0.7309,
+      "step": 23110
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.986033268961006e-05,
+      "loss": 0.7426,
+      "step": 23115
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9860146591670717e-05,
+      "loss": 0.7241,
+      "step": 23120
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.985996037070505e-05,
+      "loss": 0.7322,
+      "step": 23125
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.985977402671538e-05,
+      "loss": 0.7338,
+      "step": 23130
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9859587559704033e-05,
+      "loss": 0.7236,
+      "step": 23135
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9859400969673338e-05,
+      "loss": 0.7234,
+      "step": 23140
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9859214256625616e-05,
+      "loss": 0.7372,
+      "step": 23145
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9859027420563203e-05,
+      "loss": 0.7377,
+      "step": 23150
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9858840461488425e-05,
+      "loss": 0.7324,
+      "step": 23155
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9858653379403623e-05,
+      "loss": 0.7277,
+      "step": 23160
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.985846617431112e-05,
+      "loss": 0.7281,
+      "step": 23165
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9858278846213266e-05,
+      "loss": 0.7304,
+      "step": 23170
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9858091395112383e-05,
+      "loss": 0.7322,
+      "step": 23175
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.985790382101082e-05,
+      "loss": 0.7301,
+      "step": 23180
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9857716123910913e-05,
+      "loss": 0.7442,
+      "step": 23185
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9857528303815008e-05,
+      "loss": 0.7227,
+      "step": 23190
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9857340360725447e-05,
+      "loss": 0.7389,
+      "step": 23195
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9857152294644567e-05,
+      "loss": 0.7135,
+      "step": 23200
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9856964105574727e-05,
+      "loss": 0.7364,
+      "step": 23205
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9856775793518267e-05,
+      "loss": 0.7145,
+      "step": 23210
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.985658735847754e-05,
+      "loss": 0.7239,
+      "step": 23215
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9856398800454893e-05,
+      "loss": 0.7353,
+      "step": 23220
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9856210119452685e-05,
+      "loss": 0.7284,
+      "step": 23225
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9856021315473266e-05,
+      "loss": 0.7482,
+      "step": 23230
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9855832388518995e-05,
+      "loss": 0.7383,
+      "step": 23235
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9855643338592223e-05,
+      "loss": 0.7322,
+      "step": 23240
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9855454165695315e-05,
+      "loss": 0.74,
+      "step": 23245
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.985526486983063e-05,
+      "loss": 0.72,
+      "step": 23250
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.985507545100053e-05,
+      "loss": 0.7229,
+      "step": 23255
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9854885909207372e-05,
+      "loss": 0.7239,
+      "step": 23260
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.985469624445353e-05,
+      "loss": 0.7278,
+      "step": 23265
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9854506456741366e-05,
+      "loss": 0.7367,
+      "step": 23270
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.985431654607325e-05,
+      "loss": 0.736,
+      "step": 23275
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.985412651245155e-05,
+      "loss": 0.7341,
+      "step": 23280
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9853936355878635e-05,
+      "loss": 0.7357,
+      "step": 23285
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9853746076356878e-05,
+      "loss": 0.7429,
+      "step": 23290
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.985355567388866e-05,
+      "loss": 0.7394,
+      "step": 23295
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.985336514847635e-05,
+      "loss": 0.7319,
+      "step": 23300
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.985317450012233e-05,
+      "loss": 0.7371,
+      "step": 23305
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9852983728828972e-05,
+      "loss": 0.7366,
+      "step": 23310
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.985279283459866e-05,
+      "loss": 0.7307,
+      "step": 23315
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9852601817433777e-05,
+      "loss": 0.7248,
+      "step": 23320
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9852410677336707e-05,
+      "loss": 0.7244,
+      "step": 23325
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9852219414309833e-05,
+      "loss": 0.732,
+      "step": 23330
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9852028028355538e-05,
+      "loss": 0.7417,
+      "step": 23335
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9851836519476217e-05,
+      "loss": 0.735,
+      "step": 23340
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9851644887674254e-05,
+      "loss": 0.7374,
+      "step": 23345
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9851453132952042e-05,
+      "loss": 0.7262,
+      "step": 23350
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9851261255311974e-05,
+      "loss": 0.7089,
+      "step": 23355
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9851069254756444e-05,
+      "loss": 0.7319,
+      "step": 23360
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9850877131287844e-05,
+      "loss": 0.731,
+      "step": 23365
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9850684884908577e-05,
+      "loss": 0.7409,
+      "step": 23370
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9850492515621038e-05,
+      "loss": 0.7355,
+      "step": 23375
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9850300023427632e-05,
+      "loss": 0.7363,
+      "step": 23380
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9850107408330752e-05,
+      "loss": 0.722,
+      "step": 23385
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9849914670332807e-05,
+      "loss": 0.7396,
+      "step": 23390
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9849721809436202e-05,
+      "loss": 0.73,
+      "step": 23395
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9849528825643346e-05,
+      "loss": 0.7283,
+      "step": 23400
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9849335718956638e-05,
+      "loss": 0.7281,
+      "step": 23405
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9849142489378494e-05,
+      "loss": 0.7252,
+      "step": 23410
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9848949136911325e-05,
+      "loss": 0.7277,
+      "step": 23415
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9848755661557542e-05,
+      "loss": 0.7457,
+      "step": 23420
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.984856206331956e-05,
+      "loss": 0.739,
+      "step": 23425
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.984836834219979e-05,
+      "loss": 0.7445,
+      "step": 23430
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9848174498200656e-05,
+      "loss": 0.7234,
+      "step": 23435
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9847980531324573e-05,
+      "loss": 0.7361,
+      "step": 23440
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.984778644157396e-05,
+      "loss": 0.7365,
+      "step": 23445
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.984759222895124e-05,
+      "loss": 0.7351,
+      "step": 23450
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.984739789345884e-05,
+      "loss": 0.7112,
+      "step": 23455
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9847203435099182e-05,
+      "loss": 0.7287,
+      "step": 23460
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9847008853874687e-05,
+      "loss": 0.7191,
+      "step": 23465
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.984681414978779e-05,
+      "loss": 0.7248,
+      "step": 23470
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9846619322840915e-05,
+      "loss": 0.7421,
+      "step": 23475
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.98464243730365e-05,
+      "loss": 0.7302,
+      "step": 23480
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9846229300376968e-05,
+      "loss": 0.7388,
+      "step": 23485
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.984603410486476e-05,
+      "loss": 0.7329,
+      "step": 23490
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9845838786502312e-05,
+      "loss": 0.722,
+      "step": 23495
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9845643345292055e-05,
+      "loss": 0.7307,
+      "step": 23500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.984544778123643e-05,
+      "loss": 0.7173,
+      "step": 23505
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9845252094337884e-05,
+      "loss": 0.7229,
+      "step": 23510
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9845056284598846e-05,
+      "loss": 0.7303,
+      "step": 23515
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9844860352021768e-05,
+      "loss": 0.7287,
+      "step": 23520
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9844664296609096e-05,
+      "loss": 0.7452,
+      "step": 23525
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9844468118363268e-05,
+      "loss": 0.7311,
+      "step": 23530
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.984427181728674e-05,
+      "loss": 0.7295,
+      "step": 23535
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9844075393381955e-05,
+      "loss": 0.724,
+      "step": 23540
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9843878846651367e-05,
+      "loss": 0.7262,
+      "step": 23545
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9843682177097427e-05,
+      "loss": 0.7343,
+      "step": 23550
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.984348538472259e-05,
+      "loss": 0.7339,
+      "step": 23555
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9843288469529313e-05,
+      "loss": 0.7332,
+      "step": 23560
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9843091431520053e-05,
+      "loss": 0.7311,
+      "step": 23565
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.984289427069726e-05,
+      "loss": 0.734,
+      "step": 23570
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9842696987063402e-05,
+      "loss": 0.7287,
+      "step": 23575
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9842499580620945e-05,
+      "loss": 0.7295,
+      "step": 23580
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9842302051372342e-05,
+      "loss": 0.7305,
+      "step": 23585
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.984210439932006e-05,
+      "loss": 0.7328,
+      "step": 23590
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.984190662446657e-05,
+      "loss": 0.7246,
+      "step": 23595
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9841708726814336e-05,
+      "loss": 0.7302,
+      "step": 23600
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9841510706365824e-05,
+      "loss": 0.7306,
+      "step": 23605
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.984131256312351e-05,
+      "loss": 0.732,
+      "step": 23610
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9841114297089868e-05,
+      "loss": 0.7377,
+      "step": 23615
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9840915908267367e-05,
+      "loss": 0.7272,
+      "step": 23620
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9840717396658483e-05,
+      "loss": 0.7272,
+      "step": 23625
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9840518762265695e-05,
+      "loss": 0.7339,
+      "step": 23630
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.984032000509148e-05,
+      "loss": 0.7385,
+      "step": 23635
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9840121125138317e-05,
+      "loss": 0.7266,
+      "step": 23640
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9839922122408688e-05,
+      "loss": 0.7471,
+      "step": 23645
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9839722996905076e-05,
+      "loss": 0.7169,
+      "step": 23650
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9839523748629967e-05,
+      "loss": 0.7369,
+      "step": 23655
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9839324377585845e-05,
+      "loss": 0.7259,
+      "step": 23660
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.98391248837752e-05,
+      "loss": 0.7256,
+      "step": 23665
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9838925267200517e-05,
+      "loss": 0.731,
+      "step": 23670
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.983872552786429e-05,
+      "loss": 0.7374,
+      "step": 23675
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9838525665769013e-05,
+      "loss": 0.7307,
+      "step": 23680
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9838325680917177e-05,
+      "loss": 0.733,
+      "step": 23685
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9838125573311276e-05,
+      "loss": 0.7294,
+      "step": 23690
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9837925342953808e-05,
+      "loss": 0.7277,
+      "step": 23695
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.983772498984727e-05,
+      "loss": 0.7152,
+      "step": 23700
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9837524513994165e-05,
+      "loss": 0.74,
+      "step": 23705
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9837323915396994e-05,
+      "loss": 0.7278,
+      "step": 23710
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9837123194058257e-05,
+      "loss": 0.7226,
+      "step": 23715
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.983692234998046e-05,
+      "loss": 0.7283,
+      "step": 23720
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.983672138316611e-05,
+      "loss": 0.7174,
+      "step": 23725
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.983652029361771e-05,
+      "loss": 0.7212,
+      "step": 23730
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9836319081337776e-05,
+      "loss": 0.7294,
+      "step": 23735
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9836117746328815e-05,
+      "loss": 0.722,
+      "step": 23740
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.983591628859334e-05,
+      "loss": 0.7319,
+      "step": 23745
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.983571470813386e-05,
+      "loss": 0.7327,
+      "step": 23750
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9835513004952895e-05,
+      "loss": 0.7356,
+      "step": 23755
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9835311179052963e-05,
+      "loss": 0.735,
+      "step": 23760
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9835109230436578e-05,
+      "loss": 0.7339,
+      "step": 23765
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9834907159106264e-05,
+      "loss": 0.7256,
+      "step": 23770
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9834704965064535e-05,
+      "loss": 0.7319,
+      "step": 23775
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.983450264831392e-05,
+      "loss": 0.7345,
+      "step": 23780
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9834300208856943e-05,
+      "loss": 0.7329,
+      "step": 23785
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9834097646696125e-05,
+      "loss": 0.736,
+      "step": 23790
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9833894961834003e-05,
+      "loss": 0.7148,
+      "step": 23795
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9833692154273097e-05,
+      "loss": 0.7292,
+      "step": 23800
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9833489224015944e-05,
+      "loss": 0.7269,
+      "step": 23805
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9833286171065067e-05,
+      "loss": 0.7207,
+      "step": 23810
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.983308299542301e-05,
+      "loss": 0.7147,
+      "step": 23815
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9832879697092297e-05,
+      "loss": 0.7183,
+      "step": 23820
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9832676276075476e-05,
+      "loss": 0.7508,
+      "step": 23825
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9832472732375078e-05,
+      "loss": 0.7316,
+      "step": 23830
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9832269065993643e-05,
+      "loss": 0.7206,
+      "step": 23835
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9832065276933716e-05,
+      "loss": 0.7305,
+      "step": 23840
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9831861365197834e-05,
+      "loss": 0.7215,
+      "step": 23845
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.983165733078855e-05,
+      "loss": 0.7265,
+      "step": 23850
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9831453173708398e-05,
+      "loss": 0.7326,
+      "step": 23855
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9831248893959937e-05,
+      "loss": 0.7289,
+      "step": 23860
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.983104449154571e-05,
+      "loss": 0.73,
+      "step": 23865
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9830839966468262e-05,
+      "loss": 0.7316,
+      "step": 23870
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9830635318730155e-05,
+      "loss": 0.7308,
+      "step": 23875
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.983043054833394e-05,
+      "loss": 0.728,
+      "step": 23880
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9830225655282164e-05,
+      "loss": 0.7221,
+      "step": 23885
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9830020639577397e-05,
+      "loss": 0.7317,
+      "step": 23890
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9829815501222184e-05,
+      "loss": 0.7328,
+      "step": 23895
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.982961024021909e-05,
+      "loss": 0.7305,
+      "step": 23900
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.982940485657068e-05,
+      "loss": 0.7318,
+      "step": 23905
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.982919935027951e-05,
+      "loss": 0.7376,
+      "step": 23910
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9828993721348146e-05,
+      "loss": 0.7381,
+      "step": 23915
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.982878796977915e-05,
+      "loss": 0.736,
+      "step": 23920
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9828582095575104e-05,
+      "loss": 0.7373,
+      "step": 23925
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.982837609873856e-05,
+      "loss": 0.7325,
+      "step": 23930
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9828169979272096e-05,
+      "loss": 0.7319,
+      "step": 23935
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9827963737178282e-05,
+      "loss": 0.7268,
+      "step": 23940
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9827757372459695e-05,
+      "loss": 0.7394,
+      "step": 23945
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9827550885118902e-05,
+      "loss": 0.7353,
+      "step": 23950
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9827344275158486e-05,
+      "loss": 0.7279,
+      "step": 23955
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9827137542581022e-05,
+      "loss": 0.7338,
+      "step": 23960
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9826930687389094e-05,
+      "loss": 0.7236,
+      "step": 23965
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9826723709585274e-05,
+      "loss": 0.7292,
+      "step": 23970
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.982651660917215e-05,
+      "loss": 0.7223,
+      "step": 23975
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.982630938615231e-05,
+      "loss": 0.7318,
+      "step": 23980
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9826102040528333e-05,
+      "loss": 0.7232,
+      "step": 23985
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9825894572302808e-05,
+      "loss": 0.7331,
+      "step": 23990
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9825686981478324e-05,
+      "loss": 0.7398,
+      "step": 23995
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.982547926805747e-05,
+      "loss": 0.728,
+      "step": 24000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9825271432042838e-05,
+      "loss": 0.7241,
+      "step": 24005
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9825063473437024e-05,
+      "loss": 0.7336,
+      "step": 24010
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9824855392242622e-05,
+      "loss": 0.7336,
+      "step": 24015
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9824647188462224e-05,
+      "loss": 0.7115,
+      "step": 24020
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9824438862098435e-05,
+      "loss": 0.729,
+      "step": 24025
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9824230413153843e-05,
+      "loss": 0.7295,
+      "step": 24030
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9824021841631062e-05,
+      "loss": 0.738,
+      "step": 24035
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9823813147532686e-05,
+      "loss": 0.7152,
+      "step": 24040
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.982360433086132e-05,
+      "loss": 0.7457,
+      "step": 24045
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9823395391619575e-05,
+      "loss": 0.7193,
+      "step": 24050
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.982318632981005e-05,
+      "loss": 0.7186,
+      "step": 24055
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9822977145435357e-05,
+      "loss": 0.7187,
+      "step": 24060
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9822767838498107e-05,
+      "loss": 0.7342,
+      "step": 24065
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9822558409000908e-05,
+      "loss": 0.709,
+      "step": 24070
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.982234885694638e-05,
+      "loss": 0.7281,
+      "step": 24075
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.982213918233713e-05,
+      "loss": 0.7366,
+      "step": 24080
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.982192938517578e-05,
+      "loss": 0.7384,
+      "step": 24085
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9821719465464942e-05,
+      "loss": 0.7323,
+      "step": 24090
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.982150942320724e-05,
+      "loss": 0.7202,
+      "step": 24095
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9821299258405297e-05,
+      "loss": 0.7261,
+      "step": 24100
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9821088971061725e-05,
+      "loss": 0.7334,
+      "step": 24105
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9820878561179157e-05,
+      "loss": 0.7267,
+      "step": 24110
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9820668028760215e-05,
+      "loss": 0.7191,
+      "step": 24115
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9820457373807525e-05,
+      "loss": 0.7301,
+      "step": 24120
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.982024659632372e-05,
+      "loss": 0.726,
+      "step": 24125
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9820035696311423e-05,
+      "loss": 0.725,
+      "step": 24130
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.981982467377327e-05,
+      "loss": 0.7304,
+      "step": 24135
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9819613528711896e-05,
+      "loss": 0.7251,
+      "step": 24140
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9819402261129928e-05,
+      "loss": 0.7216,
+      "step": 24145
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9819190871030005e-05,
+      "loss": 0.7225,
+      "step": 24150
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.981897935841477e-05,
+      "loss": 0.7254,
+      "step": 24155
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9818767723286855e-05,
+      "loss": 0.7277,
+      "step": 24160
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9818555965648907e-05,
+      "loss": 0.7143,
+      "step": 24165
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.981834408550356e-05,
+      "loss": 0.7208,
+      "step": 24170
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9818132082853466e-05,
+      "loss": 0.7178,
+      "step": 24175
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.981791995770127e-05,
+      "loss": 0.7289,
+      "step": 24180
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.981770771004961e-05,
+      "loss": 0.7271,
+      "step": 24185
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.981749533990114e-05,
+      "loss": 0.7216,
+      "step": 24190
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9817282847258507e-05,
+      "loss": 0.7305,
+      "step": 24195
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.981707023212437e-05,
+      "loss": 0.7222,
+      "step": 24200
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9816857494501374e-05,
+      "loss": 0.7192,
+      "step": 24205
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9816644634392173e-05,
+      "loss": 0.7183,
+      "step": 24210
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.981643165179943e-05,
+      "loss": 0.7284,
+      "step": 24215
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9816218546725796e-05,
+      "loss": 0.7232,
+      "step": 24220
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9816005319173932e-05,
+      "loss": 0.7154,
+      "step": 24225
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9815791969146495e-05,
+      "loss": 0.7223,
+      "step": 24230
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9815578496646158e-05,
+      "loss": 0.7233,
+      "step": 24235
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9815364901675573e-05,
+      "loss": 0.7298,
+      "step": 24240
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9815151184237407e-05,
+      "loss": 0.7278,
+      "step": 24245
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.981493734433433e-05,
+      "loss": 0.749,
+      "step": 24250
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.981472338196901e-05,
+      "loss": 0.7455,
+      "step": 24255
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9814509297144114e-05,
+      "loss": 0.7256,
+      "step": 24260
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9814295089862316e-05,
+      "loss": 0.7281,
+      "step": 24265
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9814080760126285e-05,
+      "loss": 0.7142,
+      "step": 24270
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.98138663079387e-05,
+      "loss": 0.7326,
+      "step": 24275
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9813651733302232e-05,
+      "loss": 0.7188,
+      "step": 24280
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9813437036219563e-05,
+      "loss": 0.7339,
+      "step": 24285
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9813222216693365e-05,
+      "loss": 0.7191,
+      "step": 24290
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9813007274726325e-05,
+      "loss": 0.7274,
+      "step": 24295
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9812792210321123e-05,
+      "loss": 0.7322,
+      "step": 24300
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9812577023480442e-05,
+      "loss": 0.7389,
+      "step": 24305
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9812361714206963e-05,
+      "loss": 0.7266,
+      "step": 24310
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9812146282503383e-05,
+      "loss": 0.7298,
+      "step": 24315
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.981193072837238e-05,
+      "loss": 0.7233,
+      "step": 24320
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9811715051816644e-05,
+      "loss": 0.7312,
+      "step": 24325
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9811499252838873e-05,
+      "loss": 0.7203,
+      "step": 24330
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.981128333144175e-05,
+      "loss": 0.7252,
+      "step": 24335
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9811067287627978e-05,
+      "loss": 0.726,
+      "step": 24340
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9810851121400248e-05,
+      "loss": 0.7169,
+      "step": 24345
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9810634832761258e-05,
+      "loss": 0.7223,
+      "step": 24350
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9810418421713708e-05,
+      "loss": 0.716,
+      "step": 24355
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9810201888260295e-05,
+      "loss": 0.7222,
+      "step": 24360
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9809985232403727e-05,
+      "loss": 0.7243,
+      "step": 24365
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9809768454146697e-05,
+      "loss": 0.7323,
+      "step": 24370
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9809551553491918e-05,
+      "loss": 0.7194,
+      "step": 24375
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.980933453044209e-05,
+      "loss": 0.7041,
+      "step": 24380
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.980911738499993e-05,
+      "loss": 0.7282,
+      "step": 24385
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9808900117168142e-05,
+      "loss": 0.7218,
+      "step": 24390
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9808682726949435e-05,
+      "loss": 0.7241,
+      "step": 24395
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9808465214346525e-05,
+      "loss": 0.7245,
+      "step": 24400
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9808247579362123e-05,
+      "loss": 0.7286,
+      "step": 24405
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9808029821998946e-05,
+      "loss": 0.7271,
+      "step": 24410
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.980781194225971e-05,
+      "loss": 0.7094,
+      "step": 24415
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9807593940147138e-05,
+      "loss": 0.7228,
+      "step": 24420
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9807375815663944e-05,
+      "loss": 0.7295,
+      "step": 24425
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9807157568812848e-05,
+      "loss": 0.7227,
+      "step": 24430
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.980693919959658e-05,
+      "loss": 0.7277,
+      "step": 24435
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9806720708017862e-05,
+      "loss": 0.724,
+      "step": 24440
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.980650209407942e-05,
+      "loss": 0.7264,
+      "step": 24445
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.980628335778398e-05,
+      "loss": 0.7223,
+      "step": 24450
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9806064499134274e-05,
+      "loss": 0.7282,
+      "step": 24455
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.980584551813303e-05,
+      "loss": 0.7335,
+      "step": 24460
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9805626414782985e-05,
+      "loss": 0.7238,
+      "step": 24465
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9805407189086866e-05,
+      "loss": 0.725,
+      "step": 24470
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9805187841047412e-05,
+      "loss": 0.7244,
+      "step": 24475
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.980496837066736e-05,
+      "loss": 0.7163,
+      "step": 24480
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.980474877794945e-05,
+      "loss": 0.7206,
+      "step": 24485
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9804529062896417e-05,
+      "loss": 0.7304,
+      "step": 24490
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9804309225511003e-05,
+      "loss": 0.7234,
+      "step": 24495
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.980408926579596e-05,
+      "loss": 0.7284,
+      "step": 24500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.980386918375402e-05,
+      "loss": 0.7283,
+      "step": 24505
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9803648979387936e-05,
+      "loss": 0.7242,
+      "step": 24510
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9803428652700452e-05,
+      "loss": 0.7252,
+      "step": 24515
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9803208203694323e-05,
+      "loss": 0.714,
+      "step": 24520
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9802987632372296e-05,
+      "loss": 0.7245,
+      "step": 24525
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9802766938737118e-05,
+      "loss": 0.7318,
+      "step": 24530
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9802546122791552e-05,
+      "loss": 0.7227,
+      "step": 24535
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9802325184538345e-05,
+      "loss": 0.7249,
+      "step": 24540
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9802104123980262e-05,
+      "loss": 0.7334,
+      "step": 24545
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.980188294112005e-05,
+      "loss": 0.719,
+      "step": 24550
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9801661635960482e-05,
+      "loss": 0.7042,
+      "step": 24555
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.980144020850431e-05,
+      "loss": 0.7134,
+      "step": 24560
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9801218658754296e-05,
+      "loss": 0.7194,
+      "step": 24565
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.980099698671321e-05,
+      "loss": 0.7281,
+      "step": 24570
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.980077519238382e-05,
+      "loss": 0.7175,
+      "step": 24575
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9800553275768882e-05,
+      "loss": 0.7055,
+      "step": 24580
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9800331236871174e-05,
+      "loss": 0.729,
+      "step": 24585
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9800109075693465e-05,
+      "loss": 0.7287,
+      "step": 24590
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9799886792238523e-05,
+      "loss": 0.7184,
+      "step": 24595
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9799664386509127e-05,
+      "loss": 0.7204,
+      "step": 24600
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.979944185850805e-05,
+      "loss": 0.7258,
+      "step": 24605
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9799219208238063e-05,
+      "loss": 0.7309,
+      "step": 24610
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9798996435701954e-05,
+      "loss": 0.7351,
+      "step": 24615
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9798773540902494e-05,
+      "loss": 0.7255,
+      "step": 24620
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.979855052384247e-05,
+      "loss": 0.7322,
+      "step": 24625
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.979832738452466e-05,
+      "loss": 0.7216,
+      "step": 24630
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9798104122951854e-05,
+      "loss": 0.714,
+      "step": 24635
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9797880739126833e-05,
+      "loss": 0.7124,
+      "step": 24640
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9797657233052382e-05,
+      "loss": 0.7248,
+      "step": 24645
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9797433604731297e-05,
+      "loss": 0.7328,
+      "step": 24650
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.979720985416636e-05,
+      "loss": 0.7216,
+      "step": 24655
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9796985981360368e-05,
+      "loss": 0.7191,
+      "step": 24660
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9796761986316114e-05,
+      "loss": 0.7241,
+      "step": 24665
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.979653786903639e-05,
+      "loss": 0.7298,
+      "step": 24670
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9796313629524e-05,
+      "loss": 0.7132,
+      "step": 24675
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9796089267781733e-05,
+      "loss": 0.7138,
+      "step": 24680
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.979586478381239e-05,
+      "loss": 0.7203,
+      "step": 24685
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9795640177618777e-05,
+      "loss": 0.7336,
+      "step": 24690
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9795415449203688e-05,
+      "loss": 0.7229,
+      "step": 24695
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9795190598569936e-05,
+      "loss": 0.7269,
+      "step": 24700
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9794965625720318e-05,
+      "loss": 0.7182,
+      "step": 24705
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.979474053065765e-05,
+      "loss": 0.7339,
+      "step": 24710
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9794515313384734e-05,
+      "loss": 0.7137,
+      "step": 24715
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9794289973904382e-05,
+      "loss": 0.7248,
+      "step": 24720
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.979406451221941e-05,
+      "loss": 0.7297,
+      "step": 24725
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.979383892833262e-05,
+      "loss": 0.7206,
+      "step": 24730
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9793613222246838e-05,
+      "loss": 0.7334,
+      "step": 24735
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9793387393964873e-05,
+      "loss": 0.7194,
+      "step": 24740
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9793161443489547e-05,
+      "loss": 0.7238,
+      "step": 24745
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9792935370823676e-05,
+      "loss": 0.7201,
+      "step": 24750
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.979270917597008e-05,
+      "loss": 0.7341,
+      "step": 24755
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9792482858931586e-05,
+      "loss": 0.7276,
+      "step": 24760
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.979225641971102e-05,
+      "loss": 0.7258,
+      "step": 24765
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9792029858311193e-05,
+      "loss": 0.7286,
+      "step": 24770
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9791803174734946e-05,
+      "loss": 0.7297,
+      "step": 24775
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9791576368985102e-05,
+      "loss": 0.7122,
+      "step": 24780
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9791349441064495e-05,
+      "loss": 0.7112,
+      "step": 24785
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.979112239097595e-05,
+      "loss": 0.7254,
+      "step": 24790
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9790895218722305e-05,
+      "loss": 0.7348,
+      "step": 24795
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.979066792430639e-05,
+      "loss": 0.7162,
+      "step": 24800
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9790440507731046e-05,
+      "loss": 0.7231,
+      "step": 24805
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9790212968999104e-05,
+      "loss": 0.7204,
+      "step": 24810
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.978998530811341e-05,
+      "loss": 0.7307,
+      "step": 24815
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9789757525076802e-05,
+      "loss": 0.7196,
+      "step": 24820
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.978952961989212e-05,
+      "loss": 0.718,
+      "step": 24825
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.978930159256221e-05,
+      "loss": 0.7262,
+      "step": 24830
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9789073443089914e-05,
+      "loss": 0.7253,
+      "step": 24835
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9788845171478082e-05,
+      "loss": 0.7256,
+      "step": 24840
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9788616777729565e-05,
+      "loss": 0.733,
+      "step": 24845
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9788388261847204e-05,
+      "loss": 0.7234,
+      "step": 24850
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9788159623833856e-05,
+      "loss": 0.7084,
+      "step": 24855
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9787930863692374e-05,
+      "loss": 0.7158,
+      "step": 24860
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.978770198142561e-05,
+      "loss": 0.7288,
+      "step": 24865
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.978747297703642e-05,
+      "loss": 0.7243,
+      "step": 24870
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9787243850527663e-05,
+      "loss": 0.7176,
+      "step": 24875
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.97870146019022e-05,
+      "loss": 0.7227,
+      "step": 24880
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9786785231162885e-05,
+      "loss": 0.7164,
+      "step": 24885
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9786555738312584e-05,
+      "loss": 0.7098,
+      "step": 24890
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9786326123354162e-05,
+      "loss": 0.717,
+      "step": 24895
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.978609638629048e-05,
+      "loss": 0.7234,
+      "step": 24900
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9785866527124405e-05,
+      "loss": 0.7275,
+      "step": 24905
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9785636545858807e-05,
+      "loss": 0.7273,
+      "step": 24910
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9785406442496553e-05,
+      "loss": 0.724,
+      "step": 24915
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9785176217040516e-05,
+      "loss": 0.7179,
+      "step": 24920
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9784945869493568e-05,
+      "loss": 0.7189,
+      "step": 24925
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9784715399858587e-05,
+      "loss": 0.7296,
+      "step": 24930
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.978448480813844e-05,
+      "loss": 0.7116,
+      "step": 24935
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.978425409433601e-05,
+      "loss": 0.7279,
+      "step": 24940
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9784023258454176e-05,
+      "loss": 0.7318,
+      "step": 24945
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9783792300495817e-05,
+      "loss": 0.7295,
+      "step": 24950
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9783561220463813e-05,
+      "loss": 0.7301,
+      "step": 24955
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.978333001836105e-05,
+      "loss": 0.7178,
+      "step": 24960
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9783098694190414e-05,
+      "loss": 0.715,
+      "step": 24965
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9782867247954783e-05,
+      "loss": 0.7102,
+      "step": 24970
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9782635679657053e-05,
+      "loss": 0.7227,
+      "step": 24975
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.978240398930011e-05,
+      "loss": 0.7278,
+      "step": 24980
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9782172176886848e-05,
+      "loss": 0.7247,
+      "step": 24985
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9781940242420156e-05,
+      "loss": 0.7358,
+      "step": 24990
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9781708185902925e-05,
+      "loss": 0.7253,
+      "step": 24995
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9781476007338058e-05,
+      "loss": 0.72,
+      "step": 25000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.978124370672845e-05,
+      "loss": 0.7065,
+      "step": 25005
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.978101128407699e-05,
+      "loss": 0.7275,
+      "step": 25010
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.978077873938659e-05,
+      "loss": 0.7278,
+      "step": 25015
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.978054607266015e-05,
+      "loss": 0.7216,
+      "step": 25020
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9780313283900568e-05,
+      "loss": 0.7178,
+      "step": 25025
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.978008037311075e-05,
+      "loss": 0.7196,
+      "step": 25030
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9779847340293605e-05,
+      "loss": 0.727,
+      "step": 25035
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9779614185452034e-05,
+      "loss": 0.7242,
+      "step": 25040
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9779380908588954e-05,
+      "loss": 0.7126,
+      "step": 25045
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9779147509707267e-05,
+      "loss": 0.7237,
+      "step": 25050
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9778913988809895e-05,
+      "loss": 0.7226,
+      "step": 25055
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9778680345899744e-05,
+      "loss": 0.7212,
+      "step": 25060
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9778446580979735e-05,
+      "loss": 0.7286,
+      "step": 25065
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.977821269405278e-05,
+      "loss": 0.7134,
+      "step": 25070
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.97779786851218e-05,
+      "loss": 0.7286,
+      "step": 25075
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9777744554189713e-05,
+      "loss": 0.7195,
+      "step": 25080
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9777510301259437e-05,
+      "loss": 0.7198,
+      "step": 25085
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9777275926333903e-05,
+      "loss": 0.7191,
+      "step": 25090
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9777041429416034e-05,
+      "loss": 0.729,
+      "step": 25095
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9776806810508748e-05,
+      "loss": 0.7165,
+      "step": 25100
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9776572069614978e-05,
+      "loss": 0.7196,
+      "step": 25105
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9776337206737653e-05,
+      "loss": 0.7269,
+      "step": 25110
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9776102221879702e-05,
+      "loss": 0.7413,
+      "step": 25115
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9775867115044057e-05,
+      "loss": 0.7156,
+      "step": 25120
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9775631886233655e-05,
+      "loss": 0.7235,
+      "step": 25125
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9775396535451423e-05,
+      "loss": 0.7209,
+      "step": 25130
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9775161062700306e-05,
+      "loss": 0.7263,
+      "step": 25135
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9774925467983238e-05,
+      "loss": 0.7235,
+      "step": 25140
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.977468975130316e-05,
+      "loss": 0.7205,
+      "step": 25145
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.977445391266301e-05,
+      "loss": 0.7076,
+      "step": 25150
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9774217952065735e-05,
+      "loss": 0.7106,
+      "step": 25155
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9773981869514274e-05,
+      "loss": 0.7171,
+      "step": 25160
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9773745665011577e-05,
+      "loss": 0.7249,
+      "step": 25165
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.977350933856059e-05,
+      "loss": 0.7157,
+      "step": 25170
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9773272890164264e-05,
+      "loss": 0.7215,
+      "step": 25175
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9773036319825543e-05,
+      "loss": 0.7205,
+      "step": 25180
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9772799627547383e-05,
+      "loss": 0.7233,
+      "step": 25185
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9772562813332736e-05,
+      "loss": 0.7384,
+      "step": 25190
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9772325877184558e-05,
+      "loss": 0.7239,
+      "step": 25195
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9772088819105804e-05,
+      "loss": 0.714,
+      "step": 25200
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9771851639099432e-05,
+      "loss": 0.7327,
+      "step": 25205
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9771614337168404e-05,
+      "loss": 0.7293,
+      "step": 25210
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9771376913315677e-05,
+      "loss": 0.7346,
+      "step": 25215
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9771139367544212e-05,
+      "loss": 0.7217,
+      "step": 25220
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9770901699856978e-05,
+      "loss": 0.721,
+      "step": 25225
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.977066391025694e-05,
+      "loss": 0.7212,
+      "step": 25230
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.977042599874706e-05,
+      "loss": 0.7216,
+      "step": 25235
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.977018796533031e-05,
+      "loss": 0.713,
+      "step": 25240
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.976994981000966e-05,
+      "loss": 0.7287,
+      "step": 25245
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9769711532788083e-05,
+      "loss": 0.7189,
+      "step": 25250
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.976947313366855e-05,
+      "loss": 0.7122,
+      "step": 25255
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.976923461265403e-05,
+      "loss": 0.7238,
+      "step": 25260
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9768995969747513e-05,
+      "loss": 0.7196,
+      "step": 25265
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.976875720495196e-05,
+      "loss": 0.7225,
+      "step": 25270
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.976851831827036e-05,
+      "loss": 0.7267,
+      "step": 25275
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.97682793097057e-05,
+      "loss": 0.7299,
+      "step": 25280
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9768040179260944e-05,
+      "loss": 0.7266,
+      "step": 25285
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.976780092693909e-05,
+      "loss": 0.7055,
+      "step": 25290
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9767561552743117e-05,
+      "loss": 0.7121,
+      "step": 25295
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9767322056676018e-05,
+      "loss": 0.7269,
+      "step": 25300
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9767082438740776e-05,
+      "loss": 0.7321,
+      "step": 25305
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9766842698940377e-05,
+      "loss": 0.7167,
+      "step": 25310
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9766602837277818e-05,
+      "loss": 0.7099,
+      "step": 25315
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9766362853756092e-05,
+      "loss": 0.7216,
+      "step": 25320
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.976612274837819e-05,
+      "loss": 0.7357,
+      "step": 25325
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9765882521147115e-05,
+      "loss": 0.7271,
+      "step": 25330
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9765642172065857e-05,
+      "loss": 0.718,
+      "step": 25335
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9765401701137416e-05,
+      "loss": 0.7129,
+      "step": 25340
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.976516110836479e-05,
+      "loss": 0.7196,
+      "step": 25345
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.976492039375099e-05,
+      "loss": 0.7189,
+      "step": 25350
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9764679557299005e-05,
+      "loss": 0.7248,
+      "step": 25355
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.976443859901186e-05,
+      "loss": 0.7178,
+      "step": 25360
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9764197518892544e-05,
+      "loss": 0.7318,
+      "step": 25365
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9763956316944068e-05,
+      "loss": 0.7097,
+      "step": 25370
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.976371499316945e-05,
+      "loss": 0.72,
+      "step": 25375
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9763473547571693e-05,
+      "loss": 0.7177,
+      "step": 25380
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9763231980153813e-05,
+      "loss": 0.7302,
+      "step": 25385
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9762990290918825e-05,
+      "loss": 0.7062,
+      "step": 25390
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.976274847986974e-05,
+      "loss": 0.7214,
+      "step": 25395
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.976250654700958e-05,
+      "loss": 0.7086,
+      "step": 25400
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.976226449234136e-05,
+      "loss": 0.7134,
+      "step": 25405
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.97620223158681e-05,
+      "loss": 0.7186,
+      "step": 25410
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9761780017592827e-05,
+      "loss": 0.7185,
+      "step": 25415
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9761537597518557e-05,
+      "loss": 0.7308,
+      "step": 25420
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9761295055648323e-05,
+      "loss": 0.7089,
+      "step": 25425
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9761052391985145e-05,
+      "loss": 0.7085,
+      "step": 25430
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9760809606532054e-05,
+      "loss": 0.7211,
+      "step": 25435
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.976056669929207e-05,
+      "loss": 0.7112,
+      "step": 25440
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9760323670268238e-05,
+      "loss": 0.725,
+      "step": 25445
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.976008051946358e-05,
+      "loss": 0.7275,
+      "step": 25450
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9759837246881134e-05,
+      "loss": 0.7163,
+      "step": 25455
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9759593852523937e-05,
+      "loss": 0.7136,
+      "step": 25460
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.975935033639502e-05,
+      "loss": 0.7054,
+      "step": 25465
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.975910669849743e-05,
+      "loss": 0.7163,
+      "step": 25470
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.97588629388342e-05,
+      "loss": 0.718,
+      "step": 25475
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9758619057408372e-05,
+      "loss": 0.7224,
+      "step": 25480
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9758375054222992e-05,
+      "loss": 0.7172,
+      "step": 25485
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9758130929281102e-05,
+      "loss": 0.7175,
+      "step": 25490
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9757886682585744e-05,
+      "loss": 0.7251,
+      "step": 25495
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9757642314139977e-05,
+      "loss": 0.7241,
+      "step": 25500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.975739782394684e-05,
+      "loss": 0.7185,
+      "step": 25505
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9757153212009388e-05,
+      "loss": 0.7185,
+      "step": 25510
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.975690847833067e-05,
+      "loss": 0.7141,
+      "step": 25515
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9756663622913744e-05,
+      "loss": 0.7332,
+      "step": 25520
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.975641864576166e-05,
+      "loss": 0.7026,
+      "step": 25525
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.975617354687748e-05,
+      "loss": 0.7253,
+      "step": 25530
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9755928326264255e-05,
+      "loss": 0.7169,
+      "step": 25535
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9755682983925054e-05,
+      "loss": 0.7272,
+      "step": 25540
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.975543751986293e-05,
+      "loss": 0.7206,
+      "step": 25545
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.975519193408095e-05,
+      "loss": 0.7266,
+      "step": 25550
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9754946226582176e-05,
+      "loss": 0.7251,
+      "step": 25555
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9754700397369674e-05,
+      "loss": 0.7136,
+      "step": 25560
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9754454446446512e-05,
+      "loss": 0.7093,
+      "step": 25565
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9754208373815758e-05,
+      "loss": 0.7218,
+      "step": 25570
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9753962179480485e-05,
+      "loss": 0.7208,
+      "step": 25575
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.975371586344376e-05,
+      "loss": 0.7129,
+      "step": 25580
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.975346942570866e-05,
+      "loss": 0.7278,
+      "step": 25585
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9753222866278264e-05,
+      "loss": 0.7151,
+      "step": 25590
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9752976185155635e-05,
+      "loss": 0.7098,
+      "step": 25595
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9752729382343866e-05,
+      "loss": 0.7351,
+      "step": 25600
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9752482457846026e-05,
+      "loss": 0.6999,
+      "step": 25605
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9752235411665198e-05,
+      "loss": 0.7226,
+      "step": 25610
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.975198824380447e-05,
+      "loss": 0.7221,
+      "step": 25615
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.975174095426692e-05,
+      "loss": 0.706,
+      "step": 25620
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9751493543055634e-05,
+      "loss": 0.7156,
+      "step": 25625
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.97512460101737e-05,
+      "loss": 0.7167,
+      "step": 25630
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.975099835562421e-05,
+      "loss": 0.7097,
+      "step": 25635
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.975075057941025e-05,
+      "loss": 0.7305,
+      "step": 25640
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.975050268153491e-05,
+      "loss": 0.7228,
+      "step": 25645
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9750254662001284e-05,
+      "loss": 0.713,
+      "step": 25650
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.975000652081247e-05,
+      "loss": 0.7103,
+      "step": 25655
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9749758257971562e-05,
+      "loss": 0.734,
+      "step": 25660
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.974950987348166e-05,
+      "loss": 0.7157,
+      "step": 25665
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9749261367345857e-05,
+      "loss": 0.7144,
+      "step": 25670
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9749012739567258e-05,
+      "loss": 0.7229,
+      "step": 25675
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9748763990148964e-05,
+      "loss": 0.7178,
+      "step": 25680
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9748515119094083e-05,
+      "loss": 0.7212,
+      "step": 25685
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9748266126405714e-05,
+      "loss": 0.7145,
+      "step": 25690
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9748017012086965e-05,
+      "loss": 0.7102,
+      "step": 25695
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.974776777614095e-05,
+      "loss": 0.7259,
+      "step": 25700
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.974751841857077e-05,
+      "loss": 0.7203,
+      "step": 25705
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.974726893937954e-05,
+      "loss": 0.7126,
+      "step": 25710
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9747019338570373e-05,
+      "loss": 0.7132,
+      "step": 25715
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9746769616146383e-05,
+      "loss": 0.7166,
+      "step": 25720
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9746519772110688e-05,
+      "loss": 0.7243,
+      "step": 25725
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9746269806466404e-05,
+      "loss": 0.7168,
+      "step": 25730
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.974601971921665e-05,
+      "loss": 0.7238,
+      "step": 25735
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9745769510364546e-05,
+      "loss": 0.7247,
+      "step": 25740
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9745519179913214e-05,
+      "loss": 0.7097,
+      "step": 25745
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9745268727865774e-05,
+      "loss": 0.718,
+      "step": 25750
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9745018154225354e-05,
+      "loss": 0.7193,
+      "step": 25755
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9744767458995083e-05,
+      "loss": 0.7072,
+      "step": 25760
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9744516642178087e-05,
+      "loss": 0.7104,
+      "step": 25765
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.974426570377749e-05,
+      "loss": 0.7034,
+      "step": 25770
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9744014643796435e-05,
+      "loss": 0.7174,
+      "step": 25775
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9743763462238044e-05,
+      "loss": 0.7205,
+      "step": 25780
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9743512159105454e-05,
+      "loss": 0.7052,
+      "step": 25785
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.97432607344018e-05,
+      "loss": 0.7202,
+      "step": 25790
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.974300918813022e-05,
+      "loss": 0.7215,
+      "step": 25795
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9742757520293856e-05,
+      "loss": 0.7116,
+      "step": 25800
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9742505730895843e-05,
+      "loss": 0.7114,
+      "step": 25805
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.974225381993932e-05,
+      "loss": 0.7218,
+      "step": 25810
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9742001787427443e-05,
+      "loss": 0.7119,
+      "step": 25815
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9741749633363342e-05,
+      "loss": 0.7119,
+      "step": 25820
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.974149735775017e-05,
+      "loss": 0.7101,
+      "step": 25825
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9741244960591074e-05,
+      "loss": 0.7113,
+      "step": 25830
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9740992441889205e-05,
+      "loss": 0.7234,
+      "step": 25835
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.974073980164771e-05,
+      "loss": 0.7126,
+      "step": 25840
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9740487039869746e-05,
+      "loss": 0.708,
+      "step": 25845
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9740234156558463e-05,
+      "loss": 0.7133,
+      "step": 25850
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9739981151717017e-05,
+      "loss": 0.7053,
+      "step": 25855
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9739728025348562e-05,
+      "loss": 0.7104,
+      "step": 25860
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9739474777456257e-05,
+      "loss": 0.6985,
+      "step": 25865
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.973922140804327e-05,
+      "loss": 0.7128,
+      "step": 25870
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9738967917112752e-05,
+      "loss": 0.723,
+      "step": 25875
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9738714304667874e-05,
+      "loss": 0.7216,
+      "step": 25880
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9738460570711794e-05,
+      "loss": 0.7094,
+      "step": 25885
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9738206715247682e-05,
+      "loss": 0.7269,
+      "step": 25890
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.97379527382787e-05,
+      "loss": 0.711,
+      "step": 25895
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9737698639808024e-05,
+      "loss": 0.7184,
+      "step": 25900
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.973744441983882e-05,
+      "loss": 0.7158,
+      "step": 25905
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.973719007837426e-05,
+      "loss": 0.7135,
+      "step": 25910
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9736935615417517e-05,
+      "loss": 0.7241,
+      "step": 25915
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.973668103097177e-05,
+      "loss": 0.7166,
+      "step": 25920
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9736426325040194e-05,
+      "loss": 0.7133,
+      "step": 25925
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9736171497625968e-05,
+      "loss": 0.7173,
+      "step": 25930
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9735916548732262e-05,
+      "loss": 0.7129,
+      "step": 25935
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.973566147836227e-05,
+      "loss": 0.7097,
+      "step": 25940
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9735406286519168e-05,
+      "loss": 0.7121,
+      "step": 25945
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.973515097320614e-05,
+      "loss": 0.712,
+      "step": 25950
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9734895538426372e-05,
+      "loss": 0.7166,
+      "step": 25955
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9734639982183053e-05,
+      "loss": 0.7173,
+      "step": 25960
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9734384304479367e-05,
+      "loss": 0.7183,
+      "step": 25965
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9734128505318512e-05,
+      "loss": 0.7229,
+      "step": 25970
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9733872584703673e-05,
+      "loss": 0.7094,
+      "step": 25975
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9733616542638045e-05,
+      "loss": 0.7229,
+      "step": 25980
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9733360379124822e-05,
+      "loss": 0.7059,
+      "step": 25985
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9733104094167204e-05,
+      "loss": 0.7267,
+      "step": 25990
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9732847687768385e-05,
+      "loss": 0.715,
+      "step": 25995
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9732591159931564e-05,
+      "loss": 0.7281,
+      "step": 26000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9732334510659942e-05,
+      "loss": 0.7234,
+      "step": 26005
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9732077739956725e-05,
+      "loss": 0.7227,
+      "step": 26010
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.973182084782511e-05,
+      "loss": 0.713,
+      "step": 26015
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9731563834268308e-05,
+      "loss": 0.7061,
+      "step": 26020
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9731306699289522e-05,
+      "loss": 0.7169,
+      "step": 26025
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9731049442891965e-05,
+      "loss": 0.7051,
+      "step": 26030
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.973079206507884e-05,
+      "loss": 0.7237,
+      "step": 26035
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9730534565853364e-05,
+      "loss": 0.7268,
+      "step": 26040
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.973027694521875e-05,
+      "loss": 0.7253,
+      "step": 26045
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.973001920317821e-05,
+      "loss": 0.7119,
+      "step": 26050
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.972976133973496e-05,
+      "loss": 0.7248,
+      "step": 26055
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9729503354892215e-05,
+      "loss": 0.707,
+      "step": 26060
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.97292452486532e-05,
+      "loss": 0.7145,
+      "step": 26065
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.972898702102113e-05,
+      "loss": 0.691,
+      "step": 26070
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.972872867199923e-05,
+      "loss": 0.7114,
+      "step": 26075
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9728470201590725e-05,
+      "loss": 0.7066,
+      "step": 26080
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9728211609798835e-05,
+      "loss": 0.706,
+      "step": 26085
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9727952896626788e-05,
+      "loss": 0.7112,
+      "step": 26090
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9727694062077813e-05,
+      "loss": 0.7131,
+      "step": 26095
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.972743510615514e-05,
+      "loss": 0.7141,
+      "step": 26100
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9727176028862e-05,
+      "loss": 0.7122,
+      "step": 26105
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9726916830201626e-05,
+      "loss": 0.7206,
+      "step": 26110
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9726657510177246e-05,
+      "loss": 0.7126,
+      "step": 26115
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.972639806879211e-05,
+      "loss": 0.7035,
+      "step": 26120
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9726138506049438e-05,
+      "loss": 0.7141,
+      "step": 26125
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9725878821952478e-05,
+      "loss": 0.7073,
+      "step": 26130
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9725619016504466e-05,
+      "loss": 0.7022,
+      "step": 26135
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.972535908970865e-05,
+      "loss": 0.7232,
+      "step": 26140
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.972509904156827e-05,
+      "loss": 0.7103,
+      "step": 26145
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9724838872086568e-05,
+      "loss": 0.7149,
+      "step": 26150
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9724578581266793e-05,
+      "loss": 0.7188,
+      "step": 26155
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9724318169112187e-05,
+      "loss": 0.7136,
+      "step": 26160
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9724057635626008e-05,
+      "loss": 0.7275,
+      "step": 26165
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9723796980811503e-05,
+      "loss": 0.7032,
+      "step": 26170
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9723536204671922e-05,
+      "loss": 0.7073,
+      "step": 26175
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.972327530721052e-05,
+      "loss": 0.7227,
+      "step": 26180
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9723014288430557e-05,
+      "loss": 0.7072,
+      "step": 26185
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.972275314833528e-05,
+      "loss": 0.7012,
+      "step": 26190
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.972249188692796e-05,
+      "loss": 0.7119,
+      "step": 26195
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9722230504211843e-05,
+      "loss": 0.7167,
+      "step": 26200
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.97219690001902e-05,
+      "loss": 0.7065,
+      "step": 26205
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.972170737486629e-05,
+      "loss": 0.7175,
+      "step": 26210
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9721445628243375e-05,
+      "loss": 0.7083,
+      "step": 26215
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.972118376032473e-05,
+      "loss": 0.7112,
+      "step": 26220
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.972092177111361e-05,
+      "loss": 0.7144,
+      "step": 26225
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9720659660613298e-05,
+      "loss": 0.7059,
+      "step": 26230
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9720397428827052e-05,
+      "loss": 0.7114,
+      "step": 26235
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.972013507575815e-05,
+      "loss": 0.7076,
+      "step": 26240
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9719872601409863e-05,
+      "loss": 0.716,
+      "step": 26245
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9719610005785466e-05,
+      "loss": 0.7034,
+      "step": 26250
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.971934728888823e-05,
+      "loss": 0.719,
+      "step": 26255
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.971908445072145e-05,
+      "loss": 0.7179,
+      "step": 26260
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9718821491288392e-05,
+      "loss": 0.7065,
+      "step": 26265
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9718558410592336e-05,
+      "loss": 0.7042,
+      "step": 26270
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.971829520863657e-05,
+      "loss": 0.7204,
+      "step": 26275
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9718031885424378e-05,
+      "loss": 0.7159,
+      "step": 26280
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9717768440959042e-05,
+      "loss": 0.7124,
+      "step": 26285
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.971750487524385e-05,
+      "loss": 0.7193,
+      "step": 26290
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9717241188282092e-05,
+      "loss": 0.7173,
+      "step": 26295
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9716977380077058e-05,
+      "loss": 0.7115,
+      "step": 26300
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.971671345063204e-05,
+      "loss": 0.7174,
+      "step": 26305
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9716449399950327e-05,
+      "loss": 0.7161,
+      "step": 26310
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.971618522803522e-05,
+      "loss": 0.7121,
+      "step": 26315
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9715920934890007e-05,
+      "loss": 0.7117,
+      "step": 26320
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9715656520517993e-05,
+      "loss": 0.727,
+      "step": 26325
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9715391984922474e-05,
+      "loss": 0.7156,
+      "step": 26330
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.971512732810675e-05,
+      "loss": 0.7107,
+      "step": 26335
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.971486255007413e-05,
+      "loss": 0.7092,
+      "step": 26340
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9714597650827903e-05,
+      "loss": 0.7106,
+      "step": 26345
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9714332630371385e-05,
+      "loss": 0.7192,
+      "step": 26350
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9714067488707888e-05,
+      "loss": 0.7158,
+      "step": 26355
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9713802225840705e-05,
+      "loss": 0.7126,
+      "step": 26360
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9713536841773156e-05,
+      "loss": 0.703,
+      "step": 26365
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9713271336508552e-05,
+      "loss": 0.7093,
+      "step": 26370
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9713005710050203e-05,
+      "loss": 0.7107,
+      "step": 26375
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9712739962401422e-05,
+      "loss": 0.7114,
+      "step": 26380
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9712474093565526e-05,
+      "loss": 0.7065,
+      "step": 26385
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9712208103545837e-05,
+      "loss": 0.7166,
+      "step": 26390
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9711941992345666e-05,
+      "loss": 0.7023,
+      "step": 26395
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.971167575996834e-05,
+      "loss": 0.7154,
+      "step": 26400
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9711409406417176e-05,
+      "loss": 0.7117,
+      "step": 26405
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.97111429316955e-05,
+      "loss": 0.7176,
+      "step": 26410
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9710876335806634e-05,
+      "loss": 0.7235,
+      "step": 26415
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9710609618753907e-05,
+      "loss": 0.7155,
+      "step": 26420
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.971034278054065e-05,
+      "loss": 0.7175,
+      "step": 26425
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9710075821170183e-05,
+      "loss": 0.7215,
+      "step": 26430
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9709808740645847e-05,
+      "loss": 0.7105,
+      "step": 26435
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9709541538970968e-05,
+      "loss": 0.7066,
+      "step": 26440
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9709274216148883e-05,
+      "loss": 0.7134,
+      "step": 26445
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9709006772182926e-05,
+      "loss": 0.717,
+      "step": 26450
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.970873920707643e-05,
+      "loss": 0.7116,
+      "step": 26455
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9708471520832743e-05,
+      "loss": 0.7175,
+      "step": 26460
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9708203713455198e-05,
+      "loss": 0.7187,
+      "step": 26465
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9707935784947137e-05,
+      "loss": 0.7104,
+      "step": 26470
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9707667735311908e-05,
+      "loss": 0.7172,
+      "step": 26475
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9707399564552844e-05,
+      "loss": 0.7195,
+      "step": 26480
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9707131272673307e-05,
+      "loss": 0.7217,
+      "step": 26485
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.970686285967663e-05,
+      "loss": 0.7226,
+      "step": 26490
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.970659432556617e-05,
+      "loss": 0.7151,
+      "step": 26495
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9706325670345276e-05,
+      "loss": 0.7256,
+      "step": 26500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.97060568940173e-05,
+      "loss": 0.7089,
+      "step": 26505
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9705787996585594e-05,
+      "loss": 0.7106,
+      "step": 26510
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9705518978053513e-05,
+      "loss": 0.7075,
+      "step": 26515
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9705249838424416e-05,
+      "loss": 0.71,
+      "step": 26520
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.970498057770166e-05,
+      "loss": 0.7218,
+      "step": 26525
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9704711195888605e-05,
+      "loss": 0.7154,
+      "step": 26530
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.970444169298861e-05,
+      "loss": 0.7168,
+      "step": 26535
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9704172069005044e-05,
+      "loss": 0.7129,
+      "step": 26540
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.970390232394126e-05,
+      "loss": 0.7129,
+      "step": 26545
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9703632457800633e-05,
+      "loss": 0.6966,
+      "step": 26550
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.970336247058653e-05,
+      "loss": 0.7261,
+      "step": 26555
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9703092362302313e-05,
+      "loss": 0.7261,
+      "step": 26560
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9702822132951356e-05,
+      "loss": 0.7268,
+      "step": 26565
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9702551782537032e-05,
+      "loss": 0.7175,
+      "step": 26570
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9702281311062713e-05,
+      "loss": 0.721,
+      "step": 26575
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9702010718531774e-05,
+      "loss": 0.7121,
+      "step": 26580
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.970174000494759e-05,
+      "loss": 0.7119,
+      "step": 26585
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9701469170313538e-05,
+      "loss": 0.705,
+      "step": 26590
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9701198214633e-05,
+      "loss": 0.7161,
+      "step": 26595
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9700927137909358e-05,
+      "loss": 0.7082,
+      "step": 26600
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.970065594014599e-05,
+      "loss": 0.6984,
+      "step": 26605
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9700384621346282e-05,
+      "loss": 0.7177,
+      "step": 26610
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9700113181513618e-05,
+      "loss": 0.7184,
+      "step": 26615
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9699841620651384e-05,
+      "loss": 0.7018,
+      "step": 26620
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9699569938762975e-05,
+      "loss": 0.7195,
+      "step": 26625
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9699298135851773e-05,
+      "loss": 0.7042,
+      "step": 26630
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9699026211921172e-05,
+      "loss": 0.7199,
+      "step": 26635
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9698754166974564e-05,
+      "loss": 0.7159,
+      "step": 26640
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9698482001015344e-05,
+      "loss": 0.7176,
+      "step": 26645
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.969820971404691e-05,
+      "loss": 0.7268,
+      "step": 26650
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9697937306072656e-05,
+      "loss": 0.715,
+      "step": 26655
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9697664777095982e-05,
+      "loss": 0.7077,
+      "step": 26660
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9697392127120293e-05,
+      "loss": 0.7169,
+      "step": 26665
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9697119356148982e-05,
+      "loss": 0.702,
+      "step": 26670
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.969684646418546e-05,
+      "loss": 0.702,
+      "step": 26675
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9696573451233126e-05,
+      "loss": 0.706,
+      "step": 26680
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.969630031729539e-05,
+      "loss": 0.7262,
+      "step": 26685
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.969602706237566e-05,
+      "loss": 0.7122,
+      "step": 26690
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9695753686477347e-05,
+      "loss": 0.7188,
+      "step": 26695
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9695480189603856e-05,
+      "loss": 0.7105,
+      "step": 26700
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9695206571758604e-05,
+      "loss": 0.6986,
+      "step": 26705
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9694932832945005e-05,
+      "loss": 0.7104,
+      "step": 26710
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9694658973166468e-05,
+      "loss": 0.6998,
+      "step": 26715
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9694384992426422e-05,
+      "loss": 0.7084,
+      "step": 26720
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.969411089072828e-05,
+      "loss": 0.7121,
+      "step": 26725
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9693836668075458e-05,
+      "loss": 0.7208,
+      "step": 26730
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.969356232447138e-05,
+      "loss": 0.7044,
+      "step": 26735
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.969328785991947e-05,
+      "loss": 0.7174,
+      "step": 26740
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9693013274423155e-05,
+      "loss": 0.7162,
+      "step": 26745
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9692738567985853e-05,
+      "loss": 0.7106,
+      "step": 26750
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9692463740611e-05,
+      "loss": 0.7062,
+      "step": 26755
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.969218879230202e-05,
+      "loss": 0.7172,
+      "step": 26760
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.969191372306235e-05,
+      "loss": 0.706,
+      "step": 26765
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.969163853289541e-05,
+      "loss": 0.7155,
+      "step": 26770
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9691363221804645e-05,
+      "loss": 0.711,
+      "step": 26775
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9691087789793486e-05,
+      "loss": 0.7139,
+      "step": 26780
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.969081223686537e-05,
+      "loss": 0.709,
+      "step": 26785
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.969053656302373e-05,
+      "loss": 0.7077,
+      "step": 26790
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.969026076827202e-05,
+      "loss": 0.7116,
+      "step": 26795
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9689984852613664e-05,
+      "loss": 0.7099,
+      "step": 26800
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9689708816052115e-05,
+      "loss": 0.7129,
+      "step": 26805
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9689432658590813e-05,
+      "loss": 0.7045,
+      "step": 26810
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9689156380233206e-05,
+      "loss": 0.713,
+      "step": 26815
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9688879980982737e-05,
+      "loss": 0.699,
+      "step": 26820
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9688603460842862e-05,
+      "loss": 0.7026,
+      "step": 26825
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9688326819817023e-05,
+      "loss": 0.7229,
+      "step": 26830
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9688050057908677e-05,
+      "loss": 0.7094,
+      "step": 26835
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9687773175121278e-05,
+      "loss": 0.7002,
+      "step": 26840
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9687496171458276e-05,
+      "loss": 0.7107,
+      "step": 26845
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.968721904692313e-05,
+      "loss": 0.7123,
+      "step": 26850
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.96869418015193e-05,
+      "loss": 0.7008,
+      "step": 26855
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.968666443525024e-05,
+      "loss": 0.7089,
+      "step": 26860
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9686386948119414e-05,
+      "loss": 0.7024,
+      "step": 26865
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9686109340130284e-05,
+      "loss": 0.7134,
+      "step": 26870
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9685831611286312e-05,
+      "loss": 0.7107,
+      "step": 26875
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9685553761590967e-05,
+      "loss": 0.6984,
+      "step": 26880
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9685275791047713e-05,
+      "loss": 0.7097,
+      "step": 26885
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9684997699660016e-05,
+      "loss": 0.7108,
+      "step": 26890
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.968471948743135e-05,
+      "loss": 0.6984,
+      "step": 26895
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9684441154365185e-05,
+      "loss": 0.711,
+      "step": 26900
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9684162700464995e-05,
+      "loss": 0.7113,
+      "step": 26905
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.968388412573425e-05,
+      "loss": 0.7174,
+      "step": 26910
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9683605430176432e-05,
+      "loss": 0.7062,
+      "step": 26915
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.968332661379501e-05,
+      "loss": 0.707,
+      "step": 26920
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9683047676593475e-05,
+      "loss": 0.7147,
+      "step": 26925
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9682768618575296e-05,
+      "loss": 0.7005,
+      "step": 26930
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9682489439743958e-05,
+      "loss": 0.7151,
+      "step": 26935
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.968221014010295e-05,
+      "loss": 0.7122,
+      "step": 26940
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9681930719655747e-05,
+      "loss": 0.7102,
+      "step": 26945
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9681651178405844e-05,
+      "loss": 0.7184,
+      "step": 26950
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.968137151635672e-05,
+      "loss": 0.7203,
+      "step": 26955
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9681091733511876e-05,
+      "loss": 0.6984,
+      "step": 26960
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9680811829874794e-05,
+      "loss": 0.7209,
+      "step": 26965
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9680531805448973e-05,
+      "loss": 0.7011,
+      "step": 26970
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9680251660237898e-05,
+      "loss": 0.7138,
+      "step": 26975
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9679971394245074e-05,
+      "loss": 0.7199,
+      "step": 26980
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9679691007473988e-05,
+      "loss": 0.7129,
+      "step": 26985
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9679410499928147e-05,
+      "loss": 0.7058,
+      "step": 26990
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9679129871611047e-05,
+      "loss": 0.7185,
+      "step": 26995
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.967884912252619e-05,
+      "loss": 0.7158,
+      "step": 27000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9678568252677078e-05,
+      "loss": 0.7097,
+      "step": 27005
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.967828726206722e-05,
+      "loss": 0.7174,
+      "step": 27010
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9678006150700114e-05,
+      "loss": 0.6988,
+      "step": 27015
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9677724918579274e-05,
+      "loss": 0.7121,
+      "step": 27020
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9677443565708206e-05,
+      "loss": 0.7031,
+      "step": 27025
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9677162092090423e-05,
+      "loss": 0.7124,
+      "step": 27030
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9676880497729435e-05,
+      "loss": 0.7128,
+      "step": 27035
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9676598782628755e-05,
+      "loss": 0.7186,
+      "step": 27040
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9676316946791897e-05,
+      "loss": 0.6943,
+      "step": 27045
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9676034990222382e-05,
+      "loss": 0.7139,
+      "step": 27050
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9675752912923726e-05,
+      "loss": 0.6994,
+      "step": 27055
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.967547071489945e-05,
+      "loss": 0.7103,
+      "step": 27060
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.967518839615307e-05,
+      "loss": 0.7093,
+      "step": 27065
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9674905956688112e-05,
+      "loss": 0.7054,
+      "step": 27070
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9674623396508095e-05,
+      "loss": 0.7133,
+      "step": 27075
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9674340715616553e-05,
+      "loss": 0.7064,
+      "step": 27080
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.967405791401701e-05,
+      "loss": 0.7034,
+      "step": 27085
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9673774991712992e-05,
+      "loss": 0.707,
+      "step": 27090
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9673491948708034e-05,
+      "loss": 0.7021,
+      "step": 27095
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9673208785005658e-05,
+      "loss": 0.6885,
+      "step": 27100
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.967292550060941e-05,
+      "loss": 0.7111,
+      "step": 27105
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9672642095522813e-05,
+      "loss": 0.7142,
+      "step": 27110
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.967235856974941e-05,
+      "loss": 0.719,
+      "step": 27115
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9672074923292736e-05,
+      "loss": 0.7043,
+      "step": 27120
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.967179115615633e-05,
+      "loss": 0.6975,
+      "step": 27125
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9671507268343736e-05,
+      "loss": 0.7082,
+      "step": 27130
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.967122325985849e-05,
+      "loss": 0.7063,
+      "step": 27135
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9670939130704138e-05,
+      "loss": 0.7158,
+      "step": 27140
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.967065488088423e-05,
+      "loss": 0.7102,
+      "step": 27145
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9670370510402306e-05,
+      "loss": 0.7135,
+      "step": 27150
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9670086019261917e-05,
+      "loss": 0.7072,
+      "step": 27155
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9669801407466614e-05,
+      "loss": 0.7049,
+      "step": 27160
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9669516675019945e-05,
+      "loss": 0.7077,
+      "step": 27165
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9669231821925464e-05,
+      "loss": 0.7149,
+      "step": 27170
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9668946848186728e-05,
+      "loss": 0.7052,
+      "step": 27175
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.966866175380729e-05,
+      "loss": 0.7189,
+      "step": 27180
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9668376538790706e-05,
+      "loss": 0.7084,
+      "step": 27185
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9668091203140535e-05,
+      "loss": 0.7139,
+      "step": 27190
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.966780574686034e-05,
+      "loss": 0.7128,
+      "step": 27195
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9667520169953677e-05,
+      "loss": 0.7048,
+      "step": 27200
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9667234472424115e-05,
+      "loss": 0.6966,
+      "step": 27205
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.966694865427522e-05,
+      "loss": 0.7087,
+      "step": 27210
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.966666271551055e-05,
+      "loss": 0.7064,
+      "step": 27215
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9666376656133676e-05,
+      "loss": 0.7066,
+      "step": 27220
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.966609047614817e-05,
+      "loss": 0.7184,
+      "step": 27225
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9665804175557605e-05,
+      "loss": 0.7148,
+      "step": 27230
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9665517754365547e-05,
+      "loss": 0.7078,
+      "step": 27235
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9665231212575574e-05,
+      "loss": 0.6987,
+      "step": 27240
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.966494455019126e-05,
+      "loss": 0.7102,
+      "step": 27245
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9664657767216176e-05,
+      "loss": 0.7165,
+      "step": 27250
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.966437086365391e-05,
+      "loss": 0.7104,
+      "step": 27255
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9664083839508036e-05,
+      "loss": 0.7035,
+      "step": 27260
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9663796694782138e-05,
+      "loss": 0.7124,
+      "step": 27265
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9663509429479796e-05,
+      "loss": 0.7059,
+      "step": 27270
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9663222043604594e-05,
+      "loss": 0.7074,
+      "step": 27275
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.966293453716012e-05,
+      "loss": 0.7124,
+      "step": 27280
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.966264691014996e-05,
+      "loss": 0.7041,
+      "step": 27285
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9662359162577706e-05,
+      "loss": 0.7022,
+      "step": 27290
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.966207129444694e-05,
+      "loss": 0.7116,
+      "step": 27295
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9661783305761264e-05,
+      "loss": 0.7123,
+      "step": 27300
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9661495196524266e-05,
+      "loss": 0.7174,
+      "step": 27305
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.966120696673954e-05,
+      "loss": 0.7067,
+      "step": 27310
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9660918616410686e-05,
+      "loss": 0.7039,
+      "step": 27315
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9660630145541292e-05,
+      "loss": 0.7077,
+      "step": 27320
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9660341554134972e-05,
+      "loss": 0.7103,
+      "step": 27325
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9660052842195315e-05,
+      "loss": 0.7184,
+      "step": 27330
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.965976400972593e-05,
+      "loss": 0.7169,
+      "step": 27335
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.965947505673042e-05,
+      "loss": 0.7108,
+      "step": 27340
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9659185983212388e-05,
+      "loss": 0.712,
+      "step": 27345
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.965889678917544e-05,
+      "loss": 0.704,
+      "step": 27350
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9658607474623187e-05,
+      "loss": 0.7111,
+      "step": 27355
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9658318039559238e-05,
+      "loss": 0.7024,
+      "step": 27360
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9658028483987203e-05,
+      "loss": 0.7058,
+      "step": 27365
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9657738807910695e-05,
+      "loss": 0.7115,
+      "step": 27370
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9657449011333328e-05,
+      "loss": 0.6965,
+      "step": 27375
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9657159094258724e-05,
+      "loss": 0.6931,
+      "step": 27380
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9656869056690492e-05,
+      "loss": 0.7142,
+      "step": 27385
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9656578898632257e-05,
+      "loss": 0.712,
+      "step": 27390
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9656288620087634e-05,
+      "loss": 0.6977,
+      "step": 27395
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.965599822106025e-05,
+      "loss": 0.7103,
+      "step": 27400
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9655707701553722e-05,
+      "loss": 0.6949,
+      "step": 27405
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.965541706157168e-05,
+      "loss": 0.6963,
+      "step": 27410
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9655126301117748e-05,
+      "loss": 0.7054,
+      "step": 27415
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.965483542019556e-05,
+      "loss": 0.7197,
+      "step": 27420
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9654544418808732e-05,
+      "loss": 0.7061,
+      "step": 27425
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.965425329696091e-05,
+      "loss": 0.7106,
+      "step": 27430
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9653962054655714e-05,
+      "loss": 0.709,
+      "step": 27435
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9653670691896785e-05,
+      "loss": 0.6915,
+      "step": 27440
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.965337920868776e-05,
+      "loss": 0.7164,
+      "step": 27445
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.965308760503227e-05,
+      "loss": 0.7114,
+      "step": 27450
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9652795880933956e-05,
+      "loss": 0.7158,
+      "step": 27455
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9652504036396456e-05,
+      "loss": 0.7083,
+      "step": 27460
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9652212071423415e-05,
+      "loss": 0.7136,
+      "step": 27465
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9651919986018473e-05,
+      "loss": 0.7255,
+      "step": 27470
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9651627780185277e-05,
+      "loss": 0.6957,
+      "step": 27475
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.965133545392747e-05,
+      "loss": 0.7105,
+      "step": 27480
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.96510430072487e-05,
+      "loss": 0.719,
+      "step": 27485
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.965075044015262e-05,
+      "loss": 0.7048,
+      "step": 27490
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9650457752642872e-05,
+      "loss": 0.7047,
+      "step": 27495
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9650164944723116e-05,
+      "loss": 0.7132,
+      "step": 27500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9649872016397e-05,
+      "loss": 0.7058,
+      "step": 27505
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9649578967668185e-05,
+      "loss": 0.7172,
+      "step": 27510
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9649285798540322e-05,
+      "loss": 0.709,
+      "step": 27515
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.964899250901707e-05,
+      "loss": 0.7048,
+      "step": 27520
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.964869909910209e-05,
+      "loss": 0.7032,
+      "step": 27525
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.964840556879904e-05,
+      "loss": 0.7015,
+      "step": 27530
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9648111918111584e-05,
+      "loss": 0.6965,
+      "step": 27535
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9647818147043386e-05,
+      "loss": 0.7046,
+      "step": 27540
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9647524255598114e-05,
+      "loss": 0.7074,
+      "step": 27545
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9647230243779432e-05,
+      "loss": 0.7102,
+      "step": 27550
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9646936111591005e-05,
+      "loss": 0.7016,
+      "step": 27555
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9646641859036513e-05,
+      "loss": 0.7087,
+      "step": 27560
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9646347486119615e-05,
+      "loss": 0.6943,
+      "step": 27565
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9646052992843994e-05,
+      "loss": 0.7147,
+      "step": 27570
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.964575837921332e-05,
+      "loss": 0.7014,
+      "step": 27575
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.964546364523127e-05,
+      "loss": 0.7117,
+      "step": 27580
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9645168790901523e-05,
+      "loss": 0.7016,
+      "step": 27585
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.964487381622775e-05,
+      "loss": 0.6993,
+      "step": 27590
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.964457872121364e-05,
+      "loss": 0.7099,
+      "step": 27595
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9644283505862877e-05,
+      "loss": 0.7101,
+      "step": 27600
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9643988170179135e-05,
+      "loss": 0.7232,
+      "step": 27605
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9643692714166103e-05,
+      "loss": 0.6992,
+      "step": 27610
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.964339713782747e-05,
+      "loss": 0.7165,
+      "step": 27615
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9643101441166924e-05,
+      "loss": 0.7104,
+      "step": 27620
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.964280562418815e-05,
+      "loss": 0.7181,
+      "step": 27625
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.964250968689484e-05,
+      "loss": 0.6977,
+      "step": 27630
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.964221362929069e-05,
+      "loss": 0.7,
+      "step": 27635
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.964191745137939e-05,
+      "loss": 0.7074,
+      "step": 27640
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.964162115316464e-05,
+      "loss": 0.7128,
+      "step": 27645
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9641324734650134e-05,
+      "loss": 0.7009,
+      "step": 27650
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9641028195839568e-05,
+      "loss": 0.7045,
+      "step": 27655
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.964073153673665e-05,
+      "loss": 0.7047,
+      "step": 27660
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.964043475734507e-05,
+      "loss": 0.7227,
+      "step": 27665
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.964013785766854e-05,
+      "loss": 0.7198,
+      "step": 27670
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.963984083771076e-05,
+      "loss": 0.7198,
+      "step": 27675
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.963954369747544e-05,
+      "loss": 0.701,
+      "step": 27680
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9639246436966285e-05,
+      "loss": 0.7103,
+      "step": 27685
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9638949056187e-05,
+      "loss": 0.7025,
+      "step": 27690
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9638651555141302e-05,
+      "loss": 0.7136,
+      "step": 27695
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9638353933832902e-05,
+      "loss": 0.7104,
+      "step": 27700
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.963805619226551e-05,
+      "loss": 0.7005,
+      "step": 27705
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9637758330442845e-05,
+      "loss": 0.7126,
+      "step": 27710
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9637460348368615e-05,
+      "loss": 0.7192,
+      "step": 27715
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.963716224604655e-05,
+      "loss": 0.7146,
+      "step": 27720
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9636864023480363e-05,
+      "loss": 0.7037,
+      "step": 27725
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9636565680673774e-05,
+      "loss": 0.6933,
+      "step": 27730
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9636267217630506e-05,
+      "loss": 0.6951,
+      "step": 27735
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9635968634354286e-05,
+      "loss": 0.6975,
+      "step": 27740
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9635669930848837e-05,
+      "loss": 0.6993,
+      "step": 27745
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.963537110711789e-05,
+      "loss": 0.706,
+      "step": 27750
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9635072163165164e-05,
+      "loss": 0.7042,
+      "step": 27755
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9634773098994397e-05,
+      "loss": 0.71,
+      "step": 27760
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9634473914609316e-05,
+      "loss": 0.7015,
+      "step": 27765
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9634174610013662e-05,
+      "loss": 0.7075,
+      "step": 27770
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.963387518521116e-05,
+      "loss": 0.7046,
+      "step": 27775
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.963357564020555e-05,
+      "loss": 0.6948,
+      "step": 27780
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.963327597500057e-05,
+      "loss": 0.7087,
+      "step": 27785
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9632976189599957e-05,
+      "loss": 0.7062,
+      "step": 27790
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9632676284007454e-05,
+      "loss": 0.7115,
+      "step": 27795
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.96323762582268e-05,
+      "loss": 0.7032,
+      "step": 27800
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.963207611226174e-05,
+      "loss": 0.702,
+      "step": 27805
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.963177584611602e-05,
+      "loss": 0.7061,
+      "step": 27810
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9631475459793385e-05,
+      "loss": 0.7038,
+      "step": 27815
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9631174953297582e-05,
+      "loss": 0.6882,
+      "step": 27820
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9630874326632365e-05,
+      "loss": 0.6991,
+      "step": 27825
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.963057357980148e-05,
+      "loss": 0.7046,
+      "step": 27830
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9630272712808682e-05,
+      "loss": 0.7114,
+      "step": 27835
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9629971725657724e-05,
+      "loss": 0.7158,
+      "step": 27840
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.962967061835236e-05,
+      "loss": 0.7154,
+      "step": 27845
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.962936939089635e-05,
+      "loss": 0.7,
+      "step": 27850
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9629068043293454e-05,
+      "loss": 0.7074,
+      "step": 27855
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9628766575547423e-05,
+      "loss": 0.6952,
+      "step": 27860
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.962846498766203e-05,
+      "loss": 0.6958,
+      "step": 27865
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.962816327964103e-05,
+      "loss": 0.6996,
+      "step": 27870
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.962786145148819e-05,
+      "loss": 0.715,
+      "step": 27875
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9627559503207278e-05,
+      "loss": 0.6997,
+      "step": 27880
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9627257434802053e-05,
+      "loss": 0.7042,
+      "step": 27885
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9626955246276297e-05,
+      "loss": 0.6894,
+      "step": 27890
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.962665293763377e-05,
+      "loss": 0.7059,
+      "step": 27895
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.962635050887825e-05,
+      "loss": 0.7096,
+      "step": 27900
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9626047960013505e-05,
+      "loss": 0.7026,
+      "step": 27905
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9625745291043316e-05,
+      "loss": 0.7075,
+      "step": 27910
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9625442501971455e-05,
+      "loss": 0.7166,
+      "step": 27915
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.96251395928017e-05,
+      "loss": 0.7084,
+      "step": 27920
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9624836563537837e-05,
+      "loss": 0.7174,
+      "step": 27925
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9624533414183636e-05,
+      "loss": 0.707,
+      "step": 27930
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9624230144742887e-05,
+      "loss": 0.7035,
+      "step": 27935
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9623926755219375e-05,
+      "loss": 0.7034,
+      "step": 27940
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.962362324561688e-05,
+      "loss": 0.7081,
+      "step": 27945
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9623319615939192e-05,
+      "loss": 0.7145,
+      "step": 27950
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9623015866190098e-05,
+      "loss": 0.7173,
+      "step": 27955
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9622711996373388e-05,
+      "loss": 0.7109,
+      "step": 27960
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9622408006492855e-05,
+      "loss": 0.7008,
+      "step": 27965
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9622103896552294e-05,
+      "loss": 0.7052,
+      "step": 27970
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9621799666555495e-05,
+      "loss": 0.6995,
+      "step": 27975
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9621495316506253e-05,
+      "loss": 0.7009,
+      "step": 27980
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9621190846408373e-05,
+      "loss": 0.7147,
+      "step": 27985
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9620886256265646e-05,
+      "loss": 0.6961,
+      "step": 27990
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9620581546081875e-05,
+      "loss": 0.7161,
+      "step": 27995
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.962027671586086e-05,
+      "loss": 0.7141,
+      "step": 28000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.961997176560641e-05,
+      "loss": 0.7008,
+      "step": 28005
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9619666695322324e-05,
+      "loss": 0.6965,
+      "step": 28010
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9619361505012414e-05,
+      "loss": 0.6987,
+      "step": 28015
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.961905619468048e-05,
+      "loss": 0.7051,
+      "step": 28020
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.961875076433034e-05,
+      "loss": 0.7056,
+      "step": 28025
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.96184452139658e-05,
+      "loss": 0.7049,
+      "step": 28030
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9618139543590676e-05,
+      "loss": 0.6997,
+      "step": 28035
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9617833753208775e-05,
+      "loss": 0.7087,
+      "step": 28040
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.961752784282392e-05,
+      "loss": 0.6991,
+      "step": 28045
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9617221812439925e-05,
+      "loss": 0.6901,
+      "step": 28050
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9616915662060606e-05,
+      "loss": 0.7069,
+      "step": 28055
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9616609391689786e-05,
+      "loss": 0.6824,
+      "step": 28060
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9616303001331284e-05,
+      "loss": 0.7105,
+      "step": 28065
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.961599649098892e-05,
+      "loss": 0.6949,
+      "step": 28070
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.961568986066653e-05,
+      "loss": 0.7021,
+      "step": 28075
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.961538311036793e-05,
+      "loss": 0.7066,
+      "step": 28080
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.961507624009695e-05,
+      "loss": 0.7148,
+      "step": 28085
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.961476924985742e-05,
+      "loss": 0.7109,
+      "step": 28090
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9614462139653164e-05,
+      "loss": 0.7047,
+      "step": 28095
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.961415490948802e-05,
+      "loss": 0.6999,
+      "step": 28100
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.961384755936582e-05,
+      "loss": 0.7001,
+      "step": 28105
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9613540089290402e-05,
+      "loss": 0.6976,
+      "step": 28110
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9613232499265598e-05,
+      "loss": 0.7021,
+      "step": 28115
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.961292478929525e-05,
+      "loss": 0.706,
+      "step": 28120
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.961261695938319e-05,
+      "loss": 0.7109,
+      "step": 28125
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9612309009533264e-05,
+      "loss": 0.6956,
+      "step": 28130
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9612000939749317e-05,
+      "loss": 0.7031,
+      "step": 28135
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9611692750035188e-05,
+      "loss": 0.7046,
+      "step": 28140
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.961138444039472e-05,
+      "loss": 0.716,
+      "step": 28145
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.961107601083177e-05,
+      "loss": 0.7092,
+      "step": 28150
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9610767461350174e-05,
+      "loss": 0.6933,
+      "step": 28155
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9610458791953793e-05,
+      "loss": 0.7093,
+      "step": 28160
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9610150002646467e-05,
+      "loss": 0.7115,
+      "step": 28165
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.960984109343206e-05,
+      "loss": 0.7152,
+      "step": 28170
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.960953206431442e-05,
+      "loss": 0.6995,
+      "step": 28175
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9609222915297405e-05,
+      "loss": 0.6993,
+      "step": 28180
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.960891364638487e-05,
+      "loss": 0.7071,
+      "step": 28185
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9608604257580674e-05,
+      "loss": 0.714,
+      "step": 28190
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.960829474888868e-05,
+      "loss": 0.7124,
+      "step": 28195
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9607985120312744e-05,
+      "loss": 0.6928,
+      "step": 28200
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.960767537185674e-05,
+      "loss": 0.7075,
+      "step": 28205
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.960736550352452e-05,
+      "loss": 0.7178,
+      "step": 28210
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9607055515319962e-05,
+      "loss": 0.7023,
+      "step": 28215
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9606745407246925e-05,
+      "loss": 0.7027,
+      "step": 28220
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9606435179309284e-05,
+      "loss": 0.6945,
+      "step": 28225
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9606124831510903e-05,
+      "loss": 0.701,
+      "step": 28230
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9605814363855663e-05,
+      "loss": 0.7042,
+      "step": 28235
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.960550377634743e-05,
+      "loss": 0.7012,
+      "step": 28240
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9605193068990082e-05,
+      "loss": 0.7095,
+      "step": 28245
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.96048822417875e-05,
+      "loss": 0.7026,
+      "step": 28250
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9604571294743555e-05,
+      "loss": 0.6911,
+      "step": 28255
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.960426022786213e-05,
+      "loss": 0.7079,
+      "step": 28260
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9603949041147107e-05,
+      "loss": 0.705,
+      "step": 28265
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.960363773460237e-05,
+      "loss": 0.6901,
+      "step": 28270
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.96033263082318e-05,
+      "loss": 0.7036,
+      "step": 28275
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9603014762039286e-05,
+      "loss": 0.7058,
+      "step": 28280
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.960270309602871e-05,
+      "loss": 0.7041,
+      "step": 28285
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9602391310203966e-05,
+      "loss": 0.6907,
+      "step": 28290
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9602079404568942e-05,
+      "loss": 0.6996,
+      "step": 28295
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9601767379127528e-05,
+      "loss": 0.7134,
+      "step": 28300
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.960145523388362e-05,
+      "loss": 0.697,
+      "step": 28305
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9601142968841112e-05,
+      "loss": 0.7031,
+      "step": 28310
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.96008305840039e-05,
+      "loss": 0.7055,
+      "step": 28315
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.960051807937588e-05,
+      "loss": 0.6952,
+      "step": 28320
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9600205454960952e-05,
+      "loss": 0.7039,
+      "step": 28325
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.959989271076302e-05,
+      "loss": 0.6986,
+      "step": 28330
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.959957984678598e-05,
+      "loss": 0.7093,
+      "step": 28335
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9599266863033743e-05,
+      "loss": 0.7075,
+      "step": 28340
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.959895375951021e-05,
+      "loss": 0.7096,
+      "step": 28345
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9598640536219288e-05,
+      "loss": 0.7063,
+      "step": 28350
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.959832719316488e-05,
+      "loss": 0.6836,
+      "step": 28355
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9598013730350906e-05,
+      "loss": 0.694,
+      "step": 28360
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.959770014778127e-05,
+      "loss": 0.6988,
+      "step": 28365
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9597386445459883e-05,
+      "loss": 0.7105,
+      "step": 28370
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9597072623390668e-05,
+      "loss": 0.7125,
+      "step": 28375
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9596758681577533e-05,
+      "loss": 0.7098,
+      "step": 28380
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.95964446200244e-05,
+      "loss": 0.7004,
+      "step": 28385
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9596130438735178e-05,
+      "loss": 0.7045,
+      "step": 28390
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9595816137713798e-05,
+      "loss": 0.7068,
+      "step": 28395
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9595501716964176e-05,
+      "loss": 0.7069,
+      "step": 28400
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9595187176490237e-05,
+      "loss": 0.7032,
+      "step": 28405
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9594872516295907e-05,
+      "loss": 0.6978,
+      "step": 28410
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9594557736385107e-05,
+      "loss": 0.6982,
+      "step": 28415
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.959424283676177e-05,
+      "loss": 0.7054,
+      "step": 28420
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.959392781742982e-05,
+      "loss": 0.6924,
+      "step": 28425
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.959361267839319e-05,
+      "loss": 0.7021,
+      "step": 28430
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9593297419655815e-05,
+      "loss": 0.697,
+      "step": 28435
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9592982041221625e-05,
+      "loss": 0.7028,
+      "step": 28440
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9592666543094558e-05,
+      "loss": 0.6965,
+      "step": 28445
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9592350925278546e-05,
+      "loss": 0.7079,
+      "step": 28450
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.959203518777753e-05,
+      "loss": 0.709,
+      "step": 28455
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9591719330595444e-05,
+      "loss": 0.7044,
+      "step": 28460
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9591403353736238e-05,
+      "loss": 0.7064,
+      "step": 28465
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.959108725720385e-05,
+      "loss": 0.7109,
+      "step": 28470
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9590771041002225e-05,
+      "loss": 0.7046,
+      "step": 28475
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9590454705135307e-05,
+      "loss": 0.6909,
+      "step": 28480
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.959013824960704e-05,
+      "loss": 0.694,
+      "step": 28485
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9589821674421382e-05,
+      "loss": 0.6941,
+      "step": 28490
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9589504979582274e-05,
+      "loss": 0.7018,
+      "step": 28495
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.958918816509367e-05,
+      "loss": 0.707,
+      "step": 28500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9588871230959524e-05,
+      "loss": 0.7036,
+      "step": 28505
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9588554177183794e-05,
+      "loss": 0.6899,
+      "step": 28510
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9588237003770426e-05,
+      "loss": 0.6886,
+      "step": 28515
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9587919710723385e-05,
+      "loss": 0.7054,
+      "step": 28520
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.958760229804663e-05,
+      "loss": 0.6793,
+      "step": 28525
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9587284765744118e-05,
+      "loss": 0.6968,
+      "step": 28530
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9586967113819812e-05,
+      "loss": 0.7083,
+      "step": 28535
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9586649342277676e-05,
+      "loss": 0.7023,
+      "step": 28540
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9586331451121678e-05,
+      "loss": 0.7028,
+      "step": 28545
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9586013440355774e-05,
+      "loss": 0.6978,
+      "step": 28550
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9585695309983946e-05,
+      "loss": 0.712,
+      "step": 28555
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9585377060010154e-05,
+      "loss": 0.7031,
+      "step": 28560
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9585058690438366e-05,
+      "loss": 0.706,
+      "step": 28565
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9584740201272567e-05,
+      "loss": 0.7008,
+      "step": 28570
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9584421592516717e-05,
+      "loss": 0.6958,
+      "step": 28575
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9584102864174804e-05,
+      "loss": 0.693,
+      "step": 28580
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9583784016250795e-05,
+      "loss": 0.7013,
+      "step": 28585
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.958346504874867e-05,
+      "loss": 0.7131,
+      "step": 28590
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.958314596167241e-05,
+      "loss": 0.6996,
+      "step": 28595
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9582826755026002e-05,
+      "loss": 0.7065,
+      "step": 28600
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.958250742881342e-05,
+      "loss": 0.7138,
+      "step": 28605
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.958218798303865e-05,
+      "loss": 0.6975,
+      "step": 28610
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9581868417705684e-05,
+      "loss": 0.7008,
+      "step": 28615
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9581548732818503e-05,
+      "loss": 0.7037,
+      "step": 28620
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.95812289283811e-05,
+      "loss": 0.7074,
+      "step": 28625
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.958090900439746e-05,
+      "loss": 0.7006,
+      "step": 28630
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9580588960871577e-05,
+      "loss": 0.6943,
+      "step": 28635
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9580268797807447e-05,
+      "loss": 0.6954,
+      "step": 28640
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.957994851520906e-05,
+      "loss": 0.7092,
+      "step": 28645
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9579628113080414e-05,
+      "loss": 0.6977,
+      "step": 28650
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9579307591425512e-05,
+      "loss": 0.6905,
+      "step": 28655
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9578986950248345e-05,
+      "loss": 0.6923,
+      "step": 28660
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.957866618955292e-05,
+      "loss": 0.7055,
+      "step": 28665
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9578345309343234e-05,
+      "loss": 0.7013,
+      "step": 28670
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9578024309623296e-05,
+      "loss": 0.7019,
+      "step": 28675
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9577703190397105e-05,
+      "loss": 0.6951,
+      "step": 28680
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9577381951668673e-05,
+      "loss": 0.6935,
+      "step": 28685
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9577060593442006e-05,
+      "loss": 0.7024,
+      "step": 28690
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9576739115721115e-05,
+      "loss": 0.6898,
+      "step": 28695
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9576417518510007e-05,
+      "loss": 0.7008,
+      "step": 28700
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.95760958018127e-05,
+      "loss": 0.71,
+      "step": 28705
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9575773965633202e-05,
+      "loss": 0.6978,
+      "step": 28710
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9575452009975535e-05,
+      "loss": 0.6864,
+      "step": 28715
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9575129934843712e-05,
+      "loss": 0.7105,
+      "step": 28720
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9574807740241754e-05,
+      "loss": 0.7031,
+      "step": 28725
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.957448542617368e-05,
+      "loss": 0.7065,
+      "step": 28730
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.957416299264351e-05,
+      "loss": 0.698,
+      "step": 28735
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9573840439655272e-05,
+      "loss": 0.6977,
+      "step": 28740
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9573517767212982e-05,
+      "loss": 0.6932,
+      "step": 28745
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9573194975320672e-05,
+      "loss": 0.7033,
+      "step": 28750
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9572872063982372e-05,
+      "loss": 0.6968,
+      "step": 28755
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9572549033202106e-05,
+      "loss": 0.7074,
+      "step": 28760
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9572225882983903e-05,
+      "loss": 0.6979,
+      "step": 28765
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9571902613331804e-05,
+      "loss": 0.6879,
+      "step": 28770
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.957157922424983e-05,
+      "loss": 0.7069,
+      "step": 28775
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9571255715742028e-05,
+      "loss": 0.7023,
+      "step": 28780
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9570932087812428e-05,
+      "loss": 0.6973,
+      "step": 28785
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9570608340465066e-05,
+      "loss": 0.6961,
+      "step": 28790
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9570284473703983e-05,
+      "loss": 0.7064,
+      "step": 28795
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9569960487533226e-05,
+      "loss": 0.7065,
+      "step": 28800
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.956963638195683e-05,
+      "loss": 0.7045,
+      "step": 28805
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9569312156978843e-05,
+      "loss": 0.7006,
+      "step": 28810
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9568987812603307e-05,
+      "loss": 0.7137,
+      "step": 28815
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.956866334883427e-05,
+      "loss": 0.6949,
+      "step": 28820
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9568338765675786e-05,
+      "loss": 0.6884,
+      "step": 28825
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9568014063131898e-05,
+      "loss": 0.6997,
+      "step": 28830
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9567689241206658e-05,
+      "loss": 0.6988,
+      "step": 28835
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.956736429990412e-05,
+      "loss": 0.7049,
+      "step": 28840
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.956703923922834e-05,
+      "loss": 0.6851,
+      "step": 28845
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9566714059183373e-05,
+      "loss": 0.6901,
+      "step": 28850
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9566388759773277e-05,
+      "loss": 0.6993,
+      "step": 28855
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9566063341002106e-05,
+      "loss": 0.7113,
+      "step": 28860
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9565737802873927e-05,
+      "loss": 0.7004,
+      "step": 28865
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9565412145392796e-05,
+      "loss": 0.7029,
+      "step": 28870
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.956508636856278e-05,
+      "loss": 0.7187,
+      "step": 28875
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9564760472387944e-05,
+      "loss": 0.6982,
+      "step": 28880
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.956443445687235e-05,
+      "loss": 0.6879,
+      "step": 28885
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.956410832202007e-05,
+      "loss": 0.705,
+      "step": 28890
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9563782067835174e-05,
+      "loss": 0.7103,
+      "step": 28895
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.956345569432173e-05,
+      "loss": 0.6996,
+      "step": 28900
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9563129201483807e-05,
+      "loss": 0.6929,
+      "step": 28905
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9562802589325488e-05,
+      "loss": 0.6989,
+      "step": 28910
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.956247585785084e-05,
+      "loss": 0.7083,
+      "step": 28915
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.956214900706394e-05,
+      "loss": 0.7114,
+      "step": 28920
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.956182203696887e-05,
+      "loss": 0.6912,
+      "step": 28925
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.956149494756971e-05,
+      "loss": 0.6995,
+      "step": 28930
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9561167738870538e-05,
+      "loss": 0.7179,
+      "step": 28935
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.956084041087544e-05,
+      "loss": 0.7051,
+      "step": 28940
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9560512963588495e-05,
+      "loss": 0.7009,
+      "step": 28945
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9560185397013794e-05,
+      "loss": 0.6937,
+      "step": 28950
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.955985771115542e-05,
+      "loss": 0.6949,
+      "step": 28955
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9559529906017466e-05,
+      "loss": 0.7006,
+      "step": 28960
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9559201981604016e-05,
+      "loss": 0.7012,
+      "step": 28965
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.955887393791917e-05,
+      "loss": 0.6984,
+      "step": 28970
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9558545774967007e-05,
+      "loss": 0.691,
+      "step": 28975
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9558217492751637e-05,
+      "loss": 0.6967,
+      "step": 28980
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.955788909127715e-05,
+      "loss": 0.6892,
+      "step": 28985
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.955756057054764e-05,
+      "loss": 0.6912,
+      "step": 28990
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9557231930567212e-05,
+      "loss": 0.6891,
+      "step": 28995
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9556903171339963e-05,
+      "loss": 0.6953,
+      "step": 29000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.955657429286999e-05,
+      "loss": 0.7024,
+      "step": 29005
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9556245295161408e-05,
+      "loss": 0.6928,
+      "step": 29010
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9555916178218315e-05,
+      "loss": 0.6899,
+      "step": 29015
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.955558694204482e-05,
+      "loss": 0.6992,
+      "step": 29020
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9555257586645026e-05,
+      "loss": 0.6996,
+      "step": 29025
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9554928112023046e-05,
+      "loss": 0.7109,
+      "step": 29030
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.955459851818299e-05,
+      "loss": 0.6915,
+      "step": 29035
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9554268805128973e-05,
+      "loss": 0.6912,
+      "step": 29040
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9553938972865106e-05,
+      "loss": 0.6986,
+      "step": 29045
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9553609021395507e-05,
+      "loss": 0.6931,
+      "step": 29050
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.955327895072429e-05,
+      "loss": 0.7033,
+      "step": 29055
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.955294876085557e-05,
+      "loss": 0.6862,
+      "step": 29060
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.955261845179348e-05,
+      "loss": 0.6848,
+      "step": 29065
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.955228802354212e-05,
+      "loss": 0.7036,
+      "step": 29070
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9551957476105637e-05,
+      "loss": 0.6989,
+      "step": 29075
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.955162680948814e-05,
+      "loss": 0.6985,
+      "step": 29080
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9551296023693754e-05,
+      "loss": 0.702,
+      "step": 29085
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9550965118726613e-05,
+      "loss": 0.7026,
+      "step": 29090
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9550634094590844e-05,
+      "loss": 0.7079,
+      "step": 29095
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.955030295129058e-05,
+      "loss": 0.6955,
+      "step": 29100
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9549971688829946e-05,
+      "loss": 0.7023,
+      "step": 29105
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.954964030721308e-05,
+      "loss": 0.6975,
+      "step": 29110
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9549308806444112e-05,
+      "loss": 0.6999,
+      "step": 29115
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9548977186527184e-05,
+      "loss": 0.6993,
+      "step": 29120
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9548645447466433e-05,
+      "loss": 0.6932,
+      "step": 29125
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.954831358926599e-05,
+      "loss": 0.6929,
+      "step": 29130
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9547981611930006e-05,
+      "loss": 0.6993,
+      "step": 29135
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9547649515462618e-05,
+      "loss": 0.6965,
+      "step": 29140
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.954731729986797e-05,
+      "loss": 0.7047,
+      "step": 29145
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9546984965150212e-05,
+      "loss": 0.6979,
+      "step": 29150
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.954665251131348e-05,
+      "loss": 0.6965,
+      "step": 29155
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9546319938361935e-05,
+      "loss": 0.7002,
+      "step": 29160
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9545987246299715e-05,
+      "loss": 0.7062,
+      "step": 29165
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9545654435130975e-05,
+      "loss": 0.7129,
+      "step": 29170
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9545321504859875e-05,
+      "loss": 0.7058,
+      "step": 29175
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9544988455490555e-05,
+      "loss": 0.6949,
+      "step": 29180
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.954465528702718e-05,
+      "loss": 0.6898,
+      "step": 29185
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9544321999473907e-05,
+      "loss": 0.6894,
+      "step": 29190
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9543988592834892e-05,
+      "loss": 0.7039,
+      "step": 29195
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9543655067114294e-05,
+      "loss": 0.6905,
+      "step": 29200
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9543321422316276e-05,
+      "loss": 0.7155,
+      "step": 29205
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9542987658445002e-05,
+      "loss": 0.6957,
+      "step": 29210
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9542653775504633e-05,
+      "loss": 0.6958,
+      "step": 29215
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9542319773499337e-05,
+      "loss": 0.7128,
+      "step": 29220
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9541985652433283e-05,
+      "loss": 0.6959,
+      "step": 29225
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9541651412310637e-05,
+      "loss": 0.6881,
+      "step": 29230
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9541317053135575e-05,
+      "loss": 0.7027,
+      "step": 29235
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9540982574912258e-05,
+      "loss": 0.7084,
+      "step": 29240
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.954064797764487e-05,
+      "loss": 0.6854,
+      "step": 29245
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.954031326133758e-05,
+      "loss": 0.6922,
+      "step": 29250
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.953997842599457e-05,
+      "loss": 0.6835,
+      "step": 29255
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9539643471620006e-05,
+      "loss": 0.6871,
+      "step": 29260
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9539308398218083e-05,
+      "loss": 0.699,
+      "step": 29265
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9538973205792967e-05,
+      "loss": 0.7041,
+      "step": 29270
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9538637894348852e-05,
+      "loss": 0.688,
+      "step": 29275
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9538302463889914e-05,
+      "loss": 0.6984,
+      "step": 29280
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9537966914420343e-05,
+      "loss": 0.7134,
+      "step": 29285
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9537631245944323e-05,
+      "loss": 0.7064,
+      "step": 29290
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9537295458466042e-05,
+      "loss": 0.7078,
+      "step": 29295
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9536959551989692e-05,
+      "loss": 0.6979,
+      "step": 29300
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.953662352651946e-05,
+      "loss": 0.6999,
+      "step": 29305
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9536287382059543e-05,
+      "loss": 0.6946,
+      "step": 29310
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.953595111861413e-05,
+      "loss": 0.6951,
+      "step": 29315
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9535614736187425e-05,
+      "loss": 0.6773,
+      "step": 29320
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.953527823478362e-05,
+      "loss": 0.6963,
+      "step": 29325
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.953494161440691e-05,
+      "loss": 0.7047,
+      "step": 29330
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.95346048750615e-05,
+      "loss": 0.6947,
+      "step": 29335
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.953426801675159e-05,
+      "loss": 0.7114,
+      "step": 29340
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9533931039481386e-05,
+      "loss": 0.6926,
+      "step": 29345
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9533593943255087e-05,
+      "loss": 0.6994,
+      "step": 29350
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.95332567280769e-05,
+      "loss": 0.6939,
+      "step": 29355
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.953291939395104e-05,
+      "loss": 0.7154,
+      "step": 29360
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9532581940881706e-05,
+      "loss": 0.6941,
+      "step": 29365
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9532244368873114e-05,
+      "loss": 0.6986,
+      "step": 29370
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9531906677929472e-05,
+      "loss": 0.697,
+      "step": 29375
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9531568868054997e-05,
+      "loss": 0.6989,
+      "step": 29380
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9531230939253906e-05,
+      "loss": 0.6795,
+      "step": 29385
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9530892891530408e-05,
+      "loss": 0.6862,
+      "step": 29390
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.953055472488873e-05,
+      "loss": 0.6959,
+      "step": 29395
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9530216439333085e-05,
+      "loss": 0.7144,
+      "step": 29400
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9529878034867692e-05,
+      "loss": 0.6926,
+      "step": 29405
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9529539511496776e-05,
+      "loss": 0.7083,
+      "step": 29410
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9529200869224566e-05,
+      "loss": 0.6962,
+      "step": 29415
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.952886210805528e-05,
+      "loss": 0.6932,
+      "step": 29420
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.952852322799315e-05,
+      "loss": 0.6985,
+      "step": 29425
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9528184229042398e-05,
+      "loss": 0.6977,
+      "step": 29430
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.952784511120726e-05,
+      "loss": 0.7137,
+      "step": 29435
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9527505874491962e-05,
+      "loss": 0.6853,
+      "step": 29440
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.952716651890074e-05,
+      "loss": 0.6881,
+      "step": 29445
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9526827044437833e-05,
+      "loss": 0.6913,
+      "step": 29450
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9526487451107465e-05,
+      "loss": 0.702,
+      "step": 29455
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9526147738913882e-05,
+      "loss": 0.7016,
+      "step": 29460
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9525807907861322e-05,
+      "loss": 0.71,
+      "step": 29465
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.952546795795402e-05,
+      "loss": 0.6946,
+      "step": 29470
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.952512788919622e-05,
+      "loss": 0.7067,
+      "step": 29475
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9524787701592167e-05,
+      "loss": 0.6961,
+      "step": 29480
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9524447395146105e-05,
+      "loss": 0.6901,
+      "step": 29485
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.952410696986228e-05,
+      "loss": 0.6842,
+      "step": 29490
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9523766425744937e-05,
+      "loss": 0.6856,
+      "step": 29495
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9523425762798328e-05,
+      "loss": 0.695,
+      "step": 29500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9523084981026704e-05,
+      "loss": 0.6841,
+      "step": 29505
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9522744080434312e-05,
+      "loss": 0.6967,
+      "step": 29510
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9522403061025413e-05,
+      "loss": 0.6885,
+      "step": 29515
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9522061922804253e-05,
+      "loss": 0.7001,
+      "step": 29520
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9521720665775098e-05,
+      "loss": 0.701,
+      "step": 29525
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.95213792899422e-05,
+      "loss": 0.6877,
+      "step": 29530
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.952103779530982e-05,
+      "loss": 0.6861,
+      "step": 29535
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9520696181882216e-05,
+      "loss": 0.6908,
+      "step": 29540
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9520354449663653e-05,
+      "loss": 0.7054,
+      "step": 29545
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9520012598658397e-05,
+      "loss": 0.6925,
+      "step": 29550
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9519670628870708e-05,
+      "loss": 0.6985,
+      "step": 29555
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.951932854030486e-05,
+      "loss": 0.7116,
+      "step": 29560
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.951898633296511e-05,
+      "loss": 0.7004,
+      "step": 29565
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.951864400685574e-05,
+      "loss": 0.7016,
+      "step": 29570
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9518301561981016e-05,
+      "loss": 0.6956,
+      "step": 29575
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.951795899834521e-05,
+      "loss": 0.693,
+      "step": 29580
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9517616315952598e-05,
+      "loss": 0.7004,
+      "step": 29585
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.951727351480745e-05,
+      "loss": 0.6908,
+      "step": 29590
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9516930594914053e-05,
+      "loss": 0.7007,
+      "step": 29595
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9516587556276677e-05,
+      "loss": 0.6948,
+      "step": 29600
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9516244398899607e-05,
+      "loss": 0.7053,
+      "step": 29605
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9515901122787126e-05,
+      "loss": 0.7007,
+      "step": 29610
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9515557727943512e-05,
+      "loss": 0.6967,
+      "step": 29615
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9515214214373052e-05,
+      "loss": 0.6876,
+      "step": 29620
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.951487058208003e-05,
+      "loss": 0.7033,
+      "step": 29625
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.951452683106874e-05,
+      "loss": 0.6975,
+      "step": 29630
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9514182961343463e-05,
+      "loss": 0.6997,
+      "step": 29635
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9513838972908498e-05,
+      "loss": 0.7077,
+      "step": 29640
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.951349486576813e-05,
+      "loss": 0.704,
+      "step": 29645
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.951315063992665e-05,
+      "loss": 0.7011,
+      "step": 29650
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.951280629538836e-05,
+      "loss": 0.6926,
+      "step": 29655
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9512461832157554e-05,
+      "loss": 0.6906,
+      "step": 29660
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9512117250238533e-05,
+      "loss": 0.6801,
+      "step": 29665
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.951177254963559e-05,
+      "loss": 0.7152,
+      "step": 29670
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.951142773035303e-05,
+      "loss": 0.702,
+      "step": 29675
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9511082792395152e-05,
+      "loss": 0.6997,
+      "step": 29680
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9510737735766266e-05,
+      "loss": 0.6987,
+      "step": 29685
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.951039256047067e-05,
+      "loss": 0.6907,
+      "step": 29690
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9510047266512676e-05,
+      "loss": 0.6991,
+      "step": 29695
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.950970185389659e-05,
+      "loss": 0.7025,
+      "step": 29700
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9509356322626728e-05,
+      "loss": 0.6759,
+      "step": 29705
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.950901067270739e-05,
+      "loss": 0.6858,
+      "step": 29710
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9508664904142897e-05,
+      "loss": 0.6951,
+      "step": 29715
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.950831901693756e-05,
+      "loss": 0.693,
+      "step": 29720
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.950797301109569e-05,
+      "loss": 0.6941,
+      "step": 29725
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9507626886621618e-05,
+      "loss": 0.6987,
+      "step": 29730
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.950728064351965e-05,
+      "loss": 0.695,
+      "step": 29735
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.950693428179411e-05,
+      "loss": 0.6885,
+      "step": 29740
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.950658780144932e-05,
+      "loss": 0.7051,
+      "step": 29745
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.95062412024896e-05,
+      "loss": 0.697,
+      "step": 29750
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.950589448491928e-05,
+      "loss": 0.6886,
+      "step": 29755
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9505547648742687e-05,
+      "loss": 0.6947,
+      "step": 29760
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.950520069396414e-05,
+      "loss": 0.6936,
+      "step": 29765
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9504853620587977e-05,
+      "loss": 0.6851,
+      "step": 29770
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.950450642861852e-05,
+      "loss": 0.6811,
+      "step": 29775
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.950415911806011e-05,
+      "loss": 0.6854,
+      "step": 29780
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.950381168891707e-05,
+      "loss": 0.6876,
+      "step": 29785
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9503464141193747e-05,
+      "loss": 0.7096,
+      "step": 29790
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9503116474894467e-05,
+      "loss": 0.6945,
+      "step": 29795
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9502768690023574e-05,
+      "loss": 0.6954,
+      "step": 29800
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9502420786585406e-05,
+      "loss": 0.6979,
+      "step": 29805
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9502072764584303e-05,
+      "loss": 0.6893,
+      "step": 29810
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9501724624024608e-05,
+      "loss": 0.6898,
+      "step": 29815
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9501376364910664e-05,
+      "loss": 0.6868,
+      "step": 29820
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9501027987246813e-05,
+      "loss": 0.7001,
+      "step": 29825
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.950067949103741e-05,
+      "loss": 0.692,
+      "step": 29830
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9500330876286798e-05,
+      "loss": 0.7014,
+      "step": 29835
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9499982142999325e-05,
+      "loss": 0.6905,
+      "step": 29840
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.949963329117935e-05,
+      "loss": 0.6934,
+      "step": 29845
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9499284320831213e-05,
+      "loss": 0.6943,
+      "step": 29850
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.949893523195928e-05,
+      "loss": 0.7038,
+      "step": 29855
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.94985860245679e-05,
+      "loss": 0.6836,
+      "step": 29860
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9498236698661433e-05,
+      "loss": 0.7001,
+      "step": 29865
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9497887254244237e-05,
+      "loss": 0.6944,
+      "step": 29870
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.949753769132067e-05,
+      "loss": 0.7058,
+      "step": 29875
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9497188009895097e-05,
+      "loss": 0.6932,
+      "step": 29880
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9496838209971878e-05,
+      "loss": 0.6916,
+      "step": 29885
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.949648829155538e-05,
+      "loss": 0.6858,
+      "step": 29890
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9496138254649968e-05,
+      "loss": 0.6954,
+      "step": 29895
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9495788099260008e-05,
+      "loss": 0.694,
+      "step": 29900
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.949543782538987e-05,
+      "loss": 0.6968,
+      "step": 29905
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9495087433043928e-05,
+      "loss": 0.7111,
+      "step": 29910
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9494736922226546e-05,
+      "loss": 0.6957,
+      "step": 29915
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9494386292942102e-05,
+      "loss": 0.7045,
+      "step": 29920
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9494035545194975e-05,
+      "loss": 0.6927,
+      "step": 29925
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9493684678989534e-05,
+      "loss": 0.696,
+      "step": 29930
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9493333694330158e-05,
+      "loss": 0.7051,
+      "step": 29935
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.949298259122123e-05,
+      "loss": 0.6936,
+      "step": 29940
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.949263136966713e-05,
+      "loss": 0.6895,
+      "step": 29945
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9492280029672237e-05,
+      "loss": 0.6987,
+      "step": 29950
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9491928571240936e-05,
+      "loss": 0.6972,
+      "step": 29955
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9491576994377618e-05,
+      "loss": 0.6918,
+      "step": 29960
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.949122529908666e-05,
+      "loss": 0.6971,
+      "step": 29965
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9490873485372457e-05,
+      "loss": 0.6913,
+      "step": 29970
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9490521553239397e-05,
+      "loss": 0.6963,
+      "step": 29975
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.949016950269187e-05,
+      "loss": 0.6921,
+      "step": 29980
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9489817333734265e-05,
+      "loss": 0.6933,
+      "step": 29985
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9489465046370986e-05,
+      "loss": 0.6894,
+      "step": 29990
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.948911264060642e-05,
+      "loss": 0.6889,
+      "step": 29995
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9488760116444966e-05,
+      "loss": 0.7077,
+      "step": 30000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9488407473891024e-05,
+      "loss": 0.701,
+      "step": 30005
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9488054712948995e-05,
+      "loss": 0.6996,
+      "step": 30010
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.948770183362328e-05,
+      "loss": 0.6912,
+      "step": 30015
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9487348835918273e-05,
+      "loss": 0.6934,
+      "step": 30020
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9486995719838392e-05,
+      "loss": 0.7003,
+      "step": 30025
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9486642485388035e-05,
+      "loss": 0.7021,
+      "step": 30030
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9486289132571613e-05,
+      "loss": 0.6917,
+      "step": 30035
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9485935661393532e-05,
+      "loss": 0.688,
+      "step": 30040
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9485582071858202e-05,
+      "loss": 0.6983,
+      "step": 30045
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9485228363970038e-05,
+      "loss": 0.6834,
+      "step": 30050
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9484874537733448e-05,
+      "loss": 0.7053,
+      "step": 30055
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9484520593152854e-05,
+      "loss": 0.6945,
+      "step": 30060
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9484166530232666e-05,
+      "loss": 0.6912,
+      "step": 30065
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9483812348977307e-05,
+      "loss": 0.6887,
+      "step": 30070
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9483458049391188e-05,
+      "loss": 0.6833,
+      "step": 30075
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.948310363147874e-05,
+      "loss": 0.7071,
+      "step": 30080
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.948274909524438e-05,
+      "loss": 0.6747,
+      "step": 30085
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.948239444069253e-05,
+      "loss": 0.6994,
+      "step": 30090
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9482039667827614e-05,
+      "loss": 0.6911,
+      "step": 30095
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.948168477665406e-05,
+      "loss": 0.6863,
+      "step": 30100
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9481329767176303e-05,
+      "loss": 0.6882,
+      "step": 30105
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9480974639398762e-05,
+      "loss": 0.682,
+      "step": 30110
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9480619393325875e-05,
+      "loss": 0.6924,
+      "step": 30115
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.948026402896207e-05,
+      "loss": 0.6911,
+      "step": 30120
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9479908546311783e-05,
+      "loss": 0.687,
+      "step": 30125
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9479552945379454e-05,
+      "loss": 0.6909,
+      "step": 30130
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.947919722616951e-05,
+      "loss": 0.6996,
+      "step": 30135
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9478841388686397e-05,
+      "loss": 0.6915,
+      "step": 30140
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9478485432934554e-05,
+      "loss": 0.686,
+      "step": 30145
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.947812935891842e-05,
+      "loss": 0.6933,
+      "step": 30150
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9477773166642434e-05,
+      "loss": 0.6796,
+      "step": 30155
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9477416856111048e-05,
+      "loss": 0.6922,
+      "step": 30160
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9477060427328705e-05,
+      "loss": 0.6911,
+      "step": 30165
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9476703880299852e-05,
+      "loss": 0.6833,
+      "step": 30170
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.947634721502894e-05,
+      "loss": 0.7039,
+      "step": 30175
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.947599043152041e-05,
+      "loss": 0.701,
+      "step": 30180
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9475633529778726e-05,
+      "loss": 0.7022,
+      "step": 30185
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9475276509808332e-05,
+      "loss": 0.6924,
+      "step": 30190
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9474919371613687e-05,
+      "loss": 0.7013,
+      "step": 30195
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9474562115199246e-05,
+      "loss": 0.6909,
+      "step": 30200
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9474204740569466e-05,
+      "loss": 0.688,
+      "step": 30205
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.947384724772881e-05,
+      "loss": 0.6774,
+      "step": 30210
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.947348963668173e-05,
+      "loss": 0.6947,
+      "step": 30215
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9473131907432695e-05,
+      "loss": 0.7061,
+      "step": 30220
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.947277405998617e-05,
+      "loss": 0.6943,
+      "step": 30225
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.947241609434661e-05,
+      "loss": 0.6904,
+      "step": 30230
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.947205801051849e-05,
+      "loss": 0.683,
+      "step": 30235
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9471699808506276e-05,
+      "loss": 0.7006,
+      "step": 30240
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.947134148831444e-05,
+      "loss": 0.7021,
+      "step": 30245
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9470983049947446e-05,
+      "loss": 0.6998,
+      "step": 30250
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.947062449340977e-05,
+      "loss": 0.6938,
+      "step": 30255
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9470265818705885e-05,
+      "loss": 0.6973,
+      "step": 30260
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.946990702584027e-05,
+      "loss": 0.7047,
+      "step": 30265
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9469548114817395e-05,
+      "loss": 0.6991,
+      "step": 30270
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9469189085641743e-05,
+      "loss": 0.6845,
+      "step": 30275
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9468829938317798e-05,
+      "loss": 0.6874,
+      "step": 30280
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9468470672850027e-05,
+      "loss": 0.6912,
+      "step": 30285
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9468111289242925e-05,
+      "loss": 0.6955,
+      "step": 30290
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9467751787500974e-05,
+      "loss": 0.7039,
+      "step": 30295
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9467392167628655e-05,
+      "loss": 0.6855,
+      "step": 30300
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9467032429630462e-05,
+      "loss": 0.6909,
+      "step": 30305
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9466672573510875e-05,
+      "loss": 0.6952,
+      "step": 30310
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.946631259927439e-05,
+      "loss": 0.6877,
+      "step": 30315
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9465952506925497e-05,
+      "loss": 0.7036,
+      "step": 30320
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.946559229646869e-05,
+      "loss": 0.6774,
+      "step": 30325
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.946523196790846e-05,
+      "loss": 0.6999,
+      "step": 30330
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9464871521249305e-05,
+      "loss": 0.6767,
+      "step": 30335
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9464510956495724e-05,
+      "loss": 0.6865,
+      "step": 30340
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9464150273652214e-05,
+      "loss": 0.6874,
+      "step": 30345
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9463789472723272e-05,
+      "loss": 0.6899,
+      "step": 30350
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9463428553713407e-05,
+      "loss": 0.6848,
+      "step": 30355
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9463067516627116e-05,
+      "loss": 0.692,
+      "step": 30360
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9462706361468907e-05,
+      "loss": 0.6968,
+      "step": 30365
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9462345088243287e-05,
+      "loss": 0.7147,
+      "step": 30370
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.946198369695476e-05,
+      "loss": 0.6994,
+      "step": 30375
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9461622187607837e-05,
+      "loss": 0.6918,
+      "step": 30380
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.946126056020703e-05,
+      "loss": 0.6965,
+      "step": 30385
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9460898814756848e-05,
+      "loss": 0.7039,
+      "step": 30390
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9460536951261808e-05,
+      "loss": 0.6973,
+      "step": 30395
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.946017496972642e-05,
+      "loss": 0.6834,
+      "step": 30400
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.945981287015521e-05,
+      "loss": 0.6906,
+      "step": 30405
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9459450652552684e-05,
+      "loss": 0.7102,
+      "step": 30410
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9459088316923373e-05,
+      "loss": 0.6969,
+      "step": 30415
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9458725863271784e-05,
+      "loss": 0.6979,
+      "step": 30420
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9458363291602455e-05,
+      "loss": 0.6963,
+      "step": 30425
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.94580006019199e-05,
+      "loss": 0.6908,
+      "step": 30430
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9457637794228645e-05,
+      "loss": 0.6882,
+      "step": 30435
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9457274868533218e-05,
+      "loss": 0.6861,
+      "step": 30440
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.945691182483815e-05,
+      "loss": 0.7041,
+      "step": 30445
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9456548663147966e-05,
+      "loss": 0.6929,
+      "step": 30450
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.94561853834672e-05,
+      "loss": 0.6983,
+      "step": 30455
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9455821985800388e-05,
+      "loss": 0.6902,
+      "step": 30460
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9455458470152057e-05,
+      "loss": 0.6866,
+      "step": 30465
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.945509483652675e-05,
+      "loss": 0.7003,
+      "step": 30470
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9454731084928995e-05,
+      "loss": 0.682,
+      "step": 30475
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.945436721536334e-05,
+      "loss": 0.7025,
+      "step": 30480
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.945400322783432e-05,
+      "loss": 0.696,
+      "step": 30485
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9453639122346475e-05,
+      "loss": 0.6965,
+      "step": 30490
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.945327489890435e-05,
+      "loss": 0.7038,
+      "step": 30495
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9452910557512497e-05,
+      "loss": 0.6898,
+      "step": 30500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9452546098175448e-05,
+      "loss": 0.6879,
+      "step": 30505
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.945218152089776e-05,
+      "loss": 0.6739,
+      "step": 30510
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.945181682568398e-05,
+      "loss": 0.6999,
+      "step": 30515
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9451452012538653e-05,
+      "loss": 0.6957,
+      "step": 30520
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9451087081466337e-05,
+      "loss": 0.6868,
+      "step": 30525
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9450722032471584e-05,
+      "loss": 0.686,
+      "step": 30530
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.945035686555895e-05,
+      "loss": 0.6918,
+      "step": 30535
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.944999158073299e-05,
+      "loss": 0.6956,
+      "step": 30540
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9449626177998257e-05,
+      "loss": 0.6954,
+      "step": 30545
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9449260657359317e-05,
+      "loss": 0.6808,
+      "step": 30550
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.944889501882073e-05,
+      "loss": 0.6935,
+      "step": 30555
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9448529262387054e-05,
+      "loss": 0.6912,
+      "step": 30560
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9448163388062853e-05,
+      "loss": 0.6916,
+      "step": 30565
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9447797395852698e-05,
+      "loss": 0.6854,
+      "step": 30570
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9447431285761148e-05,
+      "loss": 0.7037,
+      "step": 30575
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9447065057792777e-05,
+      "loss": 0.6875,
+      "step": 30580
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.944669871195215e-05,
+      "loss": 0.7006,
+      "step": 30585
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9446332248243844e-05,
+      "loss": 0.6798,
+      "step": 30590
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.944596566667242e-05,
+      "loss": 0.7042,
+      "step": 30595
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9445598967242464e-05,
+      "loss": 0.6801,
+      "step": 30600
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9445232149958543e-05,
+      "loss": 0.6868,
+      "step": 30605
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.944486521482524e-05,
+      "loss": 0.6858,
+      "step": 30610
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.944449816184713e-05,
+      "loss": 0.6835,
+      "step": 30615
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9444130991028793e-05,
+      "loss": 0.6962,
+      "step": 30620
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.944376370237481e-05,
+      "loss": 0.6896,
+      "step": 30625
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9443396295889767e-05,
+      "loss": 0.6898,
+      "step": 30630
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9443028771578243e-05,
+      "loss": 0.6853,
+      "step": 30635
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9442661129444828e-05,
+      "loss": 0.6951,
+      "step": 30640
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9442293369494108e-05,
+      "loss": 0.6958,
+      "step": 30645
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.944192549173067e-05,
+      "loss": 0.6999,
+      "step": 30650
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9441557496159103e-05,
+      "loss": 0.6993,
+      "step": 30655
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9441189382784e-05,
+      "loss": 0.6949,
+      "step": 30660
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9440821151609956e-05,
+      "loss": 0.697,
+      "step": 30665
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9440452802641567e-05,
+      "loss": 0.6905,
+      "step": 30670
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.944008433588342e-05,
+      "loss": 0.6957,
+      "step": 30675
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.943971575134012e-05,
+      "loss": 0.6994,
+      "step": 30680
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9439347049016266e-05,
+      "loss": 0.684,
+      "step": 30685
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9438978228916456e-05,
+      "loss": 0.7001,
+      "step": 30690
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.943860929104529e-05,
+      "loss": 0.705,
+      "step": 30695
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9438240235407375e-05,
+      "loss": 0.6911,
+      "step": 30700
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9437871062007314e-05,
+      "loss": 0.6947,
+      "step": 30705
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9437501770849714e-05,
+      "loss": 0.7019,
+      "step": 30710
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9437132361939182e-05,
+      "loss": 0.6957,
+      "step": 30715
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9436762835280327e-05,
+      "loss": 0.6934,
+      "step": 30720
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.943639319087776e-05,
+      "loss": 0.6864,
+      "step": 30725
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9436023428736095e-05,
+      "loss": 0.7007,
+      "step": 30730
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9435653548859944e-05,
+      "loss": 0.6899,
+      "step": 30735
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.943528355125392e-05,
+      "loss": 0.6904,
+      "step": 30740
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.943491343592264e-05,
+      "loss": 0.7039,
+      "step": 30745
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9434543202870726e-05,
+      "loss": 0.6771,
+      "step": 30750
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9434172852102793e-05,
+      "loss": 0.6927,
+      "step": 30755
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9433802383623464e-05,
+      "loss": 0.6987,
+      "step": 30760
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.943343179743736e-05,
+      "loss": 0.6946,
+      "step": 30765
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.943306109354911e-05,
+      "loss": 0.6944,
+      "step": 30770
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9432690271963327e-05,
+      "loss": 0.6886,
+      "step": 30775
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9432319332684654e-05,
+      "loss": 0.6914,
+      "step": 30780
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9431948275717706e-05,
+      "loss": 0.6978,
+      "step": 30785
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.943157710106712e-05,
+      "loss": 0.6871,
+      "step": 30790
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.943120580873753e-05,
+      "loss": 0.6858,
+      "step": 30795
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9430834398733557e-05,
+      "loss": 0.6887,
+      "step": 30800
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9430462871059842e-05,
+      "loss": 0.6741,
+      "step": 30805
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.943009122572102e-05,
+      "loss": 0.6924,
+      "step": 30810
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9429719462721732e-05,
+      "loss": 0.6877,
+      "step": 30815
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9429347582066614e-05,
+      "loss": 0.6956,
+      "step": 30820
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9428975583760303e-05,
+      "loss": 0.6966,
+      "step": 30825
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.942860346780744e-05,
+      "loss": 0.7065,
+      "step": 30830
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9428231234212673e-05,
+      "loss": 0.699,
+      "step": 30835
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9427858882980645e-05,
+      "loss": 0.6878,
+      "step": 30840
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9427486414116e-05,
+      "loss": 0.7103,
+      "step": 30845
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9427113827623385e-05,
+      "loss": 0.6879,
+      "step": 30850
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.942674112350745e-05,
+      "loss": 0.6895,
+      "step": 30855
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9426368301772848e-05,
+      "loss": 0.6946,
+      "step": 30860
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9425995362424225e-05,
+      "loss": 0.6946,
+      "step": 30865
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.942562230546624e-05,
+      "loss": 0.6826,
+      "step": 30870
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9425249130903544e-05,
+      "loss": 0.6892,
+      "step": 30875
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9424875838740792e-05,
+      "loss": 0.704,
+      "step": 30880
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9424502428982644e-05,
+      "loss": 0.7032,
+      "step": 30885
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.942412890163376e-05,
+      "loss": 0.6946,
+      "step": 30890
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.94237552566988e-05,
+      "loss": 0.6866,
+      "step": 30895
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9423381494182422e-05,
+      "loss": 0.6775,
+      "step": 30900
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9423007614089294e-05,
+      "loss": 0.6945,
+      "step": 30905
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9422633616424083e-05,
+      "loss": 0.7001,
+      "step": 30910
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9422259501191448e-05,
+      "loss": 0.6903,
+      "step": 30915
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9421885268396064e-05,
+      "loss": 0.6696,
+      "step": 30920
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9421510918042593e-05,
+      "loss": 0.6946,
+      "step": 30925
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9421136450135715e-05,
+      "loss": 0.6918,
+      "step": 30930
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9420761864680095e-05,
+      "loss": 0.6908,
+      "step": 30935
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.942038716168041e-05,
+      "loss": 0.6876,
+      "step": 30940
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9420012341141334e-05,
+      "loss": 0.6799,
+      "step": 30945
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9419637403067547e-05,
+      "loss": 0.6858,
+      "step": 30950
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9419262347463718e-05,
+      "loss": 0.6901,
+      "step": 30955
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9418887174334538e-05,
+      "loss": 0.6952,
+      "step": 30960
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.941851188368468e-05,
+      "loss": 0.6925,
+      "step": 30965
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9418136475518833e-05,
+      "loss": 0.7038,
+      "step": 30970
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9417760949841672e-05,
+      "loss": 0.6917,
+      "step": 30975
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9417385306657894e-05,
+      "loss": 0.678,
+      "step": 30980
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9417009545972178e-05,
+      "loss": 0.6858,
+      "step": 30985
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9416633667789212e-05,
+      "loss": 0.6846,
+      "step": 30990
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9416257672113693e-05,
+      "loss": 0.6806,
+      "step": 30995
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9415881558950302e-05,
+      "loss": 0.699,
+      "step": 31000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.941550532830374e-05,
+      "loss": 0.6902,
+      "step": 31005
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.94151289801787e-05,
+      "loss": 0.6807,
+      "step": 31010
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9414752514579876e-05,
+      "loss": 0.6894,
+      "step": 31015
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9414375931511965e-05,
+      "loss": 0.6938,
+      "step": 31020
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9413999230979666e-05,
+      "loss": 0.6846,
+      "step": 31025
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.941362241298768e-05,
+      "loss": 0.696,
+      "step": 31030
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9413245477540708e-05,
+      "loss": 0.6952,
+      "step": 31035
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9412868424643452e-05,
+      "loss": 0.6884,
+      "step": 31040
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.941249125430062e-05,
+      "loss": 0.6919,
+      "step": 31045
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.941211396651691e-05,
+      "loss": 0.6936,
+      "step": 31050
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9411736561297043e-05,
+      "loss": 0.6885,
+      "step": 31055
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9411359038645715e-05,
+      "loss": 0.6955,
+      "step": 31060
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9410981398567642e-05,
+      "loss": 0.6869,
+      "step": 31065
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9410603641067537e-05,
+      "loss": 0.6859,
+      "step": 31070
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.941022576615011e-05,
+      "loss": 0.6955,
+      "step": 31075
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.940984777382008e-05,
+      "loss": 0.6881,
+      "step": 31080
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.940946966408216e-05,
+      "loss": 0.6787,
+      "step": 31085
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9409091436941067e-05,
+      "loss": 0.6915,
+      "step": 31090
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9408713092401525e-05,
+      "loss": 0.6834,
+      "step": 31095
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9408334630468247e-05,
+      "loss": 0.6873,
+      "step": 31100
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.940795605114596e-05,
+      "loss": 0.7059,
+      "step": 31105
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.940757735443939e-05,
+      "loss": 0.6817,
+      "step": 31110
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9407198540353258e-05,
+      "loss": 0.6926,
+      "step": 31115
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9406819608892288e-05,
+      "loss": 0.7024,
+      "step": 31120
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9406440560061214e-05,
+      "loss": 0.6873,
+      "step": 31125
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9406061393864764e-05,
+      "loss": 0.6941,
+      "step": 31130
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9405682110307668e-05,
+      "loss": 0.6937,
+      "step": 31135
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9405302709394658e-05,
+      "loss": 0.6838,
+      "step": 31140
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9404923191130468e-05,
+      "loss": 0.6824,
+      "step": 31145
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.940454355551983e-05,
+      "loss": 0.6908,
+      "step": 31150
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9404163802567487e-05,
+      "loss": 0.6843,
+      "step": 31155
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9403783932278176e-05,
+      "loss": 0.6951,
+      "step": 31160
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9403403944656635e-05,
+      "loss": 0.6924,
+      "step": 31165
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9403023839707603e-05,
+      "loss": 0.6865,
+      "step": 31170
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9402643617435824e-05,
+      "loss": 0.6886,
+      "step": 31175
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9402263277846046e-05,
+      "loss": 0.6936,
+      "step": 31180
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.940188282094301e-05,
+      "loss": 0.6842,
+      "step": 31185
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9401502246731464e-05,
+      "loss": 0.6774,
+      "step": 31190
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.940112155521616e-05,
+      "loss": 0.6841,
+      "step": 31195
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.940074074640184e-05,
+      "loss": 0.6917,
+      "step": 31200
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9400359820293265e-05,
+      "loss": 0.6831,
+      "step": 31205
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9399978776895177e-05,
+      "loss": 0.6859,
+      "step": 31210
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9399597616212344e-05,
+      "loss": 0.7006,
+      "step": 31215
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9399216338249508e-05,
+      "loss": 0.6939,
+      "step": 31220
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9398834943011434e-05,
+      "loss": 0.6969,
+      "step": 31225
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.939845343050288e-05,
+      "loss": 0.692,
+      "step": 31230
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9398071800728608e-05,
+      "loss": 0.6937,
+      "step": 31235
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9397690053693373e-05,
+      "loss": 0.6834,
+      "step": 31240
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9397308189401944e-05,
+      "loss": 0.6963,
+      "step": 31245
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9396926207859085e-05,
+      "loss": 0.6806,
+      "step": 31250
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9396544109069563e-05,
+      "loss": 0.6842,
+      "step": 31255
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9396161893038138e-05,
+      "loss": 0.6963,
+      "step": 31260
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9395779559769584e-05,
+      "loss": 0.678,
+      "step": 31265
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9395397109268676e-05,
+      "loss": 0.6752,
+      "step": 31270
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.939501454154018e-05,
+      "loss": 0.7012,
+      "step": 31275
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.939463185658887e-05,
+      "loss": 0.6822,
+      "step": 31280
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.939424905441952e-05,
+      "loss": 0.6843,
+      "step": 31285
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.939386613503691e-05,
+      "loss": 0.6932,
+      "step": 31290
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9393483098445815e-05,
+      "loss": 0.6933,
+      "step": 31295
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9393099944651017e-05,
+      "loss": 0.6832,
+      "step": 31300
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9392716673657292e-05,
+      "loss": 0.676,
+      "step": 31305
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9392333285469427e-05,
+      "loss": 0.6707,
+      "step": 31310
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.93919497800922e-05,
+      "loss": 0.6884,
+      "step": 31315
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9391566157530404e-05,
+      "loss": 0.6701,
+      "step": 31320
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9391182417788816e-05,
+      "loss": 0.6781,
+      "step": 31325
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9390798560872233e-05,
+      "loss": 0.6824,
+      "step": 31330
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9390414586785437e-05,
+      "loss": 0.6842,
+      "step": 31335
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9390030495533225e-05,
+      "loss": 0.6822,
+      "step": 31340
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9389646287120385e-05,
+      "loss": 0.6862,
+      "step": 31345
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.938926196155171e-05,
+      "loss": 0.6875,
+      "step": 31350
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9388877518832e-05,
+      "loss": 0.6818,
+      "step": 31355
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.938849295896605e-05,
+      "loss": 0.6905,
+      "step": 31360
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9388108281958654e-05,
+      "loss": 0.6978,
+      "step": 31365
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.938772348781462e-05,
+      "loss": 0.6946,
+      "step": 31370
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9387338576538743e-05,
+      "loss": 0.6868,
+      "step": 31375
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9386953548135824e-05,
+      "loss": 0.695,
+      "step": 31380
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9386568402610675e-05,
+      "loss": 0.6934,
+      "step": 31385
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9386183139968093e-05,
+      "loss": 0.6895,
+      "step": 31390
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.938579776021289e-05,
+      "loss": 0.6888,
+      "step": 31395
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.938541226334987e-05,
+      "loss": 0.6932,
+      "step": 31400
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9385026649383846e-05,
+      "loss": 0.696,
+      "step": 31405
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9384640918319633e-05,
+      "loss": 0.6908,
+      "step": 31410
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9384255070162036e-05,
+      "loss": 0.6927,
+      "step": 31415
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9383869104915877e-05,
+      "loss": 0.6782,
+      "step": 31420
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9383483022585966e-05,
+      "loss": 0.6727,
+      "step": 31425
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.938309682317712e-05,
+      "loss": 0.6922,
+      "step": 31430
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.938271050669416e-05,
+      "loss": 0.685,
+      "step": 31435
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.938232407314191e-05,
+      "loss": 0.6733,
+      "step": 31440
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9381937522525185e-05,
+      "loss": 0.6784,
+      "step": 31445
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9381550854848812e-05,
+      "loss": 0.6762,
+      "step": 31450
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.938116407011761e-05,
+      "loss": 0.6848,
+      "step": 31455
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9380777168336414e-05,
+      "loss": 0.6775,
+      "step": 31460
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9380390149510044e-05,
+      "loss": 0.6827,
+      "step": 31465
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9380003013643328e-05,
+      "loss": 0.6839,
+      "step": 31470
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9379615760741108e-05,
+      "loss": 0.688,
+      "step": 31475
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.93792283908082e-05,
+      "loss": 0.6985,
+      "step": 31480
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.937884090384945e-05,
+      "loss": 0.6826,
+      "step": 31485
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9378453299869683e-05,
+      "loss": 0.6825,
+      "step": 31490
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9378065578873742e-05,
+      "loss": 0.6685,
+      "step": 31495
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.937767774086646e-05,
+      "loss": 0.6808,
+      "step": 31500
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.937728978585268e-05,
+      "loss": 0.684,
+      "step": 31505
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.937690171383724e-05,
+      "loss": 0.6835,
+      "step": 31510
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9376513524824987e-05,
+      "loss": 0.6993,
+      "step": 31515
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9376125218820758e-05,
+      "loss": 0.6895,
+      "step": 31520
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9375736795829402e-05,
+      "loss": 0.6631,
+      "step": 31525
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.937534825585576e-05,
+      "loss": 0.6862,
+      "step": 31530
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9374959598904686e-05,
+      "loss": 0.6832,
+      "step": 31535
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9374570824981028e-05,
+      "loss": 0.6863,
+      "step": 31540
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9374181934089635e-05,
+      "loss": 0.6895,
+      "step": 31545
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9373792926235358e-05,
+      "loss": 0.6817,
+      "step": 31550
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9373403801423054e-05,
+      "loss": 0.6859,
+      "step": 31555
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9373014559657576e-05,
+      "loss": 0.6856,
+      "step": 31560
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9372625200943784e-05,
+      "loss": 0.6728,
+      "step": 31565
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9372235725286533e-05,
+      "loss": 0.684,
+      "step": 31570
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.937184613269068e-05,
+      "loss": 0.674,
+      "step": 31575
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9371456423161095e-05,
+      "loss": 0.6892,
+      "step": 31580
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.937106659670263e-05,
+      "loss": 0.6801,
+      "step": 31585
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9370676653320153e-05,
+      "loss": 0.7087,
+      "step": 31590
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9370286593018532e-05,
+      "loss": 0.6894,
+      "step": 31595
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.936989641580263e-05,
+      "loss": 0.6772,
+      "step": 31600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.936950612167732e-05,
+      "loss": 0.6819,
+      "step": 31605
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9369115710647465e-05,
+      "loss": 0.6801,
+      "step": 31610
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9368725182717944e-05,
+      "loss": 0.6776,
+      "step": 31615
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.936833453789362e-05,
+      "loss": 0.6885,
+      "step": 31620
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.936794377617938e-05,
+      "loss": 0.6825,
+      "step": 31625
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.936755289758009e-05,
+      "loss": 0.6824,
+      "step": 31630
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9367161902100626e-05,
+      "loss": 0.6812,
+      "step": 31635
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9366770789745873e-05,
+      "loss": 0.6912,
+      "step": 31640
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9366379560520706e-05,
+      "loss": 0.6877,
+      "step": 31645
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.936598821443001e-05,
+      "loss": 0.6915,
+      "step": 31650
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9365596751478665e-05,
+      "loss": 0.6986,
+      "step": 31655
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9365205171671558e-05,
+      "loss": 0.6947,
+      "step": 31660
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.936481347501357e-05,
+      "loss": 0.6864,
+      "step": 31665
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9364421661509592e-05,
+      "loss": 0.6901,
+      "step": 31670
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9364029731164517e-05,
+      "loss": 0.6839,
+      "step": 31675
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9363637683983225e-05,
+      "loss": 0.6925,
+      "step": 31680
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9363245519970613e-05,
+      "loss": 0.6788,
+      "step": 31685
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.936285323913158e-05,
+      "loss": 0.6803,
+      "step": 31690
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.936246084147101e-05,
+      "loss": 0.6788,
+      "step": 31695
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9362068326993804e-05,
+      "loss": 0.6836,
+      "step": 31700
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.936167569570486e-05,
+      "loss": 0.6882,
+      "step": 31705
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9361282947609075e-05,
+      "loss": 0.6887,
+      "step": 31710
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9360890082711353e-05,
+      "loss": 0.6972,
+      "step": 31715
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.936049710101659e-05,
+      "loss": 0.683,
+      "step": 31720
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9360104002529693e-05,
+      "loss": 0.687,
+      "step": 31725
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9359710787255567e-05,
+      "loss": 0.6935,
+      "step": 31730
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9359317455199117e-05,
+      "loss": 0.6844,
+      "step": 31735
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.935892400636525e-05,
+      "loss": 0.6733,
+      "step": 31740
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.935853044075888e-05,
+      "loss": 0.6823,
+      "step": 31745
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.935813675838491e-05,
+      "loss": 0.6806,
+      "step": 31750
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9357742959248256e-05,
+      "loss": 0.6826,
+      "step": 31755
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9357349043353832e-05,
+      "loss": 0.6811,
+      "step": 31760
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9356955010706552e-05,
+      "loss": 0.6884,
+      "step": 31765
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9356560861311332e-05,
+      "loss": 0.6902,
+      "step": 31770
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9356166595173094e-05,
+      "loss": 0.6713,
+      "step": 31775
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9355772212296752e-05,
+      "loss": 0.6848,
+      "step": 31780
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9355377712687228e-05,
+      "loss": 0.6796,
+      "step": 31785
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9354983096349446e-05,
+      "loss": 0.6766,
+      "step": 31790
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9354588363288324e-05,
+      "loss": 0.6825,
+      "step": 31795
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9354193513508794e-05,
+      "loss": 0.6849,
+      "step": 31800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9353798547015784e-05,
+      "loss": 0.6925,
+      "step": 31805
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.935340346381421e-05,
+      "loss": 0.6861,
+      "step": 31810
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9353008263909016e-05,
+      "loss": 0.6938,
+      "step": 31815
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.935261294730513e-05,
+      "loss": 0.6858,
+      "step": 31820
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.935221751400747e-05,
+      "loss": 0.6848,
+      "step": 31825
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9351821964020988e-05,
+      "loss": 0.6777,
+      "step": 31830
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.935142629735061e-05,
+      "loss": 0.6827,
+      "step": 31835
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.935103051400128e-05,
+      "loss": 0.6865,
+      "step": 31840
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9350634613977924e-05,
+      "loss": 0.6876,
+      "step": 31845
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9350238597285497e-05,
+      "loss": 0.682,
+      "step": 31850
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9349842463928924e-05,
+      "loss": 0.6906,
+      "step": 31855
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.934944621391316e-05,
+      "loss": 0.7,
+      "step": 31860
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9349049847243145e-05,
+      "loss": 0.6752,
+      "step": 31865
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9348653363923828e-05,
+      "loss": 0.6782,
+      "step": 31870
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9348256763960146e-05,
+      "loss": 0.6798,
+      "step": 31875
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9347860047357055e-05,
+      "loss": 0.6724,
+      "step": 31880
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9347463214119507e-05,
+      "loss": 0.684,
+      "step": 31885
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9347066264252447e-05,
+      "loss": 0.6897,
+      "step": 31890
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.934666919776083e-05,
+      "loss": 0.6765,
+      "step": 31895
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9346272014649613e-05,
+      "loss": 0.6833,
+      "step": 31900
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.934587471492375e-05,
+      "loss": 0.6889,
+      "step": 31905
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9345477298588194e-05,
+      "loss": 0.6793,
+      "step": 31910
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9345079765647912e-05,
+      "loss": 0.6881,
+      "step": 31915
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9344682116107856e-05,
+      "loss": 0.685,
+      "step": 31920
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9344284349972994e-05,
+      "loss": 0.6832,
+      "step": 31925
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9343886467248288e-05,
+      "loss": 0.6885,
+      "step": 31930
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9343488467938696e-05,
+      "loss": 0.6735,
+      "step": 31935
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.934309035204919e-05,
+      "loss": 0.6855,
+      "step": 31940
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9342692119584737e-05,
+      "loss": 0.6837,
+      "step": 31945
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.93422937705503e-05,
+      "loss": 0.6872,
+      "step": 31950
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.934189530495086e-05,
+      "loss": 0.6942,
+      "step": 31955
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.934149672279138e-05,
+      "loss": 0.6826,
+      "step": 31960
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9341098024076835e-05,
+      "loss": 0.6897,
+      "step": 31965
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.93406992088122e-05,
+      "loss": 0.6812,
+      "step": 31970
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9340300277002452e-05,
+      "loss": 0.6858,
+      "step": 31975
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9339901228652567e-05,
+      "loss": 0.6957,
+      "step": 31980
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9339502063767526e-05,
+      "loss": 0.6716,
+      "step": 31985
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9339102782352307e-05,
+      "loss": 0.6878,
+      "step": 31990
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9338703384411893e-05,
+      "loss": 0.6907,
+      "step": 31995
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.933830386995127e-05,
+      "loss": 0.692,
+      "step": 32000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.933790423897542e-05,
+      "loss": 0.6958,
+      "step": 32005
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9337504491489326e-05,
+      "loss": 0.6694,
+      "step": 32010
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.933710462749798e-05,
+      "loss": 0.6838,
+      "step": 32015
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9336704647006376e-05,
+      "loss": 0.6781,
+      "step": 32020
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9336304550019493e-05,
+      "loss": 0.6783,
+      "step": 32025
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.933590433654233e-05,
+      "loss": 0.6837,
+      "step": 32030
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9335504006579883e-05,
+      "loss": 0.6777,
+      "step": 32035
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9335103560137137e-05,
+      "loss": 0.6808,
+      "step": 32040
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.93347029972191e-05,
+      "loss": 0.6931,
+      "step": 32045
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9334302317830764e-05,
+      "loss": 0.6804,
+      "step": 32050
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.933390152197713e-05,
+      "loss": 0.6816,
+      "step": 32055
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9333500609663196e-05,
+      "loss": 0.6921,
+      "step": 32060
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.933309958089397e-05,
+      "loss": 0.6805,
+      "step": 32065
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.933269843567445e-05,
+      "loss": 0.6727,
+      "step": 32070
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.933229717400964e-05,
+      "loss": 0.6918,
+      "step": 32075
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.933189579590455e-05,
+      "loss": 0.6958,
+      "step": 32080
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9331494301364192e-05,
+      "loss": 0.6769,
+      "step": 32085
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9331092690393567e-05,
+      "loss": 0.6769,
+      "step": 32090
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9330690962997694e-05,
+      "loss": 0.6816,
+      "step": 32095
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9330289119181578e-05,
+      "loss": 0.6988,
+      "step": 32100
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.932988715895024e-05,
+      "loss": 0.6831,
+      "step": 32105
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.932948508230869e-05,
+      "loss": 0.6662,
+      "step": 32110
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.932908288926194e-05,
+      "loss": 0.6846,
+      "step": 32115
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9328680579815024e-05,
+      "loss": 0.6859,
+      "step": 32120
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9328278153972947e-05,
+      "loss": 0.6895,
+      "step": 32125
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9327875611740737e-05,
+      "loss": 0.6774,
+      "step": 32130
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9327472953123415e-05,
+      "loss": 0.6746,
+      "step": 32135
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9327070178126005e-05,
+      "loss": 0.6835,
+      "step": 32140
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9326667286753533e-05,
+      "loss": 0.6851,
+      "step": 32145
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9326264279011026e-05,
+      "loss": 0.6732,
+      "step": 32150
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.932586115490351e-05,
+      "loss": 0.6723,
+      "step": 32155
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9325457914436016e-05,
+      "loss": 0.679,
+      "step": 32160
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.932505455761358e-05,
+      "loss": 0.6773,
+      "step": 32165
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9324651084441227e-05,
+      "loss": 0.6783,
+      "step": 32170
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9324247494923996e-05,
+      "loss": 0.6771,
+      "step": 32175
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9323843789066922e-05,
+      "loss": 0.6846,
+      "step": 32180
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9323439966875043e-05,
+      "loss": 0.6769,
+      "step": 32185
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.932303602835339e-05,
+      "loss": 0.6824,
+      "step": 32190
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9322631973507016e-05,
+      "loss": 0.678,
+      "step": 32195
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9322227802340955e-05,
+      "loss": 0.6741,
+      "step": 32200
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9321823514860247e-05,
+      "loss": 0.6738,
+      "step": 32205
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9321419111069943e-05,
+      "loss": 0.6847,
+      "step": 32210
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9321014590975082e-05,
+      "loss": 0.6874,
+      "step": 32215
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9320609954580714e-05,
+      "loss": 0.6777,
+      "step": 32220
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9320205201891894e-05,
+      "loss": 0.6855,
+      "step": 32225
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9319800332913667e-05,
+      "loss": 0.687,
+      "step": 32230
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9319395347651076e-05,
+      "loss": 0.6915,
+      "step": 32235
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.931899024610919e-05,
+      "loss": 0.6719,
+      "step": 32240
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9318585028293056e-05,
+      "loss": 0.676,
+      "step": 32245
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9318179694207726e-05,
+      "loss": 0.6831,
+      "step": 32250
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9317774243858264e-05,
+      "loss": 0.6886,
+      "step": 32255
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9317368677249724e-05,
+      "loss": 0.6721,
+      "step": 32260
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9316962994387167e-05,
+      "loss": 0.688,
+      "step": 32265
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.931655719527566e-05,
+      "loss": 0.6856,
+      "step": 32270
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.931615127992026e-05,
+      "loss": 0.6843,
+      "step": 32275
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9315745248326037e-05,
+      "loss": 0.6827,
+      "step": 32280
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9315339100498048e-05,
+      "loss": 0.689,
+      "step": 32285
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9314932836441374e-05,
+      "loss": 0.6934,
+      "step": 32290
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.931452645616107e-05,
+      "loss": 0.6879,
+      "step": 32295
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9314119959662218e-05,
+      "loss": 0.6777,
+      "step": 32300
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9313713346949886e-05,
+      "loss": 0.6862,
+      "step": 32305
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9313306618029143e-05,
+      "loss": 0.6679,
+      "step": 32310
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.931289977290507e-05,
+      "loss": 0.6974,
+      "step": 32315
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9312492811582735e-05,
+      "loss": 0.6805,
+      "step": 32320
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9312085734067226e-05,
+      "loss": 0.688,
+      "step": 32325
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9311678540363615e-05,
+      "loss": 0.6739,
+      "step": 32330
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9311271230476987e-05,
+      "loss": 0.6814,
+      "step": 32335
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.931086380441242e-05,
+      "loss": 0.6824,
+      "step": 32340
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9310456262175004e-05,
+      "loss": 0.6729,
+      "step": 32345
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9310048603769816e-05,
+      "loss": 0.678,
+      "step": 32350
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9309640829201947e-05,
+      "loss": 0.6699,
+      "step": 32355
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9309232938476482e-05,
+      "loss": 0.682,
+      "step": 32360
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9308824931598514e-05,
+      "loss": 0.6938,
+      "step": 32365
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.930841680857313e-05,
+      "loss": 0.6898,
+      "step": 32370
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9308008569405424e-05,
+      "loss": 0.6847,
+      "step": 32375
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.930760021410049e-05,
+      "loss": 0.6849,
+      "step": 32380
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9307191742663424e-05,
+      "loss": 0.6733,
+      "step": 32385
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9306783155099322e-05,
+      "loss": 0.6895,
+      "step": 32390
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9306374451413278e-05,
+      "loss": 0.684,
+      "step": 32395
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9305965631610397e-05,
+      "loss": 0.6697,
+      "step": 32400
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9305556695695776e-05,
+      "loss": 0.676,
+      "step": 32405
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9305147643674524e-05,
+      "loss": 0.6787,
+      "step": 32410
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9304738475551733e-05,
+      "loss": 0.6869,
+      "step": 32415
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9304329191332522e-05,
+      "loss": 0.6889,
+      "step": 32420
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9303919791021984e-05,
+      "loss": 0.6851,
+      "step": 32425
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.930351027462524e-05,
+      "loss": 0.6723,
+      "step": 32430
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9303100642147388e-05,
+      "loss": 0.6903,
+      "step": 32435
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.930269089359355e-05,
+      "loss": 0.6838,
+      "step": 32440
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9302281028968826e-05,
+      "loss": 0.6872,
+      "step": 32445
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9301871048278343e-05,
+      "loss": 0.6717,
+      "step": 32450
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.930146095152721e-05,
+      "loss": 0.6939,
+      "step": 32455
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.930105073872054e-05,
+      "loss": 0.6796,
+      "step": 32460
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9300640409863457e-05,
+      "loss": 0.6664,
+      "step": 32465
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.930022996496108e-05,
+      "loss": 0.6818,
+      "step": 32470
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.929981940401853e-05,
+      "loss": 0.6814,
+      "step": 32475
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.929940872704093e-05,
+      "loss": 0.6836,
+      "step": 32480
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.92989979340334e-05,
+      "loss": 0.6748,
+      "step": 32485
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.929858702500107e-05,
+      "loss": 0.6904,
+      "step": 32490
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9298175999949068e-05,
+      "loss": 0.6793,
+      "step": 32495
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9297764858882516e-05,
+      "loss": 0.6823,
+      "step": 32500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.929735360180655e-05,
+      "loss": 0.6856,
+      "step": 32505
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9296942228726295e-05,
+      "loss": 0.6829,
+      "step": 32510
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9296530739646892e-05,
+      "loss": 0.6843,
+      "step": 32515
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9296119134573473e-05,
+      "loss": 0.6888,
+      "step": 32520
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9295707413511166e-05,
+      "loss": 0.6826,
+      "step": 32525
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.929529557646512e-05,
+      "loss": 0.658,
+      "step": 32530
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.929488362344046e-05,
+      "loss": 0.6874,
+      "step": 32535
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9294471554442336e-05,
+      "loss": 0.67,
+      "step": 32540
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.929405936947589e-05,
+      "loss": 0.6766,
+      "step": 32545
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9293647068546263e-05,
+      "loss": 0.6709,
+      "step": 32550
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.929323465165859e-05,
+      "loss": 0.6822,
+      "step": 32555
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.929282211881803e-05,
+      "loss": 0.6848,
+      "step": 32560
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9292409470029726e-05,
+      "loss": 0.6825,
+      "step": 32565
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9291996705298826e-05,
+      "loss": 0.6763,
+      "step": 32570
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9291583824630478e-05,
+      "loss": 0.6865,
+      "step": 32575
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9291170828029835e-05,
+      "loss": 0.6802,
+      "step": 32580
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.929075771550205e-05,
+      "loss": 0.6845,
+      "step": 32585
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9290344487052277e-05,
+      "loss": 0.6789,
+      "step": 32590
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9289931142685673e-05,
+      "loss": 0.6883,
+      "step": 32595
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9289517682407397e-05,
+      "loss": 0.6732,
+      "step": 32600
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9289104106222606e-05,
+      "loss": 0.6758,
+      "step": 32605
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9288690414136458e-05,
+      "loss": 0.6863,
+      "step": 32610
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9288276606154118e-05,
+      "loss": 0.6791,
+      "step": 32615
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9287862682280748e-05,
+      "loss": 0.6893,
+      "step": 32620
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9287448642521513e-05,
+      "loss": 0.6821,
+      "step": 32625
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.928703448688158e-05,
+      "loss": 0.6711,
+      "step": 32630
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.928662021536611e-05,
+      "loss": 0.6815,
+      "step": 32635
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.928620582798028e-05,
+      "loss": 0.6774,
+      "step": 32640
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9285791324729257e-05,
+      "loss": 0.6781,
+      "step": 32645
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9285376705618216e-05,
+      "loss": 0.6758,
+      "step": 32650
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9284961970652323e-05,
+      "loss": 0.6797,
+      "step": 32655
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.928454711983676e-05,
+      "loss": 0.688,
+      "step": 32660
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.92841321531767e-05,
+      "loss": 0.6763,
+      "step": 32665
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.928371707067732e-05,
+      "loss": 0.6864,
+      "step": 32670
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9283301872343798e-05,
+      "loss": 0.6794,
+      "step": 32675
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.928288655818132e-05,
+      "loss": 0.6819,
+      "step": 32680
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9282471128195063e-05,
+      "loss": 0.6812,
+      "step": 32685
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9282055582390208e-05,
+      "loss": 0.6842,
+      "step": 32690
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9281639920771948e-05,
+      "loss": 0.6898,
+      "step": 32695
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.928122414334546e-05,
+      "loss": 0.6829,
+      "step": 32700
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.928080825011594e-05,
+      "loss": 0.6873,
+      "step": 32705
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9280392241088576e-05,
+      "loss": 0.6806,
+      "step": 32710
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.927997611626855e-05,
+      "loss": 0.6766,
+      "step": 32715
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9279559875661067e-05,
+      "loss": 0.6757,
+      "step": 32720
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.927914351927131e-05,
+      "loss": 0.6887,
+      "step": 32725
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9278727047104477e-05,
+      "loss": 0.6716,
+      "step": 32730
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9278310459165765e-05,
+      "loss": 0.6838,
+      "step": 32735
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9277893755460373e-05,
+      "loss": 0.6891,
+      "step": 32740
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9277476935993498e-05,
+      "loss": 0.6856,
+      "step": 32745
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9277060000770342e-05,
+      "loss": 0.6721,
+      "step": 32750
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.927664294979611e-05,
+      "loss": 0.6774,
+      "step": 32755
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9276225783076002e-05,
+      "loss": 0.6924,
+      "step": 32760
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.927580850061522e-05,
+      "loss": 0.6899,
+      "step": 32765
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9275391102418976e-05,
+      "loss": 0.6789,
+      "step": 32770
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9274973588492475e-05,
+      "loss": 0.681,
+      "step": 32775
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.927455595884093e-05,
+      "loss": 0.6982,
+      "step": 32780
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9274138213469548e-05,
+      "loss": 0.6736,
+      "step": 32785
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9273720352383544e-05,
+      "loss": 0.6877,
+      "step": 32790
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9273302375588127e-05,
+      "loss": 0.6825,
+      "step": 32795
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9272884283088517e-05,
+      "loss": 0.675,
+      "step": 32800
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.927246607488993e-05,
+      "loss": 0.6817,
+      "step": 32805
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.927204775099758e-05,
+      "loss": 0.6851,
+      "step": 32810
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9271629311416687e-05,
+      "loss": 0.6895,
+      "step": 32815
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9271210756152482e-05,
+      "loss": 0.6738,
+      "step": 32820
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9270792085210176e-05,
+      "loss": 0.678,
+      "step": 32825
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9270373298594998e-05,
+      "loss": 0.6671,
+      "step": 32830
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.926995439631217e-05,
+      "loss": 0.6783,
+      "step": 32835
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.926953537836692e-05,
+      "loss": 0.6799,
+      "step": 32840
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.926911624476448e-05,
+      "loss": 0.6781,
+      "step": 32845
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.926869699551007e-05,
+      "loss": 0.6712,
+      "step": 32850
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9268277630608935e-05,
+      "loss": 0.6799,
+      "step": 32855
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9267858150066297e-05,
+      "loss": 0.6779,
+      "step": 32860
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.926743855388739e-05,
+      "loss": 0.6655,
+      "step": 32865
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9267018842077455e-05,
+      "loss": 0.682,
+      "step": 32870
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9266599014641724e-05,
+      "loss": 0.6749,
+      "step": 32875
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.926617907158544e-05,
+      "loss": 0.6799,
+      "step": 32880
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9265759012913836e-05,
+      "loss": 0.687,
+      "step": 32885
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9265338838632156e-05,
+      "loss": 0.6744,
+      "step": 32890
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9264918548745645e-05,
+      "loss": 0.6783,
+      "step": 32895
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9264498143259546e-05,
+      "loss": 0.6987,
+      "step": 32900
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9264077622179105e-05,
+      "loss": 0.685,
+      "step": 32905
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.926365698550957e-05,
+      "loss": 0.689,
+      "step": 32910
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.926323623325618e-05,
+      "loss": 0.6874,
+      "step": 32915
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.92628153654242e-05,
+      "loss": 0.6793,
+      "step": 32920
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9262394382018867e-05,
+      "loss": 0.671,
+      "step": 32925
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9261973283045443e-05,
+      "loss": 0.6773,
+      "step": 32930
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9261552068509174e-05,
+      "loss": 0.6841,
+      "step": 32935
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9261130738415327e-05,
+      "loss": 0.6842,
+      "step": 32940
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9260709292769147e-05,
+      "loss": 0.6894,
+      "step": 32945
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9260287731575902e-05,
+      "loss": 0.6733,
+      "step": 32950
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.925986605484085e-05,
+      "loss": 0.6837,
+      "step": 32955
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9259444262569244e-05,
+      "loss": 0.6783,
+      "step": 32960
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9259022354766353e-05,
+      "loss": 0.6743,
+      "step": 32965
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9258600331437445e-05,
+      "loss": 0.6773,
+      "step": 32970
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.925817819258778e-05,
+      "loss": 0.6816,
+      "step": 32975
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9257755938222628e-05,
+      "loss": 0.6867,
+      "step": 32980
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9257333568347256e-05,
+      "loss": 0.681,
+      "step": 32985
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.925691108296693e-05,
+      "loss": 0.6759,
+      "step": 32990
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.925648848208693e-05,
+      "loss": 0.6882,
+      "step": 32995
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9256065765712524e-05,
+      "loss": 0.6723,
+      "step": 33000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9255642933848983e-05,
+      "loss": 0.6877,
+      "step": 33005
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.925521998650159e-05,
+      "loss": 0.6896,
+      "step": 33010
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.925479692367562e-05,
+      "loss": 0.686,
+      "step": 33015
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.925437374537635e-05,
+      "loss": 0.6838,
+      "step": 33020
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.925395045160906e-05,
+      "loss": 0.6762,
+      "step": 33025
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.925352704237903e-05,
+      "loss": 0.6875,
+      "step": 33030
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9253103517691543e-05,
+      "loss": 0.6817,
+      "step": 33035
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.925267987755189e-05,
+      "loss": 0.6772,
+      "step": 33040
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9252256121965346e-05,
+      "loss": 0.6792,
+      "step": 33045
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.925183225093721e-05,
+      "loss": 0.6783,
+      "step": 33050
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9251408264472758e-05,
+      "loss": 0.6792,
+      "step": 33055
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9250984162577297e-05,
+      "loss": 0.677,
+      "step": 33060
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.92505599452561e-05,
+      "loss": 0.6801,
+      "step": 33065
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.925013561251447e-05,
+      "loss": 0.6756,
+      "step": 33070
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9249711164357704e-05,
+      "loss": 0.6797,
+      "step": 33075
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.924928660079109e-05,
+      "loss": 0.6734,
+      "step": 33080
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9248861921819934e-05,
+      "loss": 0.6856,
+      "step": 33085
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9248437127449525e-05,
+      "loss": 0.6666,
+      "step": 33090
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.924801221768517e-05,
+      "loss": 0.6755,
+      "step": 33095
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9247587192532167e-05,
+      "loss": 0.6781,
+      "step": 33100
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9247162051995826e-05,
+      "loss": 0.6762,
+      "step": 33105
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9246736796081444e-05,
+      "loss": 0.6822,
+      "step": 33110
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9246311424794327e-05,
+      "loss": 0.6719,
+      "step": 33115
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.924588593813979e-05,
+      "loss": 0.6879,
+      "step": 33120
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9245460336123136e-05,
+      "loss": 0.6896,
+      "step": 33125
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9245034618749673e-05,
+      "loss": 0.693,
+      "step": 33130
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9244608786024715e-05,
+      "loss": 0.6702,
+      "step": 33135
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.924418283795358e-05,
+      "loss": 0.6809,
+      "step": 33140
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9243756774541574e-05,
+      "loss": 0.6712,
+      "step": 33145
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.924333059579402e-05,
+      "loss": 0.6782,
+      "step": 33150
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9242904301716235e-05,
+      "loss": 0.6795,
+      "step": 33155
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9242477892313534e-05,
+      "loss": 0.6842,
+      "step": 33160
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9242051367591238e-05,
+      "loss": 0.6734,
+      "step": 33165
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9241624727554672e-05,
+      "loss": 0.6838,
+      "step": 33170
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9241197972209157e-05,
+      "loss": 0.6816,
+      "step": 33175
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.924077110156002e-05,
+      "loss": 0.6883,
+      "step": 33180
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9240344115612584e-05,
+      "loss": 0.6796,
+      "step": 33185
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.923991701437218e-05,
+      "loss": 0.6768,
+      "step": 33190
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9239489797844132e-05,
+      "loss": 0.6741,
+      "step": 33195
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.923906246603377e-05,
+      "loss": 0.6981,
+      "step": 33200
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9238635018946438e-05,
+      "loss": 0.6699,
+      "step": 33205
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9238207456587453e-05,
+      "loss": 0.6904,
+      "step": 33210
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.923777977896216e-05,
+      "loss": 0.6775,
+      "step": 33215
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.923735198607589e-05,
+      "loss": 0.6797,
+      "step": 33220
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9236924077933988e-05,
+      "loss": 0.6699,
+      "step": 33225
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9236496054541783e-05,
+      "loss": 0.6668,
+      "step": 33230
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9236067915904624e-05,
+      "loss": 0.7048,
+      "step": 33235
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.923563966202785e-05,
+      "loss": 0.6844,
+      "step": 33240
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.92352112929168e-05,
+      "loss": 0.6733,
+      "step": 33245
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9234782808576823e-05,
+      "loss": 0.6725,
+      "step": 33250
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9234354209013268e-05,
+      "loss": 0.6737,
+      "step": 33255
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9233925494231477e-05,
+      "loss": 0.6821,
+      "step": 33260
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9233496664236804e-05,
+      "loss": 0.6763,
+      "step": 33265
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9233067719034597e-05,
+      "loss": 0.6627,
+      "step": 33270
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.923263865863021e-05,
+      "loss": 0.6798,
+      "step": 33275
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.923220948302899e-05,
+      "loss": 0.6758,
+      "step": 33280
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.92317801922363e-05,
+      "loss": 0.6782,
+      "step": 33285
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9231350786257493e-05,
+      "loss": 0.6738,
+      "step": 33290
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9230921265097928e-05,
+      "loss": 0.6659,
+      "step": 33295
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.923049162876296e-05,
+      "loss": 0.6741,
+      "step": 33300
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9230061877257957e-05,
+      "loss": 0.6728,
+      "step": 33305
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.922963201058827e-05,
+      "loss": 0.6872,
+      "step": 33310
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9229202028759273e-05,
+      "loss": 0.6714,
+      "step": 33315
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.922877193177633e-05,
+      "loss": 0.6718,
+      "step": 33320
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.92283417196448e-05,
+      "loss": 0.6691,
+      "step": 33325
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9227911392370055e-05,
+      "loss": 0.6726,
+      "step": 33330
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9227480949957466e-05,
+      "loss": 0.6766,
+      "step": 33335
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.92270503924124e-05,
+      "loss": 0.6717,
+      "step": 33340
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9226619719740235e-05,
+      "loss": 0.6833,
+      "step": 33345
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9226188931946343e-05,
+      "loss": 0.673,
+      "step": 33350
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9225758029036093e-05,
+      "loss": 0.6793,
+      "step": 33355
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.922532701101486e-05,
+      "loss": 0.6784,
+      "step": 33360
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9224895877888036e-05,
+      "loss": 0.6834,
+      "step": 33365
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.922446462966099e-05,
+      "loss": 0.6757,
+      "step": 33370
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9224033266339103e-05,
+      "loss": 0.6835,
+      "step": 33375
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9223601787927757e-05,
+      "loss": 0.6706,
+      "step": 33380
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9223170194432338e-05,
+      "loss": 0.6859,
+      "step": 33385
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.922273848585823e-05,
+      "loss": 0.6775,
+      "step": 33390
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.922230666221082e-05,
+      "loss": 0.6857,
+      "step": 33395
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9221874723495494e-05,
+      "loss": 0.6709,
+      "step": 33400
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9221442669717645e-05,
+      "loss": 0.6805,
+      "step": 33405
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9221010500882662e-05,
+      "loss": 0.6699,
+      "step": 33410
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9220578216995936e-05,
+      "loss": 0.6814,
+      "step": 33415
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9220145818062857e-05,
+      "loss": 0.6744,
+      "step": 33420
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9219713304088833e-05,
+      "loss": 0.6704,
+      "step": 33425
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9219280675079244e-05,
+      "loss": 0.6782,
+      "step": 33430
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.92188479310395e-05,
+      "loss": 0.6838,
+      "step": 33435
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9218415071974998e-05,
+      "loss": 0.6878,
+      "step": 33440
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9217982097891136e-05,
+      "loss": 0.6821,
+      "step": 33445
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9217549008793318e-05,
+      "loss": 0.6659,
+      "step": 33450
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9217115804686945e-05,
+      "loss": 0.6681,
+      "step": 33455
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9216682485577425e-05,
+      "loss": 0.6723,
+      "step": 33460
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9216249051470165e-05,
+      "loss": 0.6764,
+      "step": 33465
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9215815502370573e-05,
+      "loss": 0.6697,
+      "step": 33470
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9215381838284056e-05,
+      "loss": 0.6762,
+      "step": 33475
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.921494805921603e-05,
+      "loss": 0.6656,
+      "step": 33480
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.92145141651719e-05,
+      "loss": 0.6623,
+      "step": 33485
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9214080156157085e-05,
+      "loss": 0.669,
+      "step": 33490
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9213646032176997e-05,
+      "loss": 0.681,
+      "step": 33495
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9213211793237056e-05,
+      "loss": 0.6767,
+      "step": 33500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9212777439342675e-05,
+      "loss": 0.6893,
+      "step": 33505
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.921234297049928e-05,
+      "loss": 0.68,
+      "step": 33510
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.921190838671229e-05,
+      "loss": 0.6661,
+      "step": 33515
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9211473687987123e-05,
+      "loss": 0.6777,
+      "step": 33520
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9211038874329208e-05,
+      "loss": 0.6834,
+      "step": 33525
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9210603945743965e-05,
+      "loss": 0.6825,
+      "step": 33530
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.921016890223683e-05,
+      "loss": 0.6884,
+      "step": 33535
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9209733743813218e-05,
+      "loss": 0.6836,
+      "step": 33540
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.920929847047857e-05,
+      "loss": 0.6715,
+      "step": 33545
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.920886308223831e-05,
+      "loss": 0.686,
+      "step": 33550
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9208427579097874e-05,
+      "loss": 0.6803,
+      "step": 33555
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9207991961062693e-05,
+      "loss": 0.676,
+      "step": 33560
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9207556228138206e-05,
+      "loss": 0.6743,
+      "step": 33565
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9207120380329844e-05,
+      "loss": 0.676,
+      "step": 33570
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9206684417643052e-05,
+      "loss": 0.6755,
+      "step": 33575
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9206248340083268e-05,
+      "loss": 0.6741,
+      "step": 33580
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9205812147655926e-05,
+      "loss": 0.6819,
+      "step": 33585
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9205375840366475e-05,
+      "loss": 0.6907,
+      "step": 33590
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.920493941822036e-05,
+      "loss": 0.686,
+      "step": 33595
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9204502881223023e-05,
+      "loss": 0.6778,
+      "step": 33600
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9204066229379912e-05,
+      "loss": 0.6646,
+      "step": 33605
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9203629462696473e-05,
+      "loss": 0.685,
+      "step": 33610
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.920319258117816e-05,
+      "loss": 0.6792,
+      "step": 33615
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9202755584830418e-05,
+      "loss": 0.6722,
+      "step": 33620
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9202318473658707e-05,
+      "loss": 0.674,
+      "step": 33625
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.920188124766847e-05,
+      "loss": 0.6647,
+      "step": 33630
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9201443906865176e-05,
+      "loss": 0.6728,
+      "step": 33635
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9201006451254273e-05,
+      "loss": 0.6771,
+      "step": 33640
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9200568880841217e-05,
+      "loss": 0.6818,
+      "step": 33645
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9200131195631476e-05,
+      "loss": 0.6776,
+      "step": 33650
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9199693395630502e-05,
+      "loss": 0.6878,
+      "step": 33655
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.919925548084377e-05,
+      "loss": 0.6835,
+      "step": 33660
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9198817451276728e-05,
+      "loss": 0.6763,
+      "step": 33665
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9198379306934855e-05,
+      "loss": 0.6692,
+      "step": 33670
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9197941047823606e-05,
+      "loss": 0.683,
+      "step": 33675
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9197502673948462e-05,
+      "loss": 0.6733,
+      "step": 33680
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9197064185314882e-05,
+      "loss": 0.6719,
+      "step": 33685
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9196625581928342e-05,
+      "loss": 0.6795,
+      "step": 33690
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9196186863794316e-05,
+      "loss": 0.688,
+      "step": 33695
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9195748030918272e-05,
+      "loss": 0.6754,
+      "step": 33700
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.919530908330569e-05,
+      "loss": 0.6793,
+      "step": 33705
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.919487002096205e-05,
+      "loss": 0.6704,
+      "step": 33710
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.919443084389282e-05,
+      "loss": 0.6709,
+      "step": 33715
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9193991552103486e-05,
+      "loss": 0.6714,
+      "step": 33720
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9193552145599533e-05,
+      "loss": 0.6768,
+      "step": 33725
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9193112624386438e-05,
+      "loss": 0.6802,
+      "step": 33730
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9192672988469682e-05,
+      "loss": 0.6766,
+      "step": 33735
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9192233237854762e-05,
+      "loss": 0.6757,
+      "step": 33740
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9191793372547153e-05,
+      "loss": 0.6787,
+      "step": 33745
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9191353392552346e-05,
+      "loss": 0.6653,
+      "step": 33750
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9190913297875834e-05,
+      "loss": 0.6825,
+      "step": 33755
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.919047308852311e-05,
+      "loss": 0.6734,
+      "step": 33760
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9190032764499658e-05,
+      "loss": 0.6659,
+      "step": 33765
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.918959232581098e-05,
+      "loss": 0.6961,
+      "step": 33770
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9189151772462567e-05,
+      "loss": 0.6761,
+      "step": 33775
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.918871110445992e-05,
+      "loss": 0.678,
+      "step": 33780
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.918827032180853e-05,
+      "loss": 0.6886,
+      "step": 33785
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9187829424513908e-05,
+      "loss": 0.6789,
+      "step": 33790
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.918738841258154e-05,
+      "loss": 0.6771,
+      "step": 33795
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.918694728601694e-05,
+      "loss": 0.6792,
+      "step": 33800
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9186506044825614e-05,
+      "loss": 0.6757,
+      "step": 33805
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.918606468901306e-05,
+      "loss": 0.674,
+      "step": 33810
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9185623218584785e-05,
+      "loss": 0.6677,
+      "step": 33815
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.91851816335463e-05,
+      "loss": 0.6793,
+      "step": 33820
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9184739933903114e-05,
+      "loss": 0.6618,
+      "step": 33825
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.918429811966074e-05,
+      "loss": 0.6858,
+      "step": 33830
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.918385619082469e-05,
+      "loss": 0.6819,
+      "step": 33835
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9183414147400475e-05,
+      "loss": 0.6744,
+      "step": 33840
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9182971989393612e-05,
+      "loss": 0.6756,
+      "step": 33845
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9182529716809618e-05,
+      "loss": 0.6863,
+      "step": 33850
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9182087329654013e-05,
+      "loss": 0.6755,
+      "step": 33855
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9181644827932317e-05,
+      "loss": 0.6854,
+      "step": 33860
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9181202211650044e-05,
+      "loss": 0.6721,
+      "step": 33865
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.918075948081273e-05,
+      "loss": 0.6893,
+      "step": 33870
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9180316635425883e-05,
+      "loss": 0.676,
+      "step": 33875
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.917987367549504e-05,
+      "loss": 0.6839,
+      "step": 33880
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9179430601025723e-05,
+      "loss": 0.6763,
+      "step": 33885
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9178987412023465e-05,
+      "loss": 0.6778,
+      "step": 33890
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9178544108493787e-05,
+      "loss": 0.6808,
+      "step": 33895
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9178100690442233e-05,
+      "loss": 0.6744,
+      "step": 33900
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9177657157874324e-05,
+      "loss": 0.6744,
+      "step": 33905
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9177213510795594e-05,
+      "loss": 0.6747,
+      "step": 33910
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9176769749211586e-05,
+      "loss": 0.6786,
+      "step": 33915
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9176325873127837e-05,
+      "loss": 0.6793,
+      "step": 33920
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9175881882549877e-05,
+      "loss": 0.676,
+      "step": 33925
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9175437777483252e-05,
+      "loss": 0.6843,
+      "step": 33930
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9174993557933504e-05,
+      "loss": 0.6809,
+      "step": 33935
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9174549223906168e-05,
+      "loss": 0.6763,
+      "step": 33940
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.91741047754068e-05,
+      "loss": 0.6707,
+      "step": 33945
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.917366021244093e-05,
+      "loss": 0.6744,
+      "step": 33950
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9173215535014123e-05,
+      "loss": 0.6882,
+      "step": 33955
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9172770743131913e-05,
+      "loss": 0.6886,
+      "step": 33960
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9172325836799852e-05,
+      "loss": 0.6636,
+      "step": 33965
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.91718808160235e-05,
+      "loss": 0.668,
+      "step": 33970
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9171435680808396e-05,
+      "loss": 0.6654,
+      "step": 33975
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9170990431160105e-05,
+      "loss": 0.6808,
+      "step": 33980
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9170545067084178e-05,
+      "loss": 0.6754,
+      "step": 33985
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9170099588586175e-05,
+      "loss": 0.6743,
+      "step": 33990
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.916965399567165e-05,
+      "loss": 0.6851,
+      "step": 33995
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9169208288346168e-05,
+      "loss": 0.6713,
+      "step": 34000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.916876246661528e-05,
+      "loss": 0.6775,
+      "step": 34005
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9168316530484563e-05,
+      "loss": 0.6735,
+      "step": 34010
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9167870479959566e-05,
+      "loss": 0.6746,
+      "step": 34015
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9167424315045866e-05,
+      "loss": 0.6726,
+      "step": 34020
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9166978035749023e-05,
+      "loss": 0.6856,
+      "step": 34025
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.916653164207461e-05,
+      "loss": 0.6901,
+      "step": 34030
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9166085134028197e-05,
+      "loss": 0.6665,
+      "step": 34035
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9165638511615346e-05,
+      "loss": 0.6872,
+      "step": 34040
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.916519177484164e-05,
+      "loss": 0.6865,
+      "step": 34045
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.916474492371265e-05,
+      "loss": 0.6835,
+      "step": 34050
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9164297958233954e-05,
+      "loss": 0.6763,
+      "step": 34055
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9163850878411118e-05,
+      "loss": 0.6885,
+      "step": 34060
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9163403684249734e-05,
+      "loss": 0.6733,
+      "step": 34065
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.916295637575537e-05,
+      "loss": 0.6922,
+      "step": 34070
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.916250895293362e-05,
+      "loss": 0.6781,
+      "step": 34075
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9162061415790053e-05,
+      "loss": 0.6725,
+      "step": 34080
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9161613764330265e-05,
+      "loss": 0.687,
+      "step": 34085
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.916116599855983e-05,
+      "loss": 0.6803,
+      "step": 34090
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.916071811848434e-05,
+      "loss": 0.6862,
+      "step": 34095
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9160270124109386e-05,
+      "loss": 0.6814,
+      "step": 34100
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.915982201544056e-05,
+      "loss": 0.6788,
+      "step": 34105
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9159373792483444e-05,
+      "loss": 0.6795,
+      "step": 34110
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9158925455243637e-05,
+      "loss": 0.6764,
+      "step": 34115
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9158477003726728e-05,
+      "loss": 0.6816,
+      "step": 34120
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9158028437938316e-05,
+      "loss": 0.6819,
+      "step": 34125
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9157579757883997e-05,
+      "loss": 0.6809,
+      "step": 34130
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9157130963569372e-05,
+      "loss": 0.6795,
+      "step": 34135
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9156682055000034e-05,
+      "loss": 0.6698,
+      "step": 34140
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9156233032181592e-05,
+      "loss": 0.6812,
+      "step": 34145
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9155783895119642e-05,
+      "loss": 0.6757,
+      "step": 34150
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.915533464381979e-05,
+      "loss": 0.6685,
+      "step": 34155
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9154885278287648e-05,
+      "loss": 0.6644,
+      "step": 34160
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.915443579852881e-05,
+      "loss": 0.672,
+      "step": 34165
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9153986204548897e-05,
+      "loss": 0.6875,
+      "step": 34170
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9153536496353505e-05,
+      "loss": 0.6897,
+      "step": 34175
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.915308667394826e-05,
+      "loss": 0.6802,
+      "step": 34180
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9152636737338764e-05,
+      "loss": 0.6751,
+      "step": 34185
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9152186686530634e-05,
+      "loss": 0.6738,
+      "step": 34190
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9151736521529485e-05,
+      "loss": 0.665,
+      "step": 34195
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9151286242340936e-05,
+      "loss": 0.6766,
+      "step": 34200
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9150835848970605e-05,
+      "loss": 0.6779,
+      "step": 34205
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9150385341424105e-05,
+      "loss": 0.6758,
+      "step": 34210
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9149934719707067e-05,
+      "loss": 0.6727,
+      "step": 34215
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9149483983825106e-05,
+      "loss": 0.6652,
+      "step": 34220
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.914903313378385e-05,
+      "loss": 0.6749,
+      "step": 34225
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.914858216958892e-05,
+      "loss": 0.6766,
+      "step": 34230
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.914813109124595e-05,
+      "loss": 0.6817,
+      "step": 34235
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9147679898760557e-05,
+      "loss": 0.6727,
+      "step": 34240
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9147228592138384e-05,
+      "loss": 0.6714,
+      "step": 34245
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.914677717138505e-05,
+      "loss": 0.6567,
+      "step": 34250
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9146325636506196e-05,
+      "loss": 0.6815,
+      "step": 34255
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.914587398750745e-05,
+      "loss": 0.674,
+      "step": 34260
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9145422224394452e-05,
+      "loss": 0.6785,
+      "step": 34265
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9144970347172837e-05,
+      "loss": 0.6736,
+      "step": 34270
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9144518355848243e-05,
+      "loss": 0.6772,
+      "step": 34275
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9144066250426306e-05,
+      "loss": 0.6819,
+      "step": 34280
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9143614030912673e-05,
+      "loss": 0.6694,
+      "step": 34285
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9143161697312986e-05,
+      "loss": 0.6745,
+      "step": 34290
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.914270924963288e-05,
+      "loss": 0.6665,
+      "step": 34295
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9142256687878012e-05,
+      "loss": 0.6682,
+      "step": 34300
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.914180401205402e-05,
+      "loss": 0.6668,
+      "step": 34305
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9141351222166556e-05,
+      "loss": 0.6707,
+      "step": 34310
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.914089831822127e-05,
+      "loss": 0.6778,
+      "step": 34315
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.914044530022381e-05,
+      "loss": 0.675,
+      "step": 34320
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9139992168179836e-05,
+      "loss": 0.6862,
+      "step": 34325
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.913953892209499e-05,
+      "loss": 0.6798,
+      "step": 34330
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9139085561974936e-05,
+      "loss": 0.6787,
+      "step": 34335
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9138632087825327e-05,
+      "loss": 0.6798,
+      "step": 34340
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9138178499651826e-05,
+      "loss": 0.6783,
+      "step": 34345
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9137724797460084e-05,
+      "loss": 0.68,
+      "step": 34350
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9137270981255765e-05,
+      "loss": 0.6732,
+      "step": 34355
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9136817051044536e-05,
+      "loss": 0.6602,
+      "step": 34360
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9136363006832057e-05,
+      "loss": 0.6773,
+      "step": 34365
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9135908848623993e-05,
+      "loss": 0.668,
+      "step": 34370
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.913545457642601e-05,
+      "loss": 0.6753,
+      "step": 34375
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.913500019024378e-05,
+      "loss": 0.6783,
+      "step": 34380
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9134545690082968e-05,
+      "loss": 0.6668,
+      "step": 34385
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9134091075949242e-05,
+      "loss": 0.6866,
+      "step": 34390
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9133636347848287e-05,
+      "loss": 0.6696,
+      "step": 34395
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.913318150578576e-05,
+      "loss": 0.6677,
+      "step": 34400
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9132726549767353e-05,
+      "loss": 0.6649,
+      "step": 34405
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.913227147979873e-05,
+      "loss": 0.6628,
+      "step": 34410
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9131816295885572e-05,
+      "loss": 0.6672,
+      "step": 34415
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.913136099803356e-05,
+      "loss": 0.6627,
+      "step": 34420
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.913090558624838e-05,
+      "loss": 0.664,
+      "step": 34425
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9130450060535702e-05,
+      "loss": 0.6744,
+      "step": 34430
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9129994420901217e-05,
+      "loss": 0.6707,
+      "step": 34435
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9129538667350613e-05,
+      "loss": 0.674,
+      "step": 34440
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.912908279988957e-05,
+      "loss": 0.6842,
+      "step": 34445
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9128626818523776e-05,
+      "loss": 0.6849,
+      "step": 34450
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9128170723258925e-05,
+      "loss": 0.677,
+      "step": 34455
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.912771451410071e-05,
+      "loss": 0.6684,
+      "step": 34460
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9127258191054814e-05,
+      "loss": 0.6778,
+      "step": 34465
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9126801754126935e-05,
+      "loss": 0.6688,
+      "step": 34470
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.912634520332277e-05,
+      "loss": 0.6732,
+      "step": 34475
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9125888538648017e-05,
+      "loss": 0.6719,
+      "step": 34480
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9125431760108366e-05,
+      "loss": 0.6831,
+      "step": 34485
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9124974867709526e-05,
+      "loss": 0.6768,
+      "step": 34490
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.912451786145719e-05,
+      "loss": 0.6741,
+      "step": 34495
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9124060741357065e-05,
+      "loss": 0.6818,
+      "step": 34500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.912360350741485e-05,
+      "loss": 0.664,
+      "step": 34505
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9123146159636253e-05,
+      "loss": 0.6627,
+      "step": 34510
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.912268869802698e-05,
+      "loss": 0.6754,
+      "step": 34515
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.912223112259274e-05,
+      "loss": 0.6827,
+      "step": 34520
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.912177343333924e-05,
+      "loss": 0.6711,
+      "step": 34525
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9121315630272192e-05,
+      "loss": 0.6721,
+      "step": 34530
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.912085771339731e-05,
+      "loss": 0.6675,
+      "step": 34535
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.91203996827203e-05,
+      "loss": 0.6718,
+      "step": 34540
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.911994153824688e-05,
+      "loss": 0.6752,
+      "step": 34545
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.911948327998278e-05,
+      "loss": 0.6821,
+      "step": 34550
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9119024907933698e-05,
+      "loss": 0.6671,
+      "step": 34555
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.911856642210536e-05,
+      "loss": 0.6768,
+      "step": 34560
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.911810782250349e-05,
+      "loss": 0.6847,
+      "step": 34565
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9117649109133807e-05,
+      "loss": 0.677,
+      "step": 34570
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.911719028200204e-05,
+      "loss": 0.6627,
+      "step": 34575
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9116731341113904e-05,
+      "loss": 0.6651,
+      "step": 34580
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9116272286475134e-05,
+      "loss": 0.6696,
+      "step": 34585
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.911581311809145e-05,
+      "loss": 0.6694,
+      "step": 34590
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9115353835968588e-05,
+      "loss": 0.6719,
+      "step": 34595
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9114894440112274e-05,
+      "loss": 0.6742,
+      "step": 34600
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9114434930528242e-05,
+      "loss": 0.6805,
+      "step": 34605
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9113975307222226e-05,
+      "loss": 0.6693,
+      "step": 34610
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.911351557019996e-05,
+      "loss": 0.6721,
+      "step": 34615
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9113055719467182e-05,
+      "loss": 0.6813,
+      "step": 34620
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9112595755029625e-05,
+      "loss": 0.6622,
+      "step": 34625
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.911213567689303e-05,
+      "loss": 0.6607,
+      "step": 34630
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9111675485063138e-05,
+      "loss": 0.67,
+      "step": 34635
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.911121517954569e-05,
+      "loss": 0.6751,
+      "step": 34640
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9110754760346434e-05,
+      "loss": 0.677,
+      "step": 34645
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.911029422747111e-05,
+      "loss": 0.6759,
+      "step": 34650
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9109833580925466e-05,
+      "loss": 0.6703,
+      "step": 34655
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9109372820715248e-05,
+      "loss": 0.6643,
+      "step": 34660
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9108911946846205e-05,
+      "loss": 0.6723,
+      "step": 34665
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9108450959324086e-05,
+      "loss": 0.6781,
+      "step": 34670
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9107989858154646e-05,
+      "loss": 0.683,
+      "step": 34675
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9107528643343637e-05,
+      "loss": 0.6863,
+      "step": 34680
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9107067314896816e-05,
+      "loss": 0.6758,
+      "step": 34685
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9106605872819935e-05,
+      "loss": 0.6667,
+      "step": 34690
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9106144317118754e-05,
+      "loss": 0.6846,
+      "step": 34695
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.910568264779903e-05,
+      "loss": 0.6652,
+      "step": 34700
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9105220864866524e-05,
+      "loss": 0.6611,
+      "step": 34705
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9104758968327e-05,
+      "loss": 0.6764,
+      "step": 34710
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9104296958186216e-05,
+      "loss": 0.6675,
+      "step": 34715
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9103834834449942e-05,
+      "loss": 0.6704,
+      "step": 34720
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9103372597123944e-05,
+      "loss": 0.6817,
+      "step": 34725
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9102910246213982e-05,
+      "loss": 0.6637,
+      "step": 34730
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9102447781725833e-05,
+      "loss": 0.6714,
+      "step": 34735
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9101985203665267e-05,
+      "loss": 0.6737,
+      "step": 34740
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.910152251203805e-05,
+      "loss": 0.6778,
+      "step": 34745
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9101059706849957e-05,
+      "loss": 0.6836,
+      "step": 34750
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9100596788106766e-05,
+      "loss": 0.6778,
+      "step": 34755
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.910013375581425e-05,
+      "loss": 0.665,
+      "step": 34760
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9099670609978184e-05,
+      "loss": 0.6706,
+      "step": 34765
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9099207350604353e-05,
+      "loss": 0.6806,
+      "step": 34770
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.909874397769853e-05,
+      "loss": 0.6701,
+      "step": 34775
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.90982804912665e-05,
+      "loss": 0.678,
+      "step": 34780
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.909781689131405e-05,
+      "loss": 0.6733,
+      "step": 34785
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9097353177846957e-05,
+      "loss": 0.6634,
+      "step": 34790
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.909688935087101e-05,
+      "loss": 0.6868,
+      "step": 34795
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9096425410392e-05,
+      "loss": 0.6753,
+      "step": 34800
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.909596135641571e-05,
+      "loss": 0.6667,
+      "step": 34805
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9095497188947933e-05,
+      "loss": 0.6788,
+      "step": 34810
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.909503290799446e-05,
+      "loss": 0.6686,
+      "step": 34815
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.909456851356108e-05,
+      "loss": 0.6725,
+      "step": 34820
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9094104005653594e-05,
+      "loss": 0.6693,
+      "step": 34825
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9093639384277794e-05,
+      "loss": 0.6753,
+      "step": 34830
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9093174649439477e-05,
+      "loss": 0.669,
+      "step": 34835
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9092709801144445e-05,
+      "loss": 0.6668,
+      "step": 34840
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9092244839398494e-05,
+      "loss": 0.6821,
+      "step": 34845
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9091779764207425e-05,
+      "loss": 0.6712,
+      "step": 34850
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.909131457557704e-05,
+      "loss": 0.6688,
+      "step": 34855
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9090849273513153e-05,
+      "loss": 0.6552,
+      "step": 34860
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.909038385802156e-05,
+      "loss": 0.6755,
+      "step": 34865
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.908991832910807e-05,
+      "loss": 0.6707,
+      "step": 34870
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.908945268677849e-05,
+      "loss": 0.6701,
+      "step": 34875
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.908898693103863e-05,
+      "loss": 0.683,
+      "step": 34880
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9088521061894306e-05,
+      "loss": 0.6689,
+      "step": 34885
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.908805507935133e-05,
+      "loss": 0.6602,
+      "step": 34890
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9087588983415508e-05,
+      "loss": 0.6715,
+      "step": 34895
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9087122774092666e-05,
+      "loss": 0.6737,
+      "step": 34900
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9086656451388616e-05,
+      "loss": 0.6675,
+      "step": 34905
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9086190015309175e-05,
+      "loss": 0.6649,
+      "step": 34910
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9085723465860164e-05,
+      "loss": 0.6778,
+      "step": 34915
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.908525680304741e-05,
+      "loss": 0.6706,
+      "step": 34920
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.908479002687672e-05,
+      "loss": 0.6769,
+      "step": 34925
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9084323137353937e-05,
+      "loss": 0.6675,
+      "step": 34930
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9083856134484876e-05,
+      "loss": 0.6625,
+      "step": 34935
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9083389018275364e-05,
+      "loss": 0.6721,
+      "step": 34940
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9082921788731233e-05,
+      "loss": 0.684,
+      "step": 34945
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.908245444585831e-05,
+      "loss": 0.6661,
+      "step": 34950
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9081986989662423e-05,
+      "loss": 0.6907,
+      "step": 34955
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9081519420149415e-05,
+      "loss": 0.6723,
+      "step": 34960
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9081051737325106e-05,
+      "loss": 0.6744,
+      "step": 34965
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9080583941195345e-05,
+      "loss": 0.6682,
+      "step": 34970
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.908011603176596e-05,
+      "loss": 0.6683,
+      "step": 34975
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.907964800904279e-05,
+      "loss": 0.6702,
+      "step": 34980
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9079179873031676e-05,
+      "loss": 0.6781,
+      "step": 34985
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9078711623738458e-05,
+      "loss": 0.6631,
+      "step": 34990
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9078243261168983e-05,
+      "loss": 0.6653,
+      "step": 34995
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.907777478532909e-05,
+      "loss": 0.6813,
+      "step": 35000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9077306196224622e-05,
+      "loss": 0.6712,
+      "step": 35005
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.907683749386143e-05,
+      "loss": 0.6754,
+      "step": 35010
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9076368678245364e-05,
+      "loss": 0.6723,
+      "step": 35015
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9075899749382268e-05,
+      "loss": 0.6808,
+      "step": 35020
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9075430707278e-05,
+      "loss": 0.6662,
+      "step": 35025
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.90749615519384e-05,
+      "loss": 0.6706,
+      "step": 35030
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9074492283369336e-05,
+      "loss": 0.6686,
+      "step": 35035
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9074022901576656e-05,
+      "loss": 0.6688,
+      "step": 35040
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9073553406566217e-05,
+      "loss": 0.6886,
+      "step": 35045
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9073083798343873e-05,
+      "loss": 0.6717,
+      "step": 35050
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9072614076915494e-05,
+      "loss": 0.674,
+      "step": 35055
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.907214424228693e-05,
+      "loss": 0.6561,
+      "step": 35060
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9071674294464046e-05,
+      "loss": 0.6818,
+      "step": 35065
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9071204233452708e-05,
+      "loss": 0.681,
+      "step": 35070
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9070734059258784e-05,
+      "loss": 0.6765,
+      "step": 35075
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9070263771888133e-05,
+      "loss": 0.6776,
+      "step": 35080
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.906979337134663e-05,
+      "loss": 0.6687,
+      "step": 35085
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9069322857640135e-05,
+      "loss": 0.673,
+      "step": 35090
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9068852230774525e-05,
+      "loss": 0.6717,
+      "step": 35095
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9068381490755673e-05,
+      "loss": 0.6761,
+      "step": 35100
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.906791063758945e-05,
+      "loss": 0.6654,
+      "step": 35105
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9067439671281732e-05,
+      "loss": 0.6723,
+      "step": 35110
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9066968591838395e-05,
+      "loss": 0.6777,
+      "step": 35115
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.906649739926532e-05,
+      "loss": 0.667,
+      "step": 35120
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.906602609356838e-05,
+      "loss": 0.6825,
+      "step": 35125
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9065554674753456e-05,
+      "loss": 0.6793,
+      "step": 35130
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9065083142826436e-05,
+      "loss": 0.6704,
+      "step": 35135
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9064611497793196e-05,
+      "loss": 0.6693,
+      "step": 35140
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9064139739659628e-05,
+      "loss": 0.6743,
+      "step": 35145
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9063667868431613e-05,
+      "loss": 0.6807,
+      "step": 35150
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.906319588411504e-05,
+      "loss": 0.666,
+      "step": 35155
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.90627237867158e-05,
+      "loss": 0.674,
+      "step": 35160
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.906225157623978e-05,
+      "loss": 0.6623,
+      "step": 35165
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9061779252692876e-05,
+      "loss": 0.667,
+      "step": 35170
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9061306816080974e-05,
+      "loss": 0.6694,
+      "step": 35175
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9060834266409977e-05,
+      "loss": 0.6756,
+      "step": 35180
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9060361603685777e-05,
+      "loss": 0.6778,
+      "step": 35185
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9059888827914273e-05,
+      "loss": 0.668,
+      "step": 35190
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.905941593910136e-05,
+      "loss": 0.6757,
+      "step": 35195
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9058942937252943e-05,
+      "loss": 0.6775,
+      "step": 35200
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.905846982237492e-05,
+      "loss": 0.6715,
+      "step": 35205
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.90579965944732e-05,
+      "loss": 0.6664,
+      "step": 35210
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.905752325355368e-05,
+      "loss": 0.6931,
+      "step": 35215
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9057049799622268e-05,
+      "loss": 0.6755,
+      "step": 35220
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9056576232684878e-05,
+      "loss": 0.6693,
+      "step": 35225
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9056102552747407e-05,
+      "loss": 0.6702,
+      "step": 35230
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9055628759815778e-05,
+      "loss": 0.6643,
+      "step": 35235
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9055154853895894e-05,
+      "loss": 0.6734,
+      "step": 35240
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.905468083499367e-05,
+      "loss": 0.6683,
+      "step": 35245
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.905420670311502e-05,
+      "loss": 0.6693,
+      "step": 35250
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.905373245826586e-05,
+      "loss": 0.6843,
+      "step": 35255
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9053258100452108e-05,
+      "loss": 0.673,
+      "step": 35260
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9052783629679684e-05,
+      "loss": 0.6731,
+      "step": 35265
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9052309045954504e-05,
+      "loss": 0.6801,
+      "step": 35270
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.905183434928249e-05,
+      "loss": 0.6586,
+      "step": 35275
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9051359539669573e-05,
+      "loss": 0.676,
+      "step": 35280
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9050884617121667e-05,
+      "loss": 0.6637,
+      "step": 35285
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.90504095816447e-05,
+      "loss": 0.6818,
+      "step": 35290
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.90499344332446e-05,
+      "loss": 0.6749,
+      "step": 35295
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.90494591719273e-05,
+      "loss": 0.6683,
+      "step": 35300
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9048983797698723e-05,
+      "loss": 0.6673,
+      "step": 35305
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9048508310564807e-05,
+      "loss": 0.6763,
+      "step": 35310
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.904803271053148e-05,
+      "loss": 0.665,
+      "step": 35315
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.904755699760467e-05,
+      "loss": 0.675,
+      "step": 35320
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9047081171790327e-05,
+      "loss": 0.6669,
+      "step": 35325
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.904660523309438e-05,
+      "loss": 0.6704,
+      "step": 35330
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.904612918152277e-05,
+      "loss": 0.6809,
+      "step": 35335
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.904565301708143e-05,
+      "loss": 0.6648,
+      "step": 35340
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9045176739776305e-05,
+      "loss": 0.679,
+      "step": 35345
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9044700349613344e-05,
+      "loss": 0.67,
+      "step": 35350
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9044223846598483e-05,
+      "loss": 0.6688,
+      "step": 35355
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9043747230737667e-05,
+      "loss": 0.676,
+      "step": 35360
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9043270502036847e-05,
+      "loss": 0.6602,
+      "step": 35365
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.904279366050197e-05,
+      "loss": 0.6648,
+      "step": 35370
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9042316706138987e-05,
+      "loss": 0.6759,
+      "step": 35375
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9041839638953847e-05,
+      "loss": 0.6726,
+      "step": 35380
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9041362458952503e-05,
+      "loss": 0.6731,
+      "step": 35385
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.904088516614091e-05,
+      "loss": 0.681,
+      "step": 35390
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.904040776052502e-05,
+      "loss": 0.6717,
+      "step": 35395
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.903993024211079e-05,
+      "loss": 0.6816,
+      "step": 35400
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.903945261090418e-05,
+      "loss": 0.6643,
+      "step": 35405
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9038974866911154e-05,
+      "loss": 0.6767,
+      "step": 35410
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9038497010137666e-05,
+      "loss": 0.6679,
+      "step": 35415
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9038019040589678e-05,
+      "loss": 0.6749,
+      "step": 35420
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.903754095827316e-05,
+      "loss": 0.6667,
+      "step": 35425
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9037062763194067e-05,
+      "loss": 0.6668,
+      "step": 35430
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9036584455358373e-05,
+      "loss": 0.6691,
+      "step": 35435
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.903610603477205e-05,
+      "loss": 0.6729,
+      "step": 35440
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9035627501441057e-05,
+      "loss": 0.6652,
+      "step": 35445
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.903514885537137e-05,
+      "loss": 0.6727,
+      "step": 35450
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.903467009656896e-05,
+      "loss": 0.6664,
+      "step": 35455
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9034191225039802e-05,
+      "loss": 0.6758,
+      "step": 35460
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.903371224078987e-05,
+      "loss": 0.6769,
+      "step": 35465
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9033233143825144e-05,
+      "loss": 0.6759,
+      "step": 35470
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9032753934151594e-05,
+      "loss": 0.6764,
+      "step": 35475
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9032274611775204e-05,
+      "loss": 0.6614,
+      "step": 35480
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9031795176701955e-05,
+      "loss": 0.6701,
+      "step": 35485
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9031315628937826e-05,
+      "loss": 0.6857,
+      "step": 35490
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9030835968488805e-05,
+      "loss": 0.661,
+      "step": 35495
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9030356195360875e-05,
+      "loss": 0.6701,
+      "step": 35500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.902987630956002e-05,
+      "loss": 0.6696,
+      "step": 35505
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9029396311092227e-05,
+      "loss": 0.6698,
+      "step": 35510
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9028916199963486e-05,
+      "loss": 0.6763,
+      "step": 35515
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.902843597617979e-05,
+      "loss": 0.6731,
+      "step": 35520
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.902795563974713e-05,
+      "loss": 0.6671,
+      "step": 35525
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.90274751906715e-05,
+      "loss": 0.6691,
+      "step": 35530
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.902699462895889e-05,
+      "loss": 0.6806,
+      "step": 35535
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.90265139546153e-05,
+      "loss": 0.6721,
+      "step": 35540
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9026033167646727e-05,
+      "loss": 0.6687,
+      "step": 35545
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.902555226805917e-05,
+      "loss": 0.6705,
+      "step": 35550
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9025071255858627e-05,
+      "loss": 0.6656,
+      "step": 35555
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9024590131051098e-05,
+      "loss": 0.6697,
+      "step": 35560
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9024108893642594e-05,
+      "loss": 0.6808,
+      "step": 35565
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9023627543639118e-05,
+      "loss": 0.6547,
+      "step": 35570
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9023146081046664e-05,
+      "loss": 0.6725,
+      "step": 35575
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9022664505871252e-05,
+      "loss": 0.6679,
+      "step": 35580
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9022182818118888e-05,
+      "loss": 0.6643,
+      "step": 35585
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9021701017795577e-05,
+      "loss": 0.6689,
+      "step": 35590
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9021219104907337e-05,
+      "loss": 0.6746,
+      "step": 35595
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9020737079460178e-05,
+      "loss": 0.6605,
+      "step": 35600
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.902025494146011e-05,
+      "loss": 0.6665,
+      "step": 35605
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9019772690913158e-05,
+      "loss": 0.6856,
+      "step": 35610
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9019290327825334e-05,
+      "loss": 0.6782,
+      "step": 35615
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.901880785220265e-05,
+      "loss": 0.6639,
+      "step": 35620
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.901832526405114e-05,
+      "loss": 0.6655,
+      "step": 35625
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9017842563376813e-05,
+      "loss": 0.6554,
+      "step": 35630
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.90173597501857e-05,
+      "loss": 0.6742,
+      "step": 35635
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9016876824483817e-05,
+      "loss": 0.6692,
+      "step": 35640
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9016393786277202e-05,
+      "loss": 0.6597,
+      "step": 35645
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.901591063557187e-05,
+      "loss": 0.6619,
+      "step": 35650
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9015427372373852e-05,
+      "loss": 0.6847,
+      "step": 35655
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.901494399668918e-05,
+      "loss": 0.6703,
+      "step": 35660
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9014460508523883e-05,
+      "loss": 0.6802,
+      "step": 35665
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9013976907883996e-05,
+      "loss": 0.679,
+      "step": 35670
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9013493194775553e-05,
+      "loss": 0.6649,
+      "step": 35675
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9013009369204586e-05,
+      "loss": 0.6622,
+      "step": 35680
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9012525431177134e-05,
+      "loss": 0.6793,
+      "step": 35685
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.901204138069924e-05,
+      "loss": 0.6812,
+      "step": 35690
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9011557217776933e-05,
+      "loss": 0.6631,
+      "step": 35695
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.901107294241626e-05,
+      "loss": 0.6803,
+      "step": 35700
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9010588554623263e-05,
+      "loss": 0.6821,
+      "step": 35705
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9010104054403988e-05,
+      "loss": 0.6689,
+      "step": 35710
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9009619441764477e-05,
+      "loss": 0.6687,
+      "step": 35715
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.900913471671078e-05,
+      "loss": 0.6817,
+      "step": 35720
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9008649879248938e-05,
+      "loss": 0.6754,
+      "step": 35725
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9008164929385008e-05,
+      "loss": 0.6722,
+      "step": 35730
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9007679867125037e-05,
+      "loss": 0.6707,
+      "step": 35735
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9007194692475073e-05,
+      "loss": 0.678,
+      "step": 35740
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.900670940544118e-05,
+      "loss": 0.6645,
+      "step": 35745
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9006224006029404e-05,
+      "loss": 0.669,
+      "step": 35750
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9005738494245806e-05,
+      "loss": 0.6923,
+      "step": 35755
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9005252870096446e-05,
+      "loss": 0.6754,
+      "step": 35760
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9004767133587373e-05,
+      "loss": 0.6682,
+      "step": 35765
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.900428128472466e-05,
+      "loss": 0.6797,
+      "step": 35770
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9003795323514363e-05,
+      "loss": 0.6692,
+      "step": 35775
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9003309249962543e-05,
+      "loss": 0.6783,
+      "step": 35780
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9002823064075268e-05,
+      "loss": 0.6785,
+      "step": 35785
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9002336765858603e-05,
+      "loss": 0.6629,
+      "step": 35790
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.900185035531862e-05,
+      "loss": 0.671,
+      "step": 35795
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9001363832461385e-05,
+      "loss": 0.6799,
+      "step": 35800
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9000877197292964e-05,
+      "loss": 0.6502,
+      "step": 35805
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9000390449819433e-05,
+      "loss": 0.6655,
+      "step": 35810
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.8999903590046863e-05,
+      "loss": 0.6694,
+      "step": 35815
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.8999416617981334e-05,
+      "loss": 0.6722,
+      "step": 35820
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.899892953362892e-05,
+      "loss": 0.6722,
+      "step": 35825
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.899844233699569e-05,
+      "loss": 0.6681,
+      "step": 35830
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.8997955028087735e-05,
+      "loss": 0.6642,
+      "step": 35835
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.899746760691113e-05,
+      "loss": 0.6764,
+      "step": 35840
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.8996980073471953e-05,
+      "loss": 0.6772,
+      "step": 35845
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.8996492427776295e-05,
+      "loss": 0.6674,
+      "step": 35850
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.899600466983023e-05,
+      "loss": 0.6693,
+      "step": 35855
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.8995516799639856e-05,
+      "loss": 0.6762,
+      "step": 35860
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.899502881721125e-05,
+      "loss": 0.6789,
+      "step": 35865
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.8994540722550504e-05,
+      "loss": 0.6865,
+      "step": 35870
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.899405251566371e-05,
+      "loss": 0.669,
+      "step": 35875
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.8993564196556962e-05,
+      "loss": 0.6605,
+      "step": 35880
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.8993075765236342e-05,
+      "loss": 0.6679,
+      "step": 35885
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.8992587221707953e-05,
+      "loss": 0.6617,
+      "step": 35890
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.8992098565977893e-05,
+      "loss": 0.6776,
+      "step": 35895
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.899160979805225e-05,
+      "loss": 0.6767,
+      "step": 35900
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.899112091793713e-05,
+      "loss": 0.6645,
+      "step": 35905
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.8990631925638627e-05,
+      "loss": 0.6684,
+      "step": 35910
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.8990142821162848e-05,
+      "loss": 0.659,
+      "step": 35915
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.898965360451589e-05,
+      "loss": 0.6534,
+      "step": 35920
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.8989164275703864e-05,
+      "loss": 0.6757,
+      "step": 35925
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.898867483473287e-05,
+      "loss": 0.6645,
+      "step": 35930
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.8988185281609013e-05,
+      "loss": 0.662,
+      "step": 35935
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.8987695616338405e-05,
+      "loss": 0.6705,
+      "step": 35940
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.8987205838927157e-05,
+      "loss": 0.6775,
+      "step": 35945
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.8986715949381378e-05,
+      "loss": 0.6535,
+      "step": 35950
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.8986225947707177e-05,
+      "loss": 0.6847,
+      "step": 35955
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.8985735833910675e-05,
+      "loss": 0.6733,
+      "step": 35960
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.898524560799798e-05,
+      "loss": 0.6581,
+      "step": 35965
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.8984755269975213e-05,
+      "loss": 0.6691,
+      "step": 35970
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.8984264819848494e-05,
+      "loss": 0.6728,
+      "step": 35975
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.8983774257623934e-05,
+      "loss": 0.6529,
+      "step": 35980
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.8983283583307662e-05,
+      "loss": 0.672,
+      "step": 35985
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.8982792796905797e-05,
+      "loss": 0.6664,
+      "step": 35990
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.8982301898424465e-05,
+      "loss": 0.6677,
+      "step": 35995
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.8981810887869784e-05,
+      "loss": 0.6655,
+      "step": 36000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.898131976524789e-05,
+      "loss": 0.6789,
+      "step": 36005
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.8980828530564907e-05,
+      "loss": 0.6613,
+      "step": 36010
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.898033718382696e-05,
+      "loss": 0.6697,
+      "step": 36015
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.8979845725040186e-05,
+      "loss": 0.6606,
+      "step": 36020
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.897935415421071e-05,
+      "loss": 0.6677,
+      "step": 36025
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.8978862471344672e-05,
+      "loss": 0.6691,
+      "step": 36030
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.8978370676448205e-05,
+      "loss": 0.6665,
+      "step": 36035
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.8977878769527444e-05,
+      "loss": 0.6656,
+      "step": 36040
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.8977386750588523e-05,
+      "loss": 0.6671,
+      "step": 36045
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.897689461963759e-05,
+      "loss": 0.6753,
+      "step": 36050
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.897640237668078e-05,
+      "loss": 0.6764,
+      "step": 36055
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.897591002172423e-05,
+      "loss": 0.6724,
+      "step": 36060
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.8975417554774093e-05,
+      "loss": 0.6721,
+      "step": 36065
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.8974924975836507e-05,
+      "loss": 0.6626,
+      "step": 36070
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.897443228491762e-05,
+      "loss": 0.6836,
+      "step": 36075
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.897393948202358e-05,
+      "loss": 0.6657,
+      "step": 36080
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.8973446567160535e-05,
+      "loss": 0.6791,
+      "step": 36085
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.8972953540334634e-05,
+      "loss": 0.6675,
+      "step": 36090
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.897246040155203e-05,
+      "loss": 0.6749,
+      "step": 36095
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.897196715081888e-05,
+      "loss": 0.6824,
+      "step": 36100
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.8971473788141327e-05,
+      "loss": 0.67,
+      "step": 36105
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.8970980313525536e-05,
+      "loss": 0.6716,
+      "step": 36110
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.897048672697766e-05,
+      "loss": 0.6763,
+      "step": 36115
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.8969993028503862e-05,
+      "loss": 0.6747,
+      "step": 36120
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.8969499218110302e-05,
+      "loss": 0.6631,
+      "step": 36125
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.8969005295803134e-05,
+      "loss": 0.6757,
+      "step": 36130
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.8968511261588528e-05,
+      "loss": 0.6712,
+      "step": 36135
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.8968017115472646e-05,
+      "loss": 0.6682,
+      "step": 36140
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.8967522857461652e-05,
+      "loss": 0.6683,
+      "step": 36145
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.8967028487561712e-05,
+      "loss": 0.6648,
+      "step": 36150
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.8966534005779e-05,
+      "loss": 0.676,
+      "step": 36155
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.896603941211968e-05,
+      "loss": 0.6775,
+      "step": 36160
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.8965544706589925e-05,
+      "loss": 0.6757,
+      "step": 36165
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.896504988919591e-05,
+      "loss": 0.6727,
+      "step": 36170
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.8964554959943803e-05,
+      "loss": 0.66,
+      "step": 36175
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.8964059918839784e-05,
+      "loss": 0.6809,
+      "step": 36180
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.896356476589003e-05,
+      "loss": 0.6648,
+      "step": 36185
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.8963069501100717e-05,
+      "loss": 0.6777,
+      "step": 36190
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.8962574124478023e-05,
+      "loss": 0.666,
+      "step": 36195
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.8962078636028135e-05,
+      "loss": 0.6665,
+      "step": 36200
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.8961583035757227e-05,
+      "loss": 0.6725,
+      "step": 36205
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.8961087323671487e-05,
+      "loss": 0.657,
+      "step": 36210
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.8960591499777104e-05,
+      "loss": 0.6542,
+      "step": 36215
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.8960095564080258e-05,
+      "loss": 0.6774,
+      "step": 36220
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.895959951658714e-05,
+      "loss": 0.6609,
+      "step": 36225
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.895910335730394e-05,
+      "loss": 0.6573,
+      "step": 36230
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.8958607086236842e-05,
+      "loss": 0.6849,
+      "step": 36235
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.895811070339205e-05,
+      "loss": 0.6641,
+      "step": 36240
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.8957614208775747e-05,
+      "loss": 0.6742,
+      "step": 36245
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.895711760239413e-05,
+      "loss": 0.6715,
+      "step": 36250
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.89566208842534e-05,
+      "loss": 0.6666,
+      "step": 36255
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.895612405435975e-05,
+      "loss": 0.663,
+      "step": 36260
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.895562711271938e-05,
+      "loss": 0.6665,
+      "step": 36265
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.895513005933849e-05,
+      "loss": 0.662,
+      "step": 36270
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.8954632894223285e-05,
+      "loss": 0.6682,
+      "step": 36275
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.895413561737996e-05,
+      "loss": 0.6796,
+      "step": 36280
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.895363822881473e-05,
+      "loss": 0.6623,
+      "step": 36285
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.89531407285338e-05,
+      "loss": 0.6719,
+      "step": 36290
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.8952643116543368e-05,
+      "loss": 0.6548,
+      "step": 36295
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.895214539284965e-05,
+      "loss": 0.6688,
+      "step": 36300
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.8951647557458856e-05,
+      "loss": 0.6704,
+      "step": 36305
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.8951149610377193e-05,
+      "loss": 0.6604,
+      "step": 36310
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.8950651551610878e-05,
+      "loss": 0.6578,
+      "step": 36315
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.8950153381166126e-05,
+      "loss": 0.6605,
+      "step": 36320
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.894965509904915e-05,
+      "loss": 0.6784,
+      "step": 36325
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.894915670526617e-05,
+      "loss": 0.6669,
+      "step": 36330
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.8948658199823402e-05,
+      "loss": 0.6708,
+      "step": 36335
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.8948159582727067e-05,
+      "loss": 0.6679,
+      "step": 36340
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.8947660853983387e-05,
+      "loss": 0.675,
+      "step": 36345
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.894716201359858e-05,
+      "loss": 0.6787,
+      "step": 36350
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.8946663061578877e-05,
+      "loss": 0.6764,
+      "step": 36355
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.89461639979305e-05,
+      "loss": 0.6746,
+      "step": 36360
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.8945664822659675e-05,
+      "loss": 0.6643,
+      "step": 36365
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.8945165535772632e-05,
+      "loss": 0.6739,
+      "step": 36370
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.89446661372756e-05,
+      "loss": 0.6826,
+      "step": 36375
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.894416662717481e-05,
+      "loss": 0.6622,
+      "step": 36380
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.8943667005476498e-05,
+      "loss": 0.6647,
+      "step": 36385
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.894316727218689e-05,
+      "loss": 0.6733,
+      "step": 36390
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.894266742731223e-05,
+      "loss": 0.6666,
+      "step": 36395
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.8942167470858747e-05,
+      "loss": 0.6664,
+      "step": 36400
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.8941667402832683e-05,
+      "loss": 0.6676,
+      "step": 36405
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8941167223240277e-05,
+      "loss": 0.6574,
+      "step": 36410
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.894066693208777e-05,
+      "loss": 0.6669,
+      "step": 36415
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.894016652938141e-05,
+      "loss": 0.6798,
+      "step": 36420
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8939666015127424e-05,
+      "loss": 0.6654,
+      "step": 36425
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.893916538933207e-05,
+      "loss": 0.6812,
+      "step": 36430
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8938664652001596e-05,
+      "loss": 0.6746,
+      "step": 36435
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8938163803142244e-05,
+      "loss": 0.6689,
+      "step": 36440
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.893766284276026e-05,
+      "loss": 0.6791,
+      "step": 36445
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8937161770861906e-05,
+      "loss": 0.6782,
+      "step": 36450
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8936660587453427e-05,
+      "loss": 0.6739,
+      "step": 36455
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.893615929254107e-05,
+      "loss": 0.6711,
+      "step": 36460
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8935657886131102e-05,
+      "loss": 0.6683,
+      "step": 36465
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8935156368229773e-05,
+      "loss": 0.6764,
+      "step": 36470
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.893465473884334e-05,
+      "loss": 0.6674,
+      "step": 36475
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8934152997978065e-05,
+      "loss": 0.6729,
+      "step": 36480
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.89336511456402e-05,
+      "loss": 0.6682,
+      "step": 36485
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.893314918183602e-05,
+      "loss": 0.6662,
+      "step": 36490
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8932647106571775e-05,
+      "loss": 0.6727,
+      "step": 36495
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.893214491985374e-05,
+      "loss": 0.6533,
+      "step": 36500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.893164262168817e-05,
+      "loss": 0.6609,
+      "step": 36505
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.893114021208134e-05,
+      "loss": 0.6734,
+      "step": 36510
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.893063769103952e-05,
+      "loss": 0.6769,
+      "step": 36515
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8930135058568977e-05,
+      "loss": 0.6694,
+      "step": 36520
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8929632314675982e-05,
+      "loss": 0.6636,
+      "step": 36525
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8929129459366805e-05,
+      "loss": 0.6729,
+      "step": 36530
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.892862649264773e-05,
+      "loss": 0.6671,
+      "step": 36535
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8928123414525018e-05,
+      "loss": 0.6701,
+      "step": 36540
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8927620225004958e-05,
+      "loss": 0.6604,
+      "step": 36545
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8927116924093824e-05,
+      "loss": 0.6804,
+      "step": 36550
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8926613511797898e-05,
+      "loss": 0.6725,
+      "step": 36555
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8926109988123455e-05,
+      "loss": 0.6677,
+      "step": 36560
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8925606353076787e-05,
+      "loss": 0.6646,
+      "step": 36565
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.892510260666417e-05,
+      "loss": 0.667,
+      "step": 36570
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8924598748891888e-05,
+      "loss": 0.6546,
+      "step": 36575
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8924094779766237e-05,
+      "loss": 0.6677,
+      "step": 36580
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8923590699293497e-05,
+      "loss": 0.6695,
+      "step": 36585
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8923086507479963e-05,
+      "loss": 0.6752,
+      "step": 36590
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8922582204331918e-05,
+      "loss": 0.6736,
+      "step": 36595
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8922077789855665e-05,
+      "loss": 0.6671,
+      "step": 36600
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8921573264057488e-05,
+      "loss": 0.6684,
+      "step": 36605
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.892106862694369e-05,
+      "loss": 0.6655,
+      "step": 36610
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8920563878520562e-05,
+      "loss": 0.6726,
+      "step": 36615
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.89200590187944e-05,
+      "loss": 0.6675,
+      "step": 36620
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8919554047771508e-05,
+      "loss": 0.6704,
+      "step": 36625
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.891904896545819e-05,
+      "loss": 0.664,
+      "step": 36630
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8918543771860734e-05,
+      "loss": 0.6727,
+      "step": 36635
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.891803846698546e-05,
+      "loss": 0.6644,
+      "step": 36640
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8917533050838663e-05,
+      "loss": 0.6693,
+      "step": 36645
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8917027523426648e-05,
+      "loss": 0.6599,
+      "step": 36650
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.891652188475573e-05,
+      "loss": 0.6841,
+      "step": 36655
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8916016134832213e-05,
+      "loss": 0.6577,
+      "step": 36660
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8915510273662406e-05,
+      "loss": 0.6729,
+      "step": 36665
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8915004301252626e-05,
+      "loss": 0.6646,
+      "step": 36670
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.891449821760918e-05,
+      "loss": 0.655,
+      "step": 36675
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8913992022738386e-05,
+      "loss": 0.6656,
+      "step": 36680
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.891348571664656e-05,
+      "loss": 0.6686,
+      "step": 36685
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8912979299340016e-05,
+      "loss": 0.6715,
+      "step": 36690
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.891247277082508e-05,
+      "loss": 0.6614,
+      "step": 36695
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8911966131108064e-05,
+      "loss": 0.6606,
+      "step": 36700
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8911459380195293e-05,
+      "loss": 0.6557,
+      "step": 36705
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8910952518093087e-05,
+      "loss": 0.6771,
+      "step": 36710
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8910445544807774e-05,
+      "loss": 0.6597,
+      "step": 36715
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8909938460345677e-05,
+      "loss": 0.6653,
+      "step": 36720
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8909431264713128e-05,
+      "loss": 0.6774,
+      "step": 36725
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.890892395791645e-05,
+      "loss": 0.6688,
+      "step": 36730
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.890841653996197e-05,
+      "loss": 0.6649,
+      "step": 36735
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.890790901085603e-05,
+      "loss": 0.6599,
+      "step": 36740
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.890740137060495e-05,
+      "loss": 0.6728,
+      "step": 36745
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.890689361921507e-05,
+      "loss": 0.6789,
+      "step": 36750
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8906385756692724e-05,
+      "loss": 0.6658,
+      "step": 36755
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.890587778304425e-05,
+      "loss": 0.6656,
+      "step": 36760
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.890536969827599e-05,
+      "loss": 0.664,
+      "step": 36765
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8904861502394275e-05,
+      "loss": 0.6735,
+      "step": 36770
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.890435319540545e-05,
+      "loss": 0.6646,
+      "step": 36775
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8903844777315855e-05,
+      "loss": 0.675,
+      "step": 36780
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.890333624813184e-05,
+      "loss": 0.6635,
+      "step": 36785
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8902827607859742e-05,
+      "loss": 0.662,
+      "step": 36790
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8902318856505916e-05,
+      "loss": 0.6626,
+      "step": 36795
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8901809994076702e-05,
+      "loss": 0.6584,
+      "step": 36800
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.890130102057845e-05,
+      "loss": 0.6758,
+      "step": 36805
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8900791936017518e-05,
+      "loss": 0.6631,
+      "step": 36810
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8900282740400247e-05,
+      "loss": 0.6652,
+      "step": 36815
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8899773433732996e-05,
+      "loss": 0.6706,
+      "step": 36820
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.889926401602212e-05,
+      "loss": 0.6721,
+      "step": 36825
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8898754487273975e-05,
+      "loss": 0.6634,
+      "step": 36830
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.889824484749492e-05,
+      "loss": 0.6667,
+      "step": 36835
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.889773509669131e-05,
+      "loss": 0.6546,
+      "step": 36840
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8897225234869505e-05,
+      "loss": 0.6578,
+      "step": 36845
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.889671526203587e-05,
+      "loss": 0.6687,
+      "step": 36850
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8896205178196766e-05,
+      "loss": 0.6827,
+      "step": 36855
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.889569498335856e-05,
+      "loss": 0.6658,
+      "step": 36860
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8895184677527615e-05,
+      "loss": 0.6733,
+      "step": 36865
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8894674260710296e-05,
+      "loss": 0.6607,
+      "step": 36870
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.889416373291298e-05,
+      "loss": 0.6596,
+      "step": 36875
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8893653094142027e-05,
+      "loss": 0.6526,
+      "step": 36880
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8893142344403812e-05,
+      "loss": 0.6461,
+      "step": 36885
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8892631483704708e-05,
+      "loss": 0.669,
+      "step": 36890
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.889212051205109e-05,
+      "loss": 0.6619,
+      "step": 36895
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8891609429449335e-05,
+      "loss": 0.6767,
+      "step": 36900
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8891098235905815e-05,
+      "loss": 0.6591,
+      "step": 36905
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8890586931426912e-05,
+      "loss": 0.6643,
+      "step": 36910
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8890075516019006e-05,
+      "loss": 0.6621,
+      "step": 36915
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8889563989688476e-05,
+      "loss": 0.655,
+      "step": 36920
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8889052352441702e-05,
+      "loss": 0.6691,
+      "step": 36925
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8888540604285075e-05,
+      "loss": 0.6639,
+      "step": 36930
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8888028745224973e-05,
+      "loss": 0.6791,
+      "step": 36935
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8887516775267788e-05,
+      "loss": 0.6662,
+      "step": 36940
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8887004694419902e-05,
+      "loss": 0.656,
+      "step": 36945
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.888649250268771e-05,
+      "loss": 0.6647,
+      "step": 36950
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8885980200077595e-05,
+      "loss": 0.6702,
+      "step": 36955
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.888546778659596e-05,
+      "loss": 0.6711,
+      "step": 36960
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8884955262249193e-05,
+      "loss": 0.6587,
+      "step": 36965
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8884442627043688e-05,
+      "loss": 0.6563,
+      "step": 36970
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.888392988098584e-05,
+      "loss": 0.6587,
+      "step": 36975
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8883417024082047e-05,
+      "loss": 0.6828,
+      "step": 36980
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8882904056338712e-05,
+      "loss": 0.6745,
+      "step": 36985
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8882390977762234e-05,
+      "loss": 0.664,
+      "step": 36990
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.888187778835901e-05,
+      "loss": 0.6646,
+      "step": 36995
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8881364488135448e-05,
+      "loss": 0.6614,
+      "step": 37000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8880851077097952e-05,
+      "loss": 0.6759,
+      "step": 37005
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8880337555252923e-05,
+      "loss": 0.6706,
+      "step": 37010
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8879823922606777e-05,
+      "loss": 0.67,
+      "step": 37015
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8879310179165915e-05,
+      "loss": 0.6757,
+      "step": 37020
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8878796324936752e-05,
+      "loss": 0.6597,
+      "step": 37025
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8878282359925698e-05,
+      "loss": 0.6629,
+      "step": 37030
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8877768284139164e-05,
+      "loss": 0.6703,
+      "step": 37035
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8877254097583566e-05,
+      "loss": 0.6724,
+      "step": 37040
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.887673980026532e-05,
+      "loss": 0.6585,
+      "step": 37045
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.887622539219084e-05,
+      "loss": 0.6677,
+      "step": 37050
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8875710873366545e-05,
+      "loss": 0.6791,
+      "step": 37055
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8875196243798858e-05,
+      "loss": 0.6647,
+      "step": 37060
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8874681503494197e-05,
+      "loss": 0.6754,
+      "step": 37065
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.887416665245899e-05,
+      "loss": 0.6721,
+      "step": 37070
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8873651690699652e-05,
+      "loss": 0.6747,
+      "step": 37075
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8873136618222615e-05,
+      "loss": 0.6533,
+      "step": 37080
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.88726214350343e-05,
+      "loss": 0.6625,
+      "step": 37085
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.887210614114114e-05,
+      "loss": 0.6599,
+      "step": 37090
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8871590736549564e-05,
+      "loss": 0.6592,
+      "step": 37095
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8871075221266003e-05,
+      "loss": 0.6733,
+      "step": 37100
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8870559595296883e-05,
+      "loss": 0.6702,
+      "step": 37105
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8870043858648644e-05,
+      "loss": 0.6707,
+      "step": 37110
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.886952801132772e-05,
+      "loss": 0.6651,
+      "step": 37115
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8869012053340546e-05,
+      "loss": 0.6823,
+      "step": 37120
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.886849598469356e-05,
+      "loss": 0.6845,
+      "step": 37125
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8867979805393203e-05,
+      "loss": 0.6625,
+      "step": 37130
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.886746351544591e-05,
+      "loss": 0.6745,
+      "step": 37135
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8866947114858128e-05,
+      "loss": 0.6689,
+      "step": 37140
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.88664306036363e-05,
+      "loss": 0.6562,
+      "step": 37145
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8865913981786867e-05,
+      "loss": 0.6657,
+      "step": 37150
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8865397249316277e-05,
+      "loss": 0.6636,
+      "step": 37155
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.886488040623098e-05,
+      "loss": 0.6691,
+      "step": 37160
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.886436345253742e-05,
+      "loss": 0.6654,
+      "step": 37165
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8863846388242047e-05,
+      "loss": 0.6655,
+      "step": 37170
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8863329213351318e-05,
+      "loss": 0.6788,
+      "step": 37175
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8862811927871686e-05,
+      "loss": 0.6551,
+      "step": 37180
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8862294531809596e-05,
+      "loss": 0.6715,
+      "step": 37185
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8861777025171512e-05,
+      "loss": 0.6657,
+      "step": 37190
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8861259407963888e-05,
+      "loss": 0.6769,
+      "step": 37195
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.886074168019318e-05,
+      "loss": 0.6512,
+      "step": 37200
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8860223841865856e-05,
+      "loss": 0.6724,
+      "step": 37205
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8859705892988366e-05,
+      "loss": 0.6696,
+      "step": 37210
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8859187833567182e-05,
+      "loss": 0.6735,
+      "step": 37215
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8858669663608764e-05,
+      "loss": 0.6598,
+      "step": 37220
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8858151383119576e-05,
+      "loss": 0.668,
+      "step": 37225
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8857632992106087e-05,
+      "loss": 0.6681,
+      "step": 37230
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8857114490574765e-05,
+      "loss": 0.6598,
+      "step": 37235
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8856595878532074e-05,
+      "loss": 0.6667,
+      "step": 37240
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8856077155984493e-05,
+      "loss": 0.6681,
+      "step": 37245
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8855558322938492e-05,
+      "loss": 0.6562,
+      "step": 37250
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.885503937940054e-05,
+      "loss": 0.6551,
+      "step": 37255
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8854520325377117e-05,
+      "loss": 0.6658,
+      "step": 37260
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8854001160874696e-05,
+      "loss": 0.6687,
+      "step": 37265
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8853481885899755e-05,
+      "loss": 0.6648,
+      "step": 37270
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8852962500458775e-05,
+      "loss": 0.6697,
+      "step": 37275
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8852443004558237e-05,
+      "loss": 0.6677,
+      "step": 37280
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8851923398204616e-05,
+      "loss": 0.6726,
+      "step": 37285
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8851403681404408e-05,
+      "loss": 0.6771,
+      "step": 37290
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8850883854164083e-05,
+      "loss": 0.6623,
+      "step": 37295
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8850363916490137e-05,
+      "loss": 0.6664,
+      "step": 37300
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8849843868389055e-05,
+      "loss": 0.6693,
+      "step": 37305
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8849323709867324e-05,
+      "loss": 0.6724,
+      "step": 37310
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8848803440931437e-05,
+      "loss": 0.6616,
+      "step": 37315
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.884828306158788e-05,
+      "loss": 0.66,
+      "step": 37320
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8847762571843153e-05,
+      "loss": 0.6731,
+      "step": 37325
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8847241971703743e-05,
+      "loss": 0.6719,
+      "step": 37330
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8846721261176153e-05,
+      "loss": 0.6562,
+      "step": 37335
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8846200440266875e-05,
+      "loss": 0.669,
+      "step": 37340
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8845679508982407e-05,
+      "loss": 0.659,
+      "step": 37345
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8845158467329248e-05,
+      "loss": 0.6667,
+      "step": 37350
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8844637315313904e-05,
+      "loss": 0.6514,
+      "step": 37355
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8844116052942878e-05,
+      "loss": 0.6614,
+      "step": 37360
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8843594680222667e-05,
+      "loss": 0.6571,
+      "step": 37365
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8843073197159782e-05,
+      "loss": 0.668,
+      "step": 37370
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8842551603760725e-05,
+      "loss": 0.6583,
+      "step": 37375
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8842029900032005e-05,
+      "loss": 0.6669,
+      "step": 37380
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8841508085980137e-05,
+      "loss": 0.6628,
+      "step": 37385
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8840986161611626e-05,
+      "loss": 0.6598,
+      "step": 37390
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8840464126932988e-05,
+      "loss": 0.6732,
+      "step": 37395
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.883994198195073e-05,
+      "loss": 0.6702,
+      "step": 37400
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8839419726671375e-05,
+      "loss": 0.6652,
+      "step": 37405
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8838897361101433e-05,
+      "loss": 0.6745,
+      "step": 37410
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8838374885247426e-05,
+      "loss": 0.6708,
+      "step": 37415
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.883785229911587e-05,
+      "loss": 0.6687,
+      "step": 37420
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8837329602713286e-05,
+      "loss": 0.6791,
+      "step": 37425
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8836806796046194e-05,
+      "loss": 0.6676,
+      "step": 37430
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.883628387912112e-05,
+      "loss": 0.659,
+      "step": 37435
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.883576085194459e-05,
+      "loss": 0.6608,
+      "step": 37440
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8835237714523127e-05,
+      "loss": 0.6629,
+      "step": 37445
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.883471446686326e-05,
+      "loss": 0.673,
+      "step": 37450
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8834191108971516e-05,
+      "loss": 0.6494,
+      "step": 37455
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8833667640854423e-05,
+      "loss": 0.6665,
+      "step": 37460
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8833144062518517e-05,
+      "loss": 0.6623,
+      "step": 37465
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8832620373970326e-05,
+      "loss": 0.6569,
+      "step": 37470
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.883209657521639e-05,
+      "loss": 0.6601,
+      "step": 37475
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.883157266626324e-05,
+      "loss": 0.6489,
+      "step": 37480
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8831048647117415e-05,
+      "loss": 0.659,
+      "step": 37485
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.883052451778545e-05,
+      "loss": 0.6613,
+      "step": 37490
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.883000027827389e-05,
+      "loss": 0.6612,
+      "step": 37495
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8829475928589272e-05,
+      "loss": 0.6553,
+      "step": 37500
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8828951468738138e-05,
+      "loss": 0.6827,
+      "step": 37505
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8828426898727035e-05,
+      "loss": 0.6611,
+      "step": 37510
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8827902218562505e-05,
+      "loss": 0.666,
+      "step": 37515
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8827377428251096e-05,
+      "loss": 0.671,
+      "step": 37520
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8826852527799355e-05,
+      "loss": 0.6716,
+      "step": 37525
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8826327517213835e-05,
+      "loss": 0.6727,
+      "step": 37530
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.882580239650108e-05,
+      "loss": 0.6767,
+      "step": 37535
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.882527716566765e-05,
+      "loss": 0.6731,
+      "step": 37540
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8824751824720086e-05,
+      "loss": 0.6648,
+      "step": 37545
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.882422637366496e-05,
+      "loss": 0.6685,
+      "step": 37550
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.882370081250881e-05,
+      "loss": 0.6575,
+      "step": 37555
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.882317514125821e-05,
+      "loss": 0.6764,
+      "step": 37560
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8822649359919704e-05,
+      "loss": 0.6693,
+      "step": 37565
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8822123468499866e-05,
+      "loss": 0.6562,
+      "step": 37570
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8821597467005247e-05,
+      "loss": 0.6586,
+      "step": 37575
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8821071355442417e-05,
+      "loss": 0.6643,
+      "step": 37580
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8820545133817935e-05,
+      "loss": 0.6618,
+      "step": 37585
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.882001880213837e-05,
+      "loss": 0.6608,
+      "step": 37590
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8819492360410286e-05,
+      "loss": 0.6822,
+      "step": 37595
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8818965808640255e-05,
+      "loss": 0.6672,
+      "step": 37600
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8818439146834845e-05,
+      "loss": 0.6629,
+      "step": 37605
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.881791237500063e-05,
+      "loss": 0.6637,
+      "step": 37610
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.881738549314418e-05,
+      "loss": 0.6562,
+      "step": 37615
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.881685850127207e-05,
+      "loss": 0.6621,
+      "step": 37620
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.881633139939087e-05,
+      "loss": 0.6602,
+      "step": 37625
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8815804187507166e-05,
+      "loss": 0.6609,
+      "step": 37630
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.881527686562753e-05,
+      "loss": 0.6623,
+      "step": 37635
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8814749433758544e-05,
+      "loss": 0.6662,
+      "step": 37640
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8814221891906785e-05,
+      "loss": 0.6606,
+      "step": 37645
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.881369424007884e-05,
+      "loss": 0.6678,
+      "step": 37650
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.881316647828129e-05,
+      "loss": 0.6612,
+      "step": 37655
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.881263860652072e-05,
+      "loss": 0.677,
+      "step": 37660
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.881211062480372e-05,
+      "loss": 0.6778,
+      "step": 37665
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.881158253313687e-05,
+      "loss": 0.6561,
+      "step": 37670
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.881105433152677e-05,
+      "loss": 0.6529,
+      "step": 37675
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8810526019979996e-05,
+      "loss": 0.6774,
+      "step": 37680
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8809997598503153e-05,
+      "loss": 0.6652,
+      "step": 37685
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.880946906710283e-05,
+      "loss": 0.6537,
+      "step": 37690
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.880894042578562e-05,
+      "loss": 0.6655,
+      "step": 37695
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8808411674558114e-05,
+      "loss": 0.6624,
+      "step": 37700
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.880788281342692e-05,
+      "loss": 0.6592,
+      "step": 37705
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.880735384239863e-05,
+      "loss": 0.6626,
+      "step": 37710
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8806824761479845e-05,
+      "loss": 0.6726,
+      "step": 37715
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.880629557067717e-05,
+      "loss": 0.6786,
+      "step": 37720
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8805766269997203e-05,
+      "loss": 0.679,
+      "step": 37725
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8805236859446546e-05,
+      "loss": 0.6692,
+      "step": 37730
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8804707339031814e-05,
+      "loss": 0.6659,
+      "step": 37735
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8804177708759605e-05,
+      "loss": 0.6757,
+      "step": 37740
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.880364796863653e-05,
+      "loss": 0.6506,
+      "step": 37745
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8803118118669203e-05,
+      "loss": 0.6731,
+      "step": 37750
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.880258815886423e-05,
+      "loss": 0.6593,
+      "step": 37755
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8802058089228222e-05,
+      "loss": 0.6665,
+      "step": 37760
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.88015279097678e-05,
+      "loss": 0.6677,
+      "step": 37765
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.880099762048957e-05,
+      "loss": 0.6765,
+      "step": 37770
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8800467221400156e-05,
+      "loss": 0.6601,
+      "step": 37775
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.879993671250617e-05,
+      "loss": 0.6703,
+      "step": 37780
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8799406093814237e-05,
+      "loss": 0.6605,
+      "step": 37785
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8798875365330974e-05,
+      "loss": 0.6661,
+      "step": 37790
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8798344527063e-05,
+      "loss": 0.6686,
+      "step": 37795
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8797813579016948e-05,
+      "loss": 0.6711,
+      "step": 37800
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8797282521199428e-05,
+      "loss": 0.6603,
+      "step": 37805
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.879675135361708e-05,
+      "loss": 0.6635,
+      "step": 37810
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.879622007627653e-05,
+      "loss": 0.6625,
+      "step": 37815
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8795688689184398e-05,
+      "loss": 0.6635,
+      "step": 37820
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.879515719234732e-05,
+      "loss": 0.669,
+      "step": 37825
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8794625585771926e-05,
+      "loss": 0.6568,
+      "step": 37830
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.879409386946485e-05,
+      "loss": 0.6544,
+      "step": 37835
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8793562043432728e-05,
+      "loss": 0.6511,
+      "step": 37840
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.879303010768219e-05,
+      "loss": 0.6674,
+      "step": 37845
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.879249806221988e-05,
+      "loss": 0.661,
+      "step": 37850
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8791965907052435e-05,
+      "loss": 0.6741,
+      "step": 37855
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8791433642186487e-05,
+      "loss": 0.6525,
+      "step": 37860
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8790901267628685e-05,
+      "loss": 0.6632,
+      "step": 37865
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.879036878338567e-05,
+      "loss": 0.6603,
+      "step": 37870
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.878983618946409e-05,
+      "loss": 0.6649,
+      "step": 37875
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8789303485870575e-05,
+      "loss": 0.6704,
+      "step": 37880
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.878877067261179e-05,
+      "loss": 0.6616,
+      "step": 37885
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8788237749694373e-05,
+      "loss": 0.6682,
+      "step": 37890
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8787704717124975e-05,
+      "loss": 0.6628,
+      "step": 37895
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.878717157491025e-05,
+      "loss": 0.6705,
+      "step": 37900
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8786638323056846e-05,
+      "loss": 0.6772,
+      "step": 37905
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8786104961571413e-05,
+      "loss": 0.6697,
+      "step": 37910
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8785571490460617e-05,
+      "loss": 0.655,
+      "step": 37915
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8785037909731104e-05,
+      "loss": 0.6614,
+      "step": 37920
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.878450421938954e-05,
+      "loss": 0.6571,
+      "step": 37925
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8783970419442573e-05,
+      "loss": 0.6669,
+      "step": 37930
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8783436509896872e-05,
+      "loss": 0.6587,
+      "step": 37935
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.87829024907591e-05,
+      "loss": 0.656,
+      "step": 37940
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8782368362035913e-05,
+      "loss": 0.6607,
+      "step": 37945
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8781834123733977e-05,
+      "loss": 0.6653,
+      "step": 37950
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8781299775859964e-05,
+      "loss": 0.6724,
+      "step": 37955
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.878076531842053e-05,
+      "loss": 0.6677,
+      "step": 37960
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8780230751422354e-05,
+      "loss": 0.6579,
+      "step": 37965
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8779696074872103e-05,
+      "loss": 0.6601,
+      "step": 37970
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8779161288776444e-05,
+      "loss": 0.6643,
+      "step": 37975
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8778626393142053e-05,
+      "loss": 0.6705,
+      "step": 37980
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.877809138797561e-05,
+      "loss": 0.6681,
+      "step": 37985
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8777556273283774e-05,
+      "loss": 0.6751,
+      "step": 37990
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8777021049073237e-05,
+      "loss": 0.6637,
+      "step": 37995
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8776485715350672e-05,
+      "loss": 0.6705,
+      "step": 38000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8775950272122756e-05,
+      "loss": 0.6715,
+      "step": 38005
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8775414719396173e-05,
+      "loss": 0.6708,
+      "step": 38010
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8774879057177605e-05,
+      "loss": 0.6757,
+      "step": 38015
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8774343285473734e-05,
+      "loss": 0.6448,
+      "step": 38020
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8773807404291244e-05,
+      "loss": 0.6534,
+      "step": 38025
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8773271413636826e-05,
+      "loss": 0.6551,
+      "step": 38030
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8772735313517163e-05,
+      "loss": 0.6626,
+      "step": 38035
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8772199103938946e-05,
+      "loss": 0.6646,
+      "step": 38040
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8771662784908865e-05,
+      "loss": 0.6468,
+      "step": 38045
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8771126356433607e-05,
+      "loss": 0.6582,
+      "step": 38050
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8770589818519873e-05,
+      "loss": 0.6537,
+      "step": 38055
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8770053171174357e-05,
+      "loss": 0.6702,
+      "step": 38060
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.876951641440375e-05,
+      "loss": 0.664,
+      "step": 38065
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.876897954821475e-05,
+      "loss": 0.6487,
+      "step": 38070
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8768442572614055e-05,
+      "loss": 0.6801,
+      "step": 38075
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8767905487608367e-05,
+      "loss": 0.6588,
+      "step": 38080
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.876736829320439e-05,
+      "loss": 0.6749,
+      "step": 38085
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8766830989408818e-05,
+      "loss": 0.6577,
+      "step": 38090
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8766293576228363e-05,
+      "loss": 0.6626,
+      "step": 38095
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.876575605366973e-05,
+      "loss": 0.6783,
+      "step": 38100
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8765218421739623e-05,
+      "loss": 0.6616,
+      "step": 38105
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8764680680444747e-05,
+      "loss": 0.6707,
+      "step": 38110
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.876414282979182e-05,
+      "loss": 0.6489,
+      "step": 38115
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8763604869787544e-05,
+      "loss": 0.6626,
+      "step": 38120
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8763066800438638e-05,
+      "loss": 0.6461,
+      "step": 38125
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.876252862175181e-05,
+      "loss": 0.6674,
+      "step": 38130
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.876199033373378e-05,
+      "loss": 0.6678,
+      "step": 38135
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8761451936391263e-05,
+      "loss": 0.657,
+      "step": 38140
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8760913429730972e-05,
+      "loss": 0.659,
+      "step": 38145
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.876037481375963e-05,
+      "loss": 0.6634,
+      "step": 38150
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.875983608848396e-05,
+      "loss": 0.6502,
+      "step": 38155
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.875929725391068e-05,
+      "loss": 0.6657,
+      "step": 38160
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8758758310046513e-05,
+      "loss": 0.6815,
+      "step": 38165
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.875821925689818e-05,
+      "loss": 0.6665,
+      "step": 38170
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8757680094472417e-05,
+      "loss": 0.655,
+      "step": 38175
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8757140822775943e-05,
+      "loss": 0.6588,
+      "step": 38180
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.875660144181549e-05,
+      "loss": 0.6653,
+      "step": 38185
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8756061951597787e-05,
+      "loss": 0.6613,
+      "step": 38190
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8755522352129564e-05,
+      "loss": 0.6574,
+      "step": 38195
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8754982643417553e-05,
+      "loss": 0.661,
+      "step": 38200
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8754442825468494e-05,
+      "loss": 0.6584,
+      "step": 38205
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8753902898289115e-05,
+      "loss": 0.6501,
+      "step": 38210
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8753362861886156e-05,
+      "loss": 0.6593,
+      "step": 38215
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8752822716266356e-05,
+      "loss": 0.6586,
+      "step": 38220
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8752282461436456e-05,
+      "loss": 0.6653,
+      "step": 38225
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.875174209740319e-05,
+      "loss": 0.662,
+      "step": 38230
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8751201624173304e-05,
+      "loss": 0.6799,
+      "step": 38235
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8750661041753547e-05,
+      "loss": 0.6633,
+      "step": 38240
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8750120350150657e-05,
+      "loss": 0.6674,
+      "step": 38245
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.874957954937138e-05,
+      "loss": 0.6512,
+      "step": 38250
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8749038639422465e-05,
+      "loss": 0.6611,
+      "step": 38255
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8748497620310668e-05,
+      "loss": 0.6675,
+      "step": 38260
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8747956492042724e-05,
+      "loss": 0.6695,
+      "step": 38265
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.87474152546254e-05,
+      "loss": 0.6564,
+      "step": 38270
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.8746873908065443e-05,
+      "loss": 0.6539,
+      "step": 38275
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.874633245236961e-05,
+      "loss": 0.6755,
+      "step": 38280
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.874579088754465e-05,
+      "loss": 0.6609,
+      "step": 38285
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8745249213597324e-05,
+      "loss": 0.6671,
+      "step": 38290
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8744707430534395e-05,
+      "loss": 0.6662,
+      "step": 38295
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8744165538362615e-05,
+      "loss": 0.6609,
+      "step": 38300
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.874362353708875e-05,
+      "loss": 0.6586,
+      "step": 38305
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8743081426719564e-05,
+      "loss": 0.6593,
+      "step": 38310
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8742539207261815e-05,
+      "loss": 0.6502,
+      "step": 38315
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8741996878722276e-05,
+      "loss": 0.669,
+      "step": 38320
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.874145444110771e-05,
+      "loss": 0.6654,
+      "step": 38325
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8740911894424883e-05,
+      "loss": 0.683,
+      "step": 38330
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8740369238680565e-05,
+      "loss": 0.6623,
+      "step": 38335
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.873982647388153e-05,
+      "loss": 0.6614,
+      "step": 38340
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.873928360003455e-05,
+      "loss": 0.6567,
+      "step": 38345
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8738740617146396e-05,
+      "loss": 0.6691,
+      "step": 38350
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.873819752522384e-05,
+      "loss": 0.6663,
+      "step": 38355
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8737654324273668e-05,
+      "loss": 0.6589,
+      "step": 38360
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8737111014302643e-05,
+      "loss": 0.652,
+      "step": 38365
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.873656759531756e-05,
+      "loss": 0.6556,
+      "step": 38370
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8736024067325188e-05,
+      "loss": 0.6596,
+      "step": 38375
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8735480430332313e-05,
+      "loss": 0.6643,
+      "step": 38380
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.873493668434572e-05,
+      "loss": 0.6706,
+      "step": 38385
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8734392829372188e-05,
+      "loss": 0.6728,
+      "step": 38390
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8733848865418504e-05,
+      "loss": 0.6574,
+      "step": 38395
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.873330479249146e-05,
+      "loss": 0.67,
+      "step": 38400
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.873276061059784e-05,
+      "loss": 0.6684,
+      "step": 38405
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8732216319744435e-05,
+      "loss": 0.6602,
+      "step": 38410
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8731671919938038e-05,
+      "loss": 0.6575,
+      "step": 38415
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8731127411185434e-05,
+      "loss": 0.6593,
+      "step": 38420
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.873058279349343e-05,
+      "loss": 0.6599,
+      "step": 38425
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.873003806686881e-05,
+      "loss": 0.6704,
+      "step": 38430
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.872949323131837e-05,
+      "loss": 0.6604,
+      "step": 38435
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.872894828684892e-05,
+      "loss": 0.6627,
+      "step": 38440
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8728403233467247e-05,
+      "loss": 0.6615,
+      "step": 38445
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8727858071180162e-05,
+      "loss": 0.6539,
+      "step": 38450
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8727312799994455e-05,
+      "loss": 0.6499,
+      "step": 38455
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.872676741991694e-05,
+      "loss": 0.6606,
+      "step": 38460
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8726221930954418e-05,
+      "loss": 0.6542,
+      "step": 38465
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8725676333113694e-05,
+      "loss": 0.6725,
+      "step": 38470
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8725130626401573e-05,
+      "loss": 0.6561,
+      "step": 38475
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8724584810824874e-05,
+      "loss": 0.65,
+      "step": 38480
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8724038886390394e-05,
+      "loss": 0.6603,
+      "step": 38485
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8723492853104956e-05,
+      "loss": 0.658,
+      "step": 38490
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8722946710975365e-05,
+      "loss": 0.6573,
+      "step": 38495
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8722400460008437e-05,
+      "loss": 0.658,
+      "step": 38500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8721854100210993e-05,
+      "loss": 0.6656,
+      "step": 38505
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8721307631589845e-05,
+      "loss": 0.6667,
+      "step": 38510
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.872076105415181e-05,
+      "loss": 0.6706,
+      "step": 38515
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.872021436790371e-05,
+      "loss": 0.6648,
+      "step": 38520
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8719667572852366e-05,
+      "loss": 0.6553,
+      "step": 38525
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8719120669004606e-05,
+      "loss": 0.6681,
+      "step": 38530
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.871857365636724e-05,
+      "loss": 0.6592,
+      "step": 38535
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8718026534947107e-05,
+      "loss": 0.663,
+      "step": 38540
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8717479304751024e-05,
+      "loss": 0.6634,
+      "step": 38545
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8716931965785826e-05,
+      "loss": 0.6615,
+      "step": 38550
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.871638451805834e-05,
+      "loss": 0.6674,
+      "step": 38555
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.871583696157539e-05,
+      "loss": 0.6662,
+      "step": 38560
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8715289296343818e-05,
+      "loss": 0.6581,
+      "step": 38565
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.8714741522370454e-05,
+      "loss": 0.6572,
+      "step": 38570
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.871419363966213e-05,
+      "loss": 0.6762,
+      "step": 38575
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.871364564822568e-05,
+      "loss": 0.6665,
+      "step": 38580
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.871309754806795e-05,
+      "loss": 0.6583,
+      "step": 38585
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.871254933919577e-05,
+      "loss": 0.6664,
+      "step": 38590
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8712001021615985e-05,
+      "loss": 0.6658,
+      "step": 38595
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8711452595335434e-05,
+      "loss": 0.6658,
+      "step": 38600
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8710904060360965e-05,
+      "loss": 0.6618,
+      "step": 38605
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8710355416699413e-05,
+      "loss": 0.6688,
+      "step": 38610
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.870980666435763e-05,
+      "loss": 0.656,
+      "step": 38615
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.870925780334246e-05,
+      "loss": 0.6697,
+      "step": 38620
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8708708833660755e-05,
+      "loss": 0.6767,
+      "step": 38625
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8708159755319363e-05,
+      "loss": 0.6726,
+      "step": 38630
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.870761056832513e-05,
+      "loss": 0.6723,
+      "step": 38635
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8707061272684916e-05,
+      "loss": 0.6682,
+      "step": 38640
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.870651186840557e-05,
+      "loss": 0.6649,
+      "step": 38645
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8705962355493944e-05,
+      "loss": 0.6705,
+      "step": 38650
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8705412733956903e-05,
+      "loss": 0.674,
+      "step": 38655
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8704863003801297e-05,
+      "loss": 0.6608,
+      "step": 38660
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.870431316503399e-05,
+      "loss": 0.6588,
+      "step": 38665
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.870376321766184e-05,
+      "loss": 0.6604,
+      "step": 38670
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8703213161691708e-05,
+      "loss": 0.6571,
+      "step": 38675
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.870266299713046e-05,
+      "loss": 0.6698,
+      "step": 38680
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8702112723984957e-05,
+      "loss": 0.6684,
+      "step": 38685
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.870156234226207e-05,
+      "loss": 0.6546,
+      "step": 38690
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.870101185196866e-05,
+      "loss": 0.6501,
+      "step": 38695
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.87004612531116e-05,
+      "loss": 0.6636,
+      "step": 38700
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.869991054569776e-05,
+      "loss": 0.6561,
+      "step": 38705
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8699359729734002e-05,
+      "loss": 0.6627,
+      "step": 38710
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8698808805227212e-05,
+      "loss": 0.6626,
+      "step": 38715
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8698257772184253e-05,
+      "loss": 0.6612,
+      "step": 38720
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8697706630612013e-05,
+      "loss": 0.6603,
+      "step": 38725
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8697155380517356e-05,
+      "loss": 0.6589,
+      "step": 38730
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8696604021907166e-05,
+      "loss": 0.6737,
+      "step": 38735
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.869605255478832e-05,
+      "loss": 0.6523,
+      "step": 38740
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8695500979167703e-05,
+      "loss": 0.6549,
+      "step": 38745
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.869494929505219e-05,
+      "loss": 0.6604,
+      "step": 38750
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8694397502448672e-05,
+      "loss": 0.6584,
+      "step": 38755
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.869384560136403e-05,
+      "loss": 0.666,
+      "step": 38760
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.869329359180515e-05,
+      "loss": 0.6549,
+      "step": 38765
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.869274147377892e-05,
+      "loss": 0.6607,
+      "step": 38770
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.869218924729223e-05,
+      "loss": 0.6671,
+      "step": 38775
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.869163691235197e-05,
+      "loss": 0.6665,
+      "step": 38780
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8691084468965027e-05,
+      "loss": 0.6458,
+      "step": 38785
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8690531917138298e-05,
+      "loss": 0.672,
+      "step": 38790
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.868997925687868e-05,
+      "loss": 0.6534,
+      "step": 38795
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8689426488193066e-05,
+      "loss": 0.6583,
+      "step": 38800
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8688873611088348e-05,
+      "loss": 0.6562,
+      "step": 38805
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8688320625571433e-05,
+      "loss": 0.6566,
+      "step": 38810
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8687767531649212e-05,
+      "loss": 0.6646,
+      "step": 38815
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8687214329328593e-05,
+      "loss": 0.6717,
+      "step": 38820
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8686661018616478e-05,
+      "loss": 0.6579,
+      "step": 38825
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8686107599519766e-05,
+      "loss": 0.6565,
+      "step": 38830
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8685554072045367e-05,
+      "loss": 0.665,
+      "step": 38835
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.868500043620018e-05,
+      "loss": 0.6673,
+      "step": 38840
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.868444669199112e-05,
+      "loss": 0.6566,
+      "step": 38845
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8683892839425098e-05,
+      "loss": 0.6648,
+      "step": 38850
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8683338878509015e-05,
+      "loss": 0.6606,
+      "step": 38855
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8682784809249792e-05,
+      "loss": 0.6537,
+      "step": 38860
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8682230631654333e-05,
+      "loss": 0.658,
+      "step": 38865
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8681676345729564e-05,
+      "loss": 0.6636,
+      "step": 38870
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8681121951482397e-05,
+      "loss": 0.674,
+      "step": 38875
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.868056744891974e-05,
+      "loss": 0.654,
+      "step": 38880
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8680012838048523e-05,
+      "loss": 0.6515,
+      "step": 38885
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8679458118875658e-05,
+      "loss": 0.6698,
+      "step": 38890
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.8678903291408075e-05,
+      "loss": 0.6554,
+      "step": 38895
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.867834835565269e-05,
+      "loss": 0.6552,
+      "step": 38900
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.867779331161643e-05,
+      "loss": 0.6598,
+      "step": 38905
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.867723815930621e-05,
+      "loss": 0.6534,
+      "step": 38910
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.867668289872898e-05,
+      "loss": 0.6705,
+      "step": 38915
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8676127529891645e-05,
+      "loss": 0.6722,
+      "step": 38920
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8675572052801145e-05,
+      "loss": 0.6598,
+      "step": 38925
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.867501646746441e-05,
+      "loss": 0.6549,
+      "step": 38930
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.867446077388837e-05,
+      "loss": 0.6567,
+      "step": 38935
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8673904972079956e-05,
+      "loss": 0.6601,
+      "step": 38940
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8673349062046112e-05,
+      "loss": 0.6528,
+      "step": 38945
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8672793043793768e-05,
+      "loss": 0.6718,
+      "step": 38950
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8672236917329857e-05,
+      "loss": 0.6638,
+      "step": 38955
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.867168068266133e-05,
+      "loss": 0.6547,
+      "step": 38960
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8671124339795118e-05,
+      "loss": 0.6668,
+      "step": 38965
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.867056788873816e-05,
+      "loss": 0.6615,
+      "step": 38970
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8670011329497408e-05,
+      "loss": 0.6676,
+      "step": 38975
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8669454662079802e-05,
+      "loss": 0.6562,
+      "step": 38980
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8668897886492284e-05,
+      "loss": 0.6605,
+      "step": 38985
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.866834100274181e-05,
+      "loss": 0.6505,
+      "step": 38990
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.866778401083532e-05,
+      "loss": 0.6608,
+      "step": 38995
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8667226910779767e-05,
+      "loss": 0.6595,
+      "step": 39000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8666669702582097e-05,
+      "loss": 0.6536,
+      "step": 39005
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.866611238624927e-05,
+      "loss": 0.6632,
+      "step": 39010
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8665554961788236e-05,
+      "loss": 0.6641,
+      "step": 39015
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.866499742920595e-05,
+      "loss": 0.6623,
+      "step": 39020
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.866443978850937e-05,
+      "loss": 0.6665,
+      "step": 39025
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8663882039705456e-05,
+      "loss": 0.6619,
+      "step": 39030
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.866332418280116e-05,
+      "loss": 0.6472,
+      "step": 39035
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8662766217803447e-05,
+      "loss": 0.6646,
+      "step": 39040
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8662208144719276e-05,
+      "loss": 0.6627,
+      "step": 39045
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8661649963555616e-05,
+      "loss": 0.6633,
+      "step": 39050
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8661091674319425e-05,
+      "loss": 0.661,
+      "step": 39055
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8660533277017672e-05,
+      "loss": 0.6558,
+      "step": 39060
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8659974771657325e-05,
+      "loss": 0.6563,
+      "step": 39065
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8659416158245348e-05,
+      "loss": 0.6509,
+      "step": 39070
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.865885743678872e-05,
+      "loss": 0.6507,
+      "step": 39075
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8658298607294402e-05,
+      "loss": 0.6634,
+      "step": 39080
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.865773966976937e-05,
+      "loss": 0.6622,
+      "step": 39085
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8657180624220602e-05,
+      "loss": 0.6491,
+      "step": 39090
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.865662147065507e-05,
+      "loss": 0.6735,
+      "step": 39095
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.865606220907975e-05,
+      "loss": 0.6561,
+      "step": 39100
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.865550283950162e-05,
+      "loss": 0.6685,
+      "step": 39105
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8654943361927662e-05,
+      "loss": 0.6717,
+      "step": 39110
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8654383776364855e-05,
+      "loss": 0.6594,
+      "step": 39115
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.865382408282018e-05,
+      "loss": 0.658,
+      "step": 39120
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8653264281300622e-05,
+      "loss": 0.6647,
+      "step": 39125
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8652704371813165e-05,
+      "loss": 0.6614,
+      "step": 39130
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8652144354364793e-05,
+      "loss": 0.6768,
+      "step": 39135
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8651584228962495e-05,
+      "loss": 0.6687,
+      "step": 39140
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.865102399561326e-05,
+      "loss": 0.661,
+      "step": 39145
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.865046365432408e-05,
+      "loss": 0.6569,
+      "step": 39150
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8649903205101948e-05,
+      "loss": 0.6555,
+      "step": 39155
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.864934264795385e-05,
+      "loss": 0.6501,
+      "step": 39160
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8648781982886784e-05,
+      "loss": 0.659,
+      "step": 39165
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8648221209907744e-05,
+      "loss": 0.6599,
+      "step": 39170
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.864766032902373e-05,
+      "loss": 0.6573,
+      "step": 39175
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8647099340241738e-05,
+      "loss": 0.6638,
+      "step": 39180
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8646538243568765e-05,
+      "loss": 0.6531,
+      "step": 39185
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8645977039011818e-05,
+      "loss": 0.6683,
+      "step": 39190
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8645415726577893e-05,
+      "loss": 0.655,
+      "step": 39195
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8644854306273997e-05,
+      "loss": 0.6621,
+      "step": 39200
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8644292778107137e-05,
+      "loss": 0.6662,
+      "step": 39205
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.8643731142084312e-05,
+      "loss": 0.6621,
+      "step": 39210
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.864316939821254e-05,
+      "loss": 0.6528,
+      "step": 39215
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.864260754649882e-05,
+      "loss": 0.6424,
+      "step": 39220
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8642045586950165e-05,
+      "loss": 0.6567,
+      "step": 39225
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8641483519573592e-05,
+      "loss": 0.6485,
+      "step": 39230
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.864092134437611e-05,
+      "loss": 0.6682,
+      "step": 39235
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.864035906136473e-05,
+      "loss": 0.662,
+      "step": 39240
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8639796670546473e-05,
+      "loss": 0.6639,
+      "step": 39245
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8639234171928355e-05,
+      "loss": 0.6642,
+      "step": 39250
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8638671565517392e-05,
+      "loss": 0.6725,
+      "step": 39255
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8638108851320607e-05,
+      "loss": 0.6624,
+      "step": 39260
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8637546029345018e-05,
+      "loss": 0.6687,
+      "step": 39265
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8636983099597646e-05,
+      "loss": 0.6756,
+      "step": 39270
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.863642006208552e-05,
+      "loss": 0.6533,
+      "step": 39275
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8635856916815664e-05,
+      "loss": 0.668,
+      "step": 39280
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.86352936637951e-05,
+      "loss": 0.6633,
+      "step": 39285
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.863473030303086e-05,
+      "loss": 0.6741,
+      "step": 39290
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.863416683452997e-05,
+      "loss": 0.6575,
+      "step": 39295
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8633603258299464e-05,
+      "loss": 0.6607,
+      "step": 39300
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8633039574346373e-05,
+      "loss": 0.6596,
+      "step": 39305
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.863247578267773e-05,
+      "loss": 0.6662,
+      "step": 39310
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8631911883300564e-05,
+      "loss": 0.6689,
+      "step": 39315
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.863134787622192e-05,
+      "loss": 0.6582,
+      "step": 39320
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8630783761448828e-05,
+      "loss": 0.6562,
+      "step": 39325
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8630219538988334e-05,
+      "loss": 0.6632,
+      "step": 39330
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8629655208847465e-05,
+      "loss": 0.6636,
+      "step": 39335
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8629090771033275e-05,
+      "loss": 0.6609,
+      "step": 39340
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8628526225552806e-05,
+      "loss": 0.6636,
+      "step": 39345
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8627961572413094e-05,
+      "loss": 0.6714,
+      "step": 39350
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.862739681162119e-05,
+      "loss": 0.6574,
+      "step": 39355
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.862683194318413e-05,
+      "loss": 0.6606,
+      "step": 39360
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.862626696710898e-05,
+      "loss": 0.6605,
+      "step": 39365
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8625701883402776e-05,
+      "loss": 0.6633,
+      "step": 39370
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8625136692072577e-05,
+      "loss": 0.6419,
+      "step": 39375
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8624571393125425e-05,
+      "loss": 0.664,
+      "step": 39380
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8624005986568385e-05,
+      "loss": 0.6578,
+      "step": 39385
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8623440472408497e-05,
+      "loss": 0.6477,
+      "step": 39390
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8622874850652834e-05,
+      "loss": 0.6674,
+      "step": 39395
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.862230912130844e-05,
+      "loss": 0.6561,
+      "step": 39400
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.862174328438238e-05,
+      "loss": 0.6549,
+      "step": 39405
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8621177339881712e-05,
+      "loss": 0.6442,
+      "step": 39410
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.86206112878135e-05,
+      "loss": 0.6687,
+      "step": 39415
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.86200451281848e-05,
+      "loss": 0.6558,
+      "step": 39420
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8619478861002682e-05,
+      "loss": 0.6692,
+      "step": 39425
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8618912486274214e-05,
+      "loss": 0.6553,
+      "step": 39430
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8618346004006456e-05,
+      "loss": 0.6594,
+      "step": 39435
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.861777941420648e-05,
+      "loss": 0.6524,
+      "step": 39440
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.861721271688135e-05,
+      "loss": 0.663,
+      "step": 39445
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8616645912038146e-05,
+      "loss": 0.6586,
+      "step": 39450
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.861607899968393e-05,
+      "loss": 0.6665,
+      "step": 39455
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8615511979825784e-05,
+      "loss": 0.6594,
+      "step": 39460
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.861494485247078e-05,
+      "loss": 0.6615,
+      "step": 39465
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8614377617625993e-05,
+      "loss": 0.6606,
+      "step": 39470
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.86138102752985e-05,
+      "loss": 0.6627,
+      "step": 39475
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.861324282549538e-05,
+      "loss": 0.6571,
+      "step": 39480
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8612675268223717e-05,
+      "loss": 0.6481,
+      "step": 39485
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.861210760349059e-05,
+      "loss": 0.6571,
+      "step": 39490
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8611539831303074e-05,
+      "loss": 0.6564,
+      "step": 39495
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8610971951668265e-05,
+      "loss": 0.654,
+      "step": 39500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.861040396459324e-05,
+      "loss": 0.6607,
+      "step": 39505
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8609835870085096e-05,
+      "loss": 0.6527,
+      "step": 39510
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8609267668150913e-05,
+      "loss": 0.6537,
+      "step": 39515
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.860869935879778e-05,
+      "loss": 0.6512,
+      "step": 39520
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8608130942032792e-05,
+      "loss": 0.6716,
+      "step": 39525
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.8607562417863038e-05,
+      "loss": 0.6491,
+      "step": 39530
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.8606993786295614e-05,
+      "loss": 0.6617,
+      "step": 39535
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.8606425047337616e-05,
+      "loss": 0.6587,
+      "step": 39540
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.8605856200996135e-05,
+      "loss": 0.6489,
+      "step": 39545
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.8605287247278273e-05,
+      "loss": 0.665,
+      "step": 39550
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.8604718186191128e-05,
+      "loss": 0.6683,
+      "step": 39555
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.86041490177418e-05,
+      "loss": 0.6449,
+      "step": 39560
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.8603579741937387e-05,
+      "loss": 0.6628,
+      "step": 39565
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.8603010358785e-05,
+      "loss": 0.6584,
+      "step": 39570
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.8602440868291736e-05,
+      "loss": 0.6451,
+      "step": 39575
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.8601871270464708e-05,
+      "loss": 0.6539,
+      "step": 39580
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.8601301565311012e-05,
+      "loss": 0.668,
+      "step": 39585
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.8600731752837765e-05,
+      "loss": 0.6609,
+      "step": 39590
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.8600161833052074e-05,
+      "loss": 0.6563,
+      "step": 39595
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.8599591805961047e-05,
+      "loss": 0.6634,
+      "step": 39600
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.8599021671571804e-05,
+      "loss": 0.66,
+      "step": 39605
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.8598451429891455e-05,
+      "loss": 0.6632,
+      "step": 39610
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.8597881080927108e-05,
+      "loss": 0.6682,
+      "step": 39615
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.859731062468589e-05,
+      "loss": 0.642,
+      "step": 39620
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.8596740061174912e-05,
+      "loss": 0.6611,
+      "step": 39625
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.85961693904013e-05,
+      "loss": 0.6604,
+      "step": 39630
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.8595598612372162e-05,
+      "loss": 0.6409,
+      "step": 39635
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.8595027727094628e-05,
+      "loss": 0.6629,
+      "step": 39640
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.8594456734575824e-05,
+      "loss": 0.6534,
+      "step": 39645
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.8593885634822866e-05,
+      "loss": 0.6495,
+      "step": 39650
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.8593314427842886e-05,
+      "loss": 0.657,
+      "step": 39655
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.859274311364301e-05,
+      "loss": 0.6561,
+      "step": 39660
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.8592171692230367e-05,
+      "loss": 0.6553,
+      "step": 39665
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.859160016361208e-05,
+      "loss": 0.6748,
+      "step": 39670
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.8591028527795288e-05,
+      "loss": 0.658,
+      "step": 39675
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.8590456784787122e-05,
+      "loss": 0.6571,
+      "step": 39680
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.8589884934594713e-05,
+      "loss": 0.6661,
+      "step": 39685
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.8589312977225196e-05,
+      "loss": 0.6674,
+      "step": 39690
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.858874091268571e-05,
+      "loss": 0.6682,
+      "step": 39695
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.858816874098339e-05,
+      "loss": 0.6663,
+      "step": 39700
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.8587596462125378e-05,
+      "loss": 0.66,
+      "step": 39705
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.8587024076118808e-05,
+      "loss": 0.6686,
+      "step": 39710
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.8586451582970834e-05,
+      "loss": 0.6664,
+      "step": 39715
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.8585878982688588e-05,
+      "loss": 0.6614,
+      "step": 39720
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.858530627527922e-05,
+      "loss": 0.6523,
+      "step": 39725
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.8584733460749872e-05,
+      "loss": 0.658,
+      "step": 39730
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.8584160539107693e-05,
+      "loss": 0.6575,
+      "step": 39735
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.858358751035983e-05,
+      "loss": 0.6556,
+      "step": 39740
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.858301437451344e-05,
+      "loss": 0.6525,
+      "step": 39745
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.8582441131575658e-05,
+      "loss": 0.6564,
+      "step": 39750
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.8581867781553654e-05,
+      "loss": 0.6519,
+      "step": 39755
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.8581294324454573e-05,
+      "loss": 0.6704,
+      "step": 39760
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.858072076028557e-05,
+      "loss": 0.6664,
+      "step": 39765
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.8580147089053805e-05,
+      "loss": 0.6683,
+      "step": 39770
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.857957331076643e-05,
+      "loss": 0.6616,
+      "step": 39775
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.857899942543061e-05,
+      "loss": 0.6494,
+      "step": 39780
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.8578425433053505e-05,
+      "loss": 0.6636,
+      "step": 39785
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.8577851333642273e-05,
+      "loss": 0.6616,
+      "step": 39790
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.8577277127204077e-05,
+      "loss": 0.6629,
+      "step": 39795
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.8576702813746087e-05,
+      "loss": 0.6681,
+      "step": 39800
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.8576128393275463e-05,
+      "loss": 0.6561,
+      "step": 39805
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.8575553865799376e-05,
+      "loss": 0.6578,
+      "step": 39810
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.8574979231324993e-05,
+      "loss": 0.6601,
+      "step": 39815
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.8574404489859483e-05,
+      "loss": 0.6592,
+      "step": 39820
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.857382964141002e-05,
+      "loss": 0.6611,
+      "step": 39825
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.857325468598377e-05,
+      "loss": 0.6469,
+      "step": 39830
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.8572679623587914e-05,
+      "loss": 0.6612,
+      "step": 39835
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.8572104454229626e-05,
+      "loss": 0.6564,
+      "step": 39840
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.8571529177916075e-05,
+      "loss": 0.6494,
+      "step": 39845
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.8570953794654446e-05,
+      "loss": 0.6672,
+      "step": 39850
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.857037830445192e-05,
+      "loss": 0.6577,
+      "step": 39855
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.856980270731567e-05,
+      "loss": 0.6522,
+      "step": 39860
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.8569227003252887e-05,
+      "loss": 0.6521,
+      "step": 39865
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.8568651192270743e-05,
+      "loss": 0.6592,
+      "step": 39870
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.856807527437643e-05,
+      "loss": 0.6617,
+      "step": 39875
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.8567499249577135e-05,
+      "loss": 0.6564,
+      "step": 39880
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.856692311788004e-05,
+      "loss": 0.6454,
+      "step": 39885
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.8566346879292337e-05,
+      "loss": 0.6685,
+      "step": 39890
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.8565770533821213e-05,
+      "loss": 0.6622,
+      "step": 39895
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.8565194081473863e-05,
+      "loss": 0.6516,
+      "step": 39900
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.8564617522257474e-05,
+      "loss": 0.6549,
+      "step": 39905
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.856404085617925e-05,
+      "loss": 0.6411,
+      "step": 39910
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.8563464083246372e-05,
+      "loss": 0.6598,
+      "step": 39915
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.8562887203466044e-05,
+      "loss": 0.6602,
+      "step": 39920
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.8562310216845463e-05,
+      "loss": 0.6661,
+      "step": 39925
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.8561733123391835e-05,
+      "loss": 0.6542,
+      "step": 39930
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.8561155923112347e-05,
+      "loss": 0.6583,
+      "step": 39935
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.8560578616014214e-05,
+      "loss": 0.6509,
+      "step": 39940
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.856000120210463e-05,
+      "loss": 0.6679,
+      "step": 39945
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.8559423681390803e-05,
+      "loss": 0.6515,
+      "step": 39950
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.8558846053879937e-05,
+      "loss": 0.6626,
+      "step": 39955
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.855826831957924e-05,
+      "loss": 0.6554,
+      "step": 39960
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.855769047849592e-05,
+      "loss": 0.6655,
+      "step": 39965
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.8557112530637195e-05,
+      "loss": 0.6664,
+      "step": 39970
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.855653447601026e-05,
+      "loss": 0.6438,
+      "step": 39975
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.855595631462234e-05,
+      "loss": 0.6497,
+      "step": 39980
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.8555378046480647e-05,
+      "loss": 0.6572,
+      "step": 39985
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.855479967159239e-05,
+      "loss": 0.6625,
+      "step": 39990
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.855422118996479e-05,
+      "loss": 0.6453,
+      "step": 39995
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.855364260160507e-05,
+      "loss": 0.6512,
+      "step": 40000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.855306390652044e-05,
+      "loss": 0.6533,
+      "step": 40005
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.8552485104718122e-05,
+      "loss": 0.6595,
+      "step": 40010
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.855190619620534e-05,
+      "loss": 0.6626,
+      "step": 40015
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.8551327180989318e-05,
+      "loss": 0.6617,
+      "step": 40020
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.8550748059077278e-05,
+      "loss": 0.6529,
+      "step": 40025
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.8550168830476448e-05,
+      "loss": 0.6612,
+      "step": 40030
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.8549589495194058e-05,
+      "loss": 0.657,
+      "step": 40035
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.854901005323733e-05,
+      "loss": 0.6517,
+      "step": 40040
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.8548430504613497e-05,
+      "loss": 0.6569,
+      "step": 40045
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.8547850849329788e-05,
+      "loss": 0.6534,
+      "step": 40050
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.8547271087393436e-05,
+      "loss": 0.6659,
+      "step": 40055
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.854669121881168e-05,
+      "loss": 0.6606,
+      "step": 40060
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.8546111243591746e-05,
+      "loss": 0.6534,
+      "step": 40065
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.854553116174088e-05,
+      "loss": 0.6672,
+      "step": 40070
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.854495097326631e-05,
+      "loss": 0.667,
+      "step": 40075
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.8544370678175282e-05,
+      "loss": 0.6579,
+      "step": 40080
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.8543790276475037e-05,
+      "loss": 0.6479,
+      "step": 40085
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.8543209768172815e-05,
+      "loss": 0.6524,
+      "step": 40090
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.8542629153275854e-05,
+      "loss": 0.6402,
+      "step": 40095
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.8542048431791404e-05,
+      "loss": 0.6499,
+      "step": 40100
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.854146760372671e-05,
+      "loss": 0.6651,
+      "step": 40105
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.854088666908902e-05,
+      "loss": 0.645,
+      "step": 40110
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.854030562788558e-05,
+      "loss": 0.66,
+      "step": 40115
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.853972448012364e-05,
+      "loss": 0.6553,
+      "step": 40120
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.8539143225810453e-05,
+      "loss": 0.6532,
+      "step": 40125
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.853856186495327e-05,
+      "loss": 0.6591,
+      "step": 40130
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.8537980397559342e-05,
+      "loss": 0.6607,
+      "step": 40135
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.853739882363593e-05,
+      "loss": 0.6504,
+      "step": 40140
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.8536817143190284e-05,
+      "loss": 0.6598,
+      "step": 40145
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.8536235356229667e-05,
+      "loss": 0.6638,
+      "step": 40150
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.8535653462761337e-05,
+      "loss": 0.6538,
+      "step": 40155
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.8535071462792557e-05,
+      "loss": 0.6494,
+      "step": 40160
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.8534489356330576e-05,
+      "loss": 0.6646,
+      "step": 40165
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.8533907143382672e-05,
+      "loss": 0.6619,
+      "step": 40170
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.8533324823956104e-05,
+      "loss": 0.6534,
+      "step": 40175
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.8532742398058134e-05,
+      "loss": 0.6508,
+      "step": 40180
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.8532159865696034e-05,
+      "loss": 0.6587,
+      "step": 40185
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.8531577226877068e-05,
+      "loss": 0.6555,
+      "step": 40190
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.853099448160851e-05,
+      "loss": 0.6615,
+      "step": 40195
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.8530411629897628e-05,
+      "loss": 0.6482,
+      "step": 40200
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.85298286717517e-05,
+      "loss": 0.6528,
+      "step": 40205
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.852924560717799e-05,
+      "loss": 0.6524,
+      "step": 40210
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.852866243618378e-05,
+      "loss": 0.654,
+      "step": 40215
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.8528079158776342e-05,
+      "loss": 0.659,
+      "step": 40220
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.852749577496296e-05,
+      "loss": 0.6551,
+      "step": 40225
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.8526912284750908e-05,
+      "loss": 0.6445,
+      "step": 40230
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.8526328688147465e-05,
+      "loss": 0.6555,
+      "step": 40235
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.8525744985159916e-05,
+      "loss": 0.6533,
+      "step": 40240
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.852516117579554e-05,
+      "loss": 0.6716,
+      "step": 40245
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.8524577260061628e-05,
+      "loss": 0.6648,
+      "step": 40250
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.8523993237965458e-05,
+      "loss": 0.6527,
+      "step": 40255
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.8523409109514324e-05,
+      "loss": 0.6575,
+      "step": 40260
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.852282487471551e-05,
+      "loss": 0.6606,
+      "step": 40265
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.8522240533576302e-05,
+      "loss": 0.6574,
+      "step": 40270
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.8521656086103998e-05,
+      "loss": 0.666,
+      "step": 40275
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.852107153230589e-05,
+      "loss": 0.6602,
+      "step": 40280
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.852048687218927e-05,
+      "loss": 0.6456,
+      "step": 40285
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.8519902105761426e-05,
+      "loss": 0.6615,
+      "step": 40290
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.8519317233029664e-05,
+      "loss": 0.6568,
+      "step": 40295
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.8518732254001275e-05,
+      "loss": 0.6625,
+      "step": 40300
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.8518147168683566e-05,
+      "loss": 0.6632,
+      "step": 40305
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.8517561977083825e-05,
+      "loss": 0.6501,
+      "step": 40310
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.851697667920936e-05,
+      "loss": 0.6543,
+      "step": 40315
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.851639127506748e-05,
+      "loss": 0.6629,
+      "step": 40320
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.8515805764665478e-05,
+      "loss": 0.6532,
+      "step": 40325
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.851522014801067e-05,
+      "loss": 0.6588,
+      "step": 40330
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.8514634425110353e-05,
+      "loss": 0.6556,
+      "step": 40335
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.851404859597184e-05,
+      "loss": 0.654,
+      "step": 40340
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.851346266060244e-05,
+      "loss": 0.6637,
+      "step": 40345
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.8512876619009467e-05,
+      "loss": 0.6516,
+      "step": 40350
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.8512290471200225e-05,
+      "loss": 0.6708,
+      "step": 40355
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.8511704217182038e-05,
+      "loss": 0.6613,
+      "step": 40360
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.8511117856962213e-05,
+      "loss": 0.6589,
+      "step": 40365
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.8510531390548067e-05,
+      "loss": 0.6457,
+      "step": 40370
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.850994481794692e-05,
+      "loss": 0.6593,
+      "step": 40375
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.8509358139166088e-05,
+      "loss": 0.6471,
+      "step": 40380
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.8508771354212894e-05,
+      "loss": 0.6585,
+      "step": 40385
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.8508184463094657e-05,
+      "loss": 0.6526,
+      "step": 40390
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.8507597465818702e-05,
+      "loss": 0.6568,
+      "step": 40395
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.8507010362392352e-05,
+      "loss": 0.651,
+      "step": 40400
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.850642315282293e-05,
+      "loss": 0.6581,
+      "step": 40405
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.8505835837117763e-05,
+      "loss": 0.6514,
+      "step": 40410
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.8505248415284186e-05,
+      "loss": 0.6576,
+      "step": 40415
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.850466088732952e-05,
+      "loss": 0.6465,
+      "step": 40420
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.85040732532611e-05,
+      "loss": 0.6597,
+      "step": 40425
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.8503485513086256e-05,
+      "loss": 0.6526,
+      "step": 40430
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.850289766681232e-05,
+      "loss": 0.6524,
+      "step": 40435
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.850230971444663e-05,
+      "loss": 0.6662,
+      "step": 40440
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.850172165599652e-05,
+      "loss": 0.6554,
+      "step": 40445
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.850113349146933e-05,
+      "loss": 0.6532,
+      "step": 40450
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.8500545220872396e-05,
+      "loss": 0.6651,
+      "step": 40455
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.8499956844213058e-05,
+      "loss": 0.6443,
+      "step": 40460
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.849936836149866e-05,
+      "loss": 0.6558,
+      "step": 40465
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.849877977273654e-05,
+      "loss": 0.6435,
+      "step": 40470
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.8498191077934042e-05,
+      "loss": 0.647,
+      "step": 40475
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.8497602277098517e-05,
+      "loss": 0.6554,
+      "step": 40480
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.8497013370237303e-05,
+      "loss": 0.6625,
+      "step": 40485
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.849642435735776e-05,
+      "loss": 0.6514,
+      "step": 40490
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.8495835238467224e-05,
+      "loss": 0.6722,
+      "step": 40495
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.8495246013573057e-05,
+      "loss": 0.649,
+      "step": 40500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.84946566826826e-05,
+      "loss": 0.6621,
+      "step": 40505
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.8494067245803214e-05,
+      "loss": 0.6607,
+      "step": 40510
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.849347770294225e-05,
+      "loss": 0.6574,
+      "step": 40515
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.8492888054107066e-05,
+      "loss": 0.6526,
+      "step": 40520
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.8492298299305015e-05,
+      "loss": 0.6616,
+      "step": 40525
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.8491708438543465e-05,
+      "loss": 0.6552,
+      "step": 40530
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.849111847182976e-05,
+      "loss": 0.6595,
+      "step": 40535
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.8490528399171278e-05,
+      "loss": 0.6664,
+      "step": 40540
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.848993822057537e-05,
+      "loss": 0.6513,
+      "step": 40545
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.8489347936049403e-05,
+      "loss": 0.6507,
+      "step": 40550
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.8488757545600746e-05,
+      "loss": 0.6621,
+      "step": 40555
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.8488167049236758e-05,
+      "loss": 0.6473,
+      "step": 40560
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.8487576446964812e-05,
+      "loss": 0.6586,
+      "step": 40565
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.8486985738792277e-05,
+      "loss": 0.6622,
+      "step": 40570
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.848639492472652e-05,
+      "loss": 0.6603,
+      "step": 40575
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.8485804004774914e-05,
+      "loss": 0.6512,
+      "step": 40580
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.8485212978944834e-05,
+      "loss": 0.6552,
+      "step": 40585
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.848462184724365e-05,
+      "loss": 0.6604,
+      "step": 40590
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.8484030609678744e-05,
+      "loss": 0.6612,
+      "step": 40595
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.8483439266257485e-05,
+      "loss": 0.6639,
+      "step": 40600
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.848284781698726e-05,
+      "loss": 0.6508,
+      "step": 40605
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.8482256261875442e-05,
+      "loss": 0.6534,
+      "step": 40610
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.8481664600929413e-05,
+      "loss": 0.6471,
+      "step": 40615
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.848107283415656e-05,
+      "loss": 0.6593,
+      "step": 40620
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.848048096156426e-05,
+      "loss": 0.6575,
+      "step": 40625
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.8479888983159903e-05,
+      "loss": 0.6488,
+      "step": 40630
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.8479296898950873e-05,
+      "loss": 0.6602,
+      "step": 40635
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.8478704708944555e-05,
+      "loss": 0.6599,
+      "step": 40640
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.847811241314834e-05,
+      "loss": 0.6514,
+      "step": 40645
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.847752001156962e-05,
+      "loss": 0.6621,
+      "step": 40650
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.8476927504215786e-05,
+      "loss": 0.6587,
+      "step": 40655
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.847633489109423e-05,
+      "loss": 0.6541,
+      "step": 40660
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.8475742172212347e-05,
+      "loss": 0.6461,
+      "step": 40665
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.8475149347577526e-05,
+      "loss": 0.6511,
+      "step": 40670
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.8474556417197175e-05,
+      "loss": 0.6624,
+      "step": 40675
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.8473963381078685e-05,
+      "loss": 0.6658,
+      "step": 40680
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.847337023922945e-05,
+      "loss": 0.6599,
+      "step": 40685
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.8472776991656886e-05,
+      "loss": 0.6706,
+      "step": 40690
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.847218363836838e-05,
+      "loss": 0.645,
+      "step": 40695
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.8471590179371346e-05,
+      "loss": 0.6599,
+      "step": 40700
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.847099661467318e-05,
+      "loss": 0.6652,
+      "step": 40705
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.8470402944281295e-05,
+      "loss": 0.6611,
+      "step": 40710
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.8469809168203093e-05,
+      "loss": 0.6425,
+      "step": 40715
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.846921528644599e-05,
+      "loss": 0.6419,
+      "step": 40720
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.8468621299017388e-05,
+      "loss": 0.6413,
+      "step": 40725
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.84680272059247e-05,
+      "loss": 0.6665,
+      "step": 40730
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.846743300717534e-05,
+      "loss": 0.6614,
+      "step": 40735
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.8466838702776722e-05,
+      "loss": 0.6551,
+      "step": 40740
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.8466244292736267e-05,
+      "loss": 0.6487,
+      "step": 40745
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.8465649777061377e-05,
+      "loss": 0.6511,
+      "step": 40750
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.8465055155759483e-05,
+      "loss": 0.6505,
+      "step": 40755
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.8464460428838002e-05,
+      "loss": 0.6628,
+      "step": 40760
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.846386559630435e-05,
+      "loss": 0.6443,
+      "step": 40765
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.846327065816595e-05,
+      "loss": 0.6618,
+      "step": 40770
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.8462675614430223e-05,
+      "loss": 0.647,
+      "step": 40775
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.8462080465104598e-05,
+      "loss": 0.6585,
+      "step": 40780
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8461485210196503e-05,
+      "loss": 0.6568,
+      "step": 40785
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8460889849713357e-05,
+      "loss": 0.6566,
+      "step": 40790
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8460294383662595e-05,
+      "loss": 0.6625,
+      "step": 40795
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8459698812051642e-05,
+      "loss": 0.657,
+      "step": 40800
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8459103134887936e-05,
+      "loss": 0.6601,
+      "step": 40805
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.84585073521789e-05,
+      "loss": 0.6654,
+      "step": 40810
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8457911463931975e-05,
+      "loss": 0.6613,
+      "step": 40815
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.845731547015459e-05,
+      "loss": 0.6502,
+      "step": 40820
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.845671937085419e-05,
+      "loss": 0.6621,
+      "step": 40825
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.84561231660382e-05,
+      "loss": 0.648,
+      "step": 40830
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8455526855714074e-05,
+      "loss": 0.6479,
+      "step": 40835
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.845493043988924e-05,
+      "loss": 0.664,
+      "step": 40840
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8454333918571145e-05,
+      "loss": 0.6421,
+      "step": 40845
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.845373729176723e-05,
+      "loss": 0.664,
+      "step": 40850
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.845314055948494e-05,
+      "loss": 0.6531,
+      "step": 40855
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8452543721731723e-05,
+      "loss": 0.6533,
+      "step": 40860
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8451946778515024e-05,
+      "loss": 0.6421,
+      "step": 40865
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8451349729842288e-05,
+      "loss": 0.6561,
+      "step": 40870
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8450752575720967e-05,
+      "loss": 0.6491,
+      "step": 40875
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8450155316158512e-05,
+      "loss": 0.6523,
+      "step": 40880
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8449557951162376e-05,
+      "loss": 0.674,
+      "step": 40885
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8448960480740007e-05,
+      "loss": 0.6426,
+      "step": 40890
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8448362904898868e-05,
+      "loss": 0.6704,
+      "step": 40895
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.844776522364641e-05,
+      "loss": 0.648,
+      "step": 40900
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8447167436990093e-05,
+      "loss": 0.6602,
+      "step": 40905
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8446569544937372e-05,
+      "loss": 0.6607,
+      "step": 40910
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8445971547495707e-05,
+      "loss": 0.6582,
+      "step": 40915
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8445373444672562e-05,
+      "loss": 0.6498,
+      "step": 40920
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8444775236475402e-05,
+      "loss": 0.648,
+      "step": 40925
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8444176922911686e-05,
+      "loss": 0.6457,
+      "step": 40930
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.844357850398888e-05,
+      "loss": 0.6567,
+      "step": 40935
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8442979979714453e-05,
+      "loss": 0.6589,
+      "step": 40940
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.844238135009587e-05,
+      "loss": 0.6581,
+      "step": 40945
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8441782615140603e-05,
+      "loss": 0.6613,
+      "step": 40950
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.844118377485612e-05,
+      "loss": 0.6492,
+      "step": 40955
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8440584829249894e-05,
+      "loss": 0.6513,
+      "step": 40960
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.84399857783294e-05,
+      "loss": 0.6457,
+      "step": 40965
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8439386622102108e-05,
+      "loss": 0.6629,
+      "step": 40970
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.84387873605755e-05,
+      "loss": 0.6363,
+      "step": 40975
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8438187993757045e-05,
+      "loss": 0.6463,
+      "step": 40980
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8437588521654228e-05,
+      "loss": 0.6576,
+      "step": 40985
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8436988944274522e-05,
+      "loss": 0.6512,
+      "step": 40990
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.843638926162542e-05,
+      "loss": 0.6415,
+      "step": 40995
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.843578947371439e-05,
+      "loss": 0.6552,
+      "step": 41000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8435189580548924e-05,
+      "loss": 0.6565,
+      "step": 41005
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8434589582136503e-05,
+      "loss": 0.6655,
+      "step": 41010
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8433989478484622e-05,
+      "loss": 0.6557,
+      "step": 41015
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8433389269600757e-05,
+      "loss": 0.6554,
+      "step": 41020
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8432788955492404e-05,
+      "loss": 0.6652,
+      "step": 41025
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.843218853616705e-05,
+      "loss": 0.6621,
+      "step": 41030
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.843158801163219e-05,
+      "loss": 0.6552,
+      "step": 41035
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8430987381895312e-05,
+      "loss": 0.6588,
+      "step": 41040
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8430386646963915e-05,
+      "loss": 0.6586,
+      "step": 41045
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.842978580684549e-05,
+      "loss": 0.6344,
+      "step": 41050
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8429184861547537e-05,
+      "loss": 0.6669,
+      "step": 41055
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.842858381107755e-05,
+      "loss": 0.6621,
+      "step": 41060
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8427982655443034e-05,
+      "loss": 0.6603,
+      "step": 41065
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8427381394651485e-05,
+      "loss": 0.6602,
+      "step": 41070
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.842678002871041e-05,
+      "loss": 0.6482,
+      "step": 41075
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8426178557627305e-05,
+      "loss": 0.6649,
+      "step": 41080
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8425576981409682e-05,
+      "loss": 0.6539,
+      "step": 41085
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8424975300065046e-05,
+      "loss": 0.6479,
+      "step": 41090
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8424373513600896e-05,
+      "loss": 0.6557,
+      "step": 41095
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8423771622024752e-05,
+      "loss": 0.6632,
+      "step": 41100
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.842316962534412e-05,
+      "loss": 0.6628,
+      "step": 41105
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8422567523566503e-05,
+      "loss": 0.6509,
+      "step": 41110
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8421965316699424e-05,
+      "loss": 0.6444,
+      "step": 41115
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8421363004750393e-05,
+      "loss": 0.6552,
+      "step": 41120
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8420760587726925e-05,
+      "loss": 0.6522,
+      "step": 41125
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8420158065636538e-05,
+      "loss": 0.6439,
+      "step": 41130
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8419555438486747e-05,
+      "loss": 0.656,
+      "step": 41135
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.841895270628507e-05,
+      "loss": 0.6583,
+      "step": 41140
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8418349869039033e-05,
+      "loss": 0.6557,
+      "step": 41145
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8417746926756157e-05,
+      "loss": 0.656,
+      "step": 41150
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8417143879443956e-05,
+      "loss": 0.6449,
+      "step": 41155
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8416540727109967e-05,
+      "loss": 0.6564,
+      "step": 41160
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8415937469761703e-05,
+      "loss": 0.6538,
+      "step": 41165
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.84153341074067e-05,
+      "loss": 0.6515,
+      "step": 41170
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.841473064005249e-05,
+      "loss": 0.6682,
+      "step": 41175
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.841412706770659e-05,
+      "loss": 0.6564,
+      "step": 41180
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8413523390376535e-05,
+      "loss": 0.6656,
+      "step": 41185
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8412919608069862e-05,
+      "loss": 0.6432,
+      "step": 41190
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8412315720794102e-05,
+      "loss": 0.6496,
+      "step": 41195
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8411711728556787e-05,
+      "loss": 0.651,
+      "step": 41200
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8411107631365457e-05,
+      "loss": 0.6563,
+      "step": 41205
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.841050342922765e-05,
+      "loss": 0.6444,
+      "step": 41210
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8409899122150898e-05,
+      "loss": 0.657,
+      "step": 41215
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.840929471014275e-05,
+      "loss": 0.6521,
+      "step": 41220
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8408690193210737e-05,
+      "loss": 0.6598,
+      "step": 41225
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8408085571362412e-05,
+      "loss": 0.6604,
+      "step": 41230
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8407480844605315e-05,
+      "loss": 0.6487,
+      "step": 41235
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.840687601294699e-05,
+      "loss": 0.6506,
+      "step": 41240
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8406271076394983e-05,
+      "loss": 0.6451,
+      "step": 41245
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8405666034956842e-05,
+      "loss": 0.649,
+      "step": 41250
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8405060888640122e-05,
+      "loss": 0.6531,
+      "step": 41255
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8404455637452364e-05,
+      "loss": 0.6465,
+      "step": 41260
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8403850281401128e-05,
+      "loss": 0.6484,
+      "step": 41265
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8403244820493966e-05,
+      "loss": 0.652,
+      "step": 41270
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8402639254738422e-05,
+      "loss": 0.6562,
+      "step": 41275
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8402033584142067e-05,
+      "loss": 0.6464,
+      "step": 41280
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.840142780871245e-05,
+      "loss": 0.6663,
+      "step": 41285
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8400821928457126e-05,
+      "loss": 0.6537,
+      "step": 41290
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.840021594338366e-05,
+      "loss": 0.6452,
+      "step": 41295
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8399609853499614e-05,
+      "loss": 0.6464,
+      "step": 41300
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.839900365881255e-05,
+      "loss": 0.6634,
+      "step": 41305
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8398397359330023e-05,
+      "loss": 0.6433,
+      "step": 41310
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.839779095505961e-05,
+      "loss": 0.6415,
+      "step": 41315
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8397184446008868e-05,
+      "loss": 0.6455,
+      "step": 41320
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.839657783218537e-05,
+      "loss": 0.6517,
+      "step": 41325
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.839597111359669e-05,
+      "loss": 0.652,
+      "step": 41330
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8395364290250383e-05,
+      "loss": 0.6569,
+      "step": 41335
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8394757362154027e-05,
+      "loss": 0.6645,
+      "step": 41340
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8394150329315203e-05,
+      "loss": 0.6592,
+      "step": 41345
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8393543191741473e-05,
+      "loss": 0.6524,
+      "step": 41350
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8392935949440417e-05,
+      "loss": 0.6573,
+      "step": 41355
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8392328602419617e-05,
+      "loss": 0.6524,
+      "step": 41360
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8391721150686642e-05,
+      "loss": 0.6551,
+      "step": 41365
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8391113594249076e-05,
+      "loss": 0.6441,
+      "step": 41370
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8390505933114503e-05,
+      "loss": 0.6678,
+      "step": 41375
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8389898167290497e-05,
+      "loss": 0.6493,
+      "step": 41380
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8389290296784647e-05,
+      "loss": 0.6516,
+      "step": 41385
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8388682321604535e-05,
+      "loss": 0.654,
+      "step": 41390
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8388074241757748e-05,
+      "loss": 0.6534,
+      "step": 41395
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8387466057251872e-05,
+      "loss": 0.66,
+      "step": 41400
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8386857768094492e-05,
+      "loss": 0.6464,
+      "step": 41405
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8386249374293208e-05,
+      "loss": 0.6516,
+      "step": 41410
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.83856408758556e-05,
+      "loss": 0.6551,
+      "step": 41415
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8385032272789266e-05,
+      "loss": 0.6533,
+      "step": 41420
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8384423565101798e-05,
+      "loss": 0.6634,
+      "step": 41425
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8383814752800793e-05,
+      "loss": 0.6619,
+      "step": 41430
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8383205835893845e-05,
+      "loss": 0.6618,
+      "step": 41435
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.838259681438855e-05,
+      "loss": 0.6566,
+      "step": 41440
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8381987688292514e-05,
+      "loss": 0.6644,
+      "step": 41445
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8381378457613327e-05,
+      "loss": 0.6562,
+      "step": 41450
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8380769122358595e-05,
+      "loss": 0.6492,
+      "step": 41455
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8380159682535925e-05,
+      "loss": 0.6439,
+      "step": 41460
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8379550138152916e-05,
+      "loss": 0.6521,
+      "step": 41465
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8378940489217178e-05,
+      "loss": 0.6608,
+      "step": 41470
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8378330735736313e-05,
+      "loss": 0.6488,
+      "step": 41475
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8377720877717925e-05,
+      "loss": 0.6548,
+      "step": 41480
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8377110915169634e-05,
+      "loss": 0.6732,
+      "step": 41485
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8376500848099043e-05,
+      "loss": 0.6417,
+      "step": 41490
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8375890676513767e-05,
+      "loss": 0.6435,
+      "step": 41495
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.837528040042142e-05,
+      "loss": 0.6642,
+      "step": 41500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8374670019829614e-05,
+      "loss": 0.6418,
+      "step": 41505
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8374059534745962e-05,
+      "loss": 0.6485,
+      "step": 41510
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8373448945178086e-05,
+      "loss": 0.6503,
+      "step": 41515
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8372838251133606e-05,
+      "loss": 0.6531,
+      "step": 41520
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8372227452620137e-05,
+      "loss": 0.6623,
+      "step": 41525
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8371616549645303e-05,
+      "loss": 0.6438,
+      "step": 41530
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8371005542216728e-05,
+      "loss": 0.646,
+      "step": 41535
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8370394430342027e-05,
+      "loss": 0.6467,
+      "step": 41540
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8369783214028832e-05,
+      "loss": 0.6584,
+      "step": 41545
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.836917189328477e-05,
+      "loss": 0.64,
+      "step": 41550
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8368560468117464e-05,
+      "loss": 0.6489,
+      "step": 41555
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8367948938534547e-05,
+      "loss": 0.6509,
+      "step": 41560
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.836733730454365e-05,
+      "loss": 0.6422,
+      "step": 41565
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8366725566152394e-05,
+      "loss": 0.6484,
+      "step": 41570
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.836611372336843e-05,
+      "loss": 0.6672,
+      "step": 41575
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8365501776199375e-05,
+      "loss": 0.6487,
+      "step": 41580
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8364889724652872e-05,
+      "loss": 0.6642,
+      "step": 41585
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8364277568736556e-05,
+      "loss": 0.641,
+      "step": 41590
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8363665308458065e-05,
+      "loss": 0.6585,
+      "step": 41595
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.836305294382504e-05,
+      "loss": 0.6576,
+      "step": 41600
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8362440474845124e-05,
+      "loss": 0.652,
+      "step": 41605
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.836182790152595e-05,
+      "loss": 0.6445,
+      "step": 41610
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.836121522387517e-05,
+      "loss": 0.6461,
+      "step": 41615
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8360602441900423e-05,
+      "loss": 0.6485,
+      "step": 41620
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8359989555609355e-05,
+      "loss": 0.647,
+      "step": 41625
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8359376565009618e-05,
+      "loss": 0.6461,
+      "step": 41630
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8358763470108852e-05,
+      "loss": 0.6458,
+      "step": 41635
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.835815027091472e-05,
+      "loss": 0.6526,
+      "step": 41640
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8357536967434856e-05,
+      "loss": 0.6625,
+      "step": 41645
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.835692355967693e-05,
+      "loss": 0.6536,
+      "step": 41650
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8356310047648578e-05,
+      "loss": 0.6454,
+      "step": 41655
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8355696431357465e-05,
+      "loss": 0.6613,
+      "step": 41660
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.835508271081125e-05,
+      "loss": 0.6497,
+      "step": 41665
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8354468886017582e-05,
+      "loss": 0.6522,
+      "step": 41670
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8353854956984127e-05,
+      "loss": 0.6565,
+      "step": 41675
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8353240923718538e-05,
+      "loss": 0.6458,
+      "step": 41680
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8352626786228486e-05,
+      "loss": 0.6555,
+      "step": 41685
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.835201254452162e-05,
+      "loss": 0.6595,
+      "step": 41690
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8351398198605617e-05,
+      "loss": 0.6568,
+      "step": 41695
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8350783748488135e-05,
+      "loss": 0.636,
+      "step": 41700
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8350169194176846e-05,
+      "loss": 0.6501,
+      "step": 41705
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8349554535679413e-05,
+      "loss": 0.6478,
+      "step": 41710
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8348939773003508e-05,
+      "loss": 0.6449,
+      "step": 41715
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8348324906156796e-05,
+      "loss": 0.6525,
+      "step": 41720
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8347709935146958e-05,
+      "loss": 0.6468,
+      "step": 41725
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8347094859981658e-05,
+      "loss": 0.6561,
+      "step": 41730
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8346479680668575e-05,
+      "loss": 0.6491,
+      "step": 41735
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8345864397215387e-05,
+      "loss": 0.6638,
+      "step": 41740
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8345249009629768e-05,
+      "loss": 0.6558,
+      "step": 41745
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.834463351791939e-05,
+      "loss": 0.659,
+      "step": 41750
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8344017922091944e-05,
+      "loss": 0.6615,
+      "step": 41755
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8343402222155106e-05,
+      "loss": 0.6521,
+      "step": 41760
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.834278641811656e-05,
+      "loss": 0.6446,
+      "step": 41765
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8342170509983985e-05,
+      "loss": 0.6497,
+      "step": 41770
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.834155449776507e-05,
+      "loss": 0.6634,
+      "step": 41775
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8340938381467498e-05,
+      "loss": 0.6468,
+      "step": 41780
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8340322161098956e-05,
+      "loss": 0.6534,
+      "step": 41785
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8339705836667135e-05,
+      "loss": 0.6597,
+      "step": 41790
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8339089408179724e-05,
+      "loss": 0.6503,
+      "step": 41795
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8338472875644417e-05,
+      "loss": 0.6496,
+      "step": 41800
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8337856239068902e-05,
+      "loss": 0.6517,
+      "step": 41805
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8337239498460876e-05,
+      "loss": 0.6538,
+      "step": 41810
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8336622653828032e-05,
+      "loss": 0.658,
+      "step": 41815
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8336005705178065e-05,
+      "loss": 0.6507,
+      "step": 41820
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.833538865251868e-05,
+      "loss": 0.6582,
+      "step": 41825
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.833477149585757e-05,
+      "loss": 0.6619,
+      "step": 41830
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8334154235202434e-05,
+      "loss": 0.6641,
+      "step": 41835
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8333536870560983e-05,
+      "loss": 0.6635,
+      "step": 41840
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.833291940194091e-05,
+      "loss": 0.6495,
+      "step": 41845
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8332301829349918e-05,
+      "loss": 0.6563,
+      "step": 41850
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.833168415279572e-05,
+      "loss": 0.661,
+      "step": 41855
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.833106637228602e-05,
+      "loss": 0.6416,
+      "step": 41860
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.833044848782853e-05,
+      "loss": 0.6485,
+      "step": 41865
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.832983049943095e-05,
+      "loss": 0.6486,
+      "step": 41870
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8329212407100996e-05,
+      "loss": 0.6532,
+      "step": 41875
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8328594210846382e-05,
+      "loss": 0.6419,
+      "step": 41880
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.832797591067482e-05,
+      "loss": 0.6535,
+      "step": 41885
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8327357506594023e-05,
+      "loss": 0.6538,
+      "step": 41890
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8326738998611708e-05,
+      "loss": 0.6564,
+      "step": 41895
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8326120386735595e-05,
+      "loss": 0.6372,
+      "step": 41900
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8325501670973397e-05,
+      "loss": 0.6529,
+      "step": 41905
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8324882851332834e-05,
+      "loss": 0.6555,
+      "step": 41910
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8324263927821634e-05,
+      "loss": 0.6669,
+      "step": 41915
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.832364490044751e-05,
+      "loss": 0.6521,
+      "step": 41920
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8323025769218198e-05,
+      "loss": 0.6457,
+      "step": 41925
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.832240653414141e-05,
+      "loss": 0.6625,
+      "step": 41930
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.832178719522488e-05,
+      "loss": 0.6623,
+      "step": 41935
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8321167752476335e-05,
+      "loss": 0.6697,
+      "step": 41940
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8320548205903497e-05,
+      "loss": 0.6341,
+      "step": 41945
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8319928555514108e-05,
+      "loss": 0.6483,
+      "step": 41950
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.831930880131589e-05,
+      "loss": 0.643,
+      "step": 41955
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8318688943316577e-05,
+      "loss": 0.6533,
+      "step": 41960
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8318068981523906e-05,
+      "loss": 0.6665,
+      "step": 41965
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8317448915945613e-05,
+      "loss": 0.6486,
+      "step": 41970
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.831682874658943e-05,
+      "loss": 0.6562,
+      "step": 41975
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.83162084734631e-05,
+      "loss": 0.6488,
+      "step": 41980
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8315588096574358e-05,
+      "loss": 0.6543,
+      "step": 41985
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8314967615930948e-05,
+      "loss": 0.6575,
+      "step": 41990
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.831434703154061e-05,
+      "loss": 0.6569,
+      "step": 41995
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8313726343411085e-05,
+      "loss": 0.6301,
+      "step": 42000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8313105551550125e-05,
+      "loss": 0.6437,
+      "step": 42005
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8312484655965466e-05,
+      "loss": 0.6487,
+      "step": 42010
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.831186365666486e-05,
+      "loss": 0.6587,
+      "step": 42015
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8311242553656056e-05,
+      "loss": 0.6595,
+      "step": 42020
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.83106213469468e-05,
+      "loss": 0.6386,
+      "step": 42025
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.831000003654485e-05,
+      "loss": 0.6462,
+      "step": 42030
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.830937862245795e-05,
+      "loss": 0.6319,
+      "step": 42035
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8308757104693854e-05,
+      "loss": 0.6573,
+      "step": 42040
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8308135483260324e-05,
+      "loss": 0.6556,
+      "step": 42045
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8307513758165113e-05,
+      "loss": 0.6578,
+      "step": 42050
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8306891929415973e-05,
+      "loss": 0.6485,
+      "step": 42055
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8306269997020666e-05,
+      "loss": 0.659,
+      "step": 42060
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8305647960986958e-05,
+      "loss": 0.654,
+      "step": 42065
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.83050258213226e-05,
+      "loss": 0.6445,
+      "step": 42070
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.830440357803536e-05,
+      "loss": 0.6471,
+      "step": 42075
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8303781231133e-05,
+      "loss": 0.6465,
+      "step": 42080
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.830315878062329e-05,
+      "loss": 0.6667,
+      "step": 42085
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8302536226513993e-05,
+      "loss": 0.6563,
+      "step": 42090
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8301913568812874e-05,
+      "loss": 0.6507,
+      "step": 42095
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8301290807527703e-05,
+      "loss": 0.6493,
+      "step": 42100
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8300667942666252e-05,
+      "loss": 0.6493,
+      "step": 42105
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8300044974236294e-05,
+      "loss": 0.6444,
+      "step": 42110
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8299421902245598e-05,
+      "loss": 0.6493,
+      "step": 42115
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8298798726701942e-05,
+      "loss": 0.6529,
+      "step": 42120
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.82981754476131e-05,
+      "loss": 0.6469,
+      "step": 42125
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8297552064986846e-05,
+      "loss": 0.643,
+      "step": 42130
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8296928578830963e-05,
+      "loss": 0.6493,
+      "step": 42135
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8296304989153225e-05,
+      "loss": 0.6562,
+      "step": 42140
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8295681295961416e-05,
+      "loss": 0.6448,
+      "step": 42145
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.829505749926332e-05,
+      "loss": 0.6478,
+      "step": 42150
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.829443359906671e-05,
+      "loss": 0.6484,
+      "step": 42155
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8293809595379386e-05,
+      "loss": 0.6403,
+      "step": 42160
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8293185488209122e-05,
+      "loss": 0.6458,
+      "step": 42165
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.829256127756371e-05,
+      "loss": 0.644,
+      "step": 42170
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8291936963450933e-05,
+      "loss": 0.6468,
+      "step": 42175
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8291312545878587e-05,
+      "loss": 0.6404,
+      "step": 42180
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8290688024854465e-05,
+      "loss": 0.6493,
+      "step": 42185
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.829006340038635e-05,
+      "loss": 0.6433,
+      "step": 42190
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8289438672482044e-05,
+      "loss": 0.6607,
+      "step": 42195
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8288813841149333e-05,
+      "loss": 0.6561,
+      "step": 42200
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8288188906396024e-05,
+      "loss": 0.6421,
+      "step": 42205
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8287563868229906e-05,
+      "loss": 0.6556,
+      "step": 42210
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8286938726658778e-05,
+      "loss": 0.6419,
+      "step": 42215
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8286313481690446e-05,
+      "loss": 0.6558,
+      "step": 42220
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8285688133332704e-05,
+      "loss": 0.6435,
+      "step": 42225
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.828506268159336e-05,
+      "loss": 0.6559,
+      "step": 42230
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8284437126480215e-05,
+      "loss": 0.6551,
+      "step": 42235
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8283811468001078e-05,
+      "loss": 0.6505,
+      "step": 42240
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8283185706163752e-05,
+      "loss": 0.668,
+      "step": 42245
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8282559840976043e-05,
+      "loss": 0.643,
+      "step": 42250
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8281933872445764e-05,
+      "loss": 0.6536,
+      "step": 42255
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.828130780058072e-05,
+      "loss": 0.6427,
+      "step": 42260
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8280681625388732e-05,
+      "loss": 0.6526,
+      "step": 42265
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.82800553468776e-05,
+      "loss": 0.6462,
+      "step": 42270
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8279428965055153e-05,
+      "loss": 0.6581,
+      "step": 42275
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8278802479929197e-05,
+      "loss": 0.6309,
+      "step": 42280
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8278175891507546e-05,
+      "loss": 0.6608,
+      "step": 42285
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8277549199798023e-05,
+      "loss": 0.6461,
+      "step": 42290
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.827692240480845e-05,
+      "loss": 0.6502,
+      "step": 42295
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.827629550654664e-05,
+      "loss": 0.6659,
+      "step": 42300
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8275668505020424e-05,
+      "loss": 0.655,
+      "step": 42305
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.827504140023762e-05,
+      "loss": 0.652,
+      "step": 42310
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.827441419220605e-05,
+      "loss": 0.6436,
+      "step": 42315
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8273786880933543e-05,
+      "loss": 0.6497,
+      "step": 42320
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8273159466427926e-05,
+      "loss": 0.6586,
+      "step": 42325
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8272531948697026e-05,
+      "loss": 0.6614,
+      "step": 42330
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8271904327748677e-05,
+      "loss": 0.6466,
+      "step": 42335
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.82712766035907e-05,
+      "loss": 0.6389,
+      "step": 42340
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.827064877623094e-05,
+      "loss": 0.65,
+      "step": 42345
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.827002084567722e-05,
+      "loss": 0.6602,
+      "step": 42350
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8269392811937384e-05,
+      "loss": 0.6594,
+      "step": 42355
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8268764675019257e-05,
+      "loss": 0.6466,
+      "step": 42360
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8268136434930687e-05,
+      "loss": 0.6508,
+      "step": 42365
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8267508091679508e-05,
+      "loss": 0.6595,
+      "step": 42370
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8266879645273557e-05,
+      "loss": 0.6476,
+      "step": 42375
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8266251095720677e-05,
+      "loss": 0.6453,
+      "step": 42380
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8265622443028715e-05,
+      "loss": 0.6508,
+      "step": 42385
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.826499368720551e-05,
+      "loss": 0.6504,
+      "step": 42390
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8264364828258908e-05,
+      "loss": 0.6516,
+      "step": 42395
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8263735866196758e-05,
+      "loss": 0.6452,
+      "step": 42400
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8263106801026903e-05,
+      "loss": 0.6451,
+      "step": 42405
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8262477632757193e-05,
+      "loss": 0.6529,
+      "step": 42410
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8261848361395484e-05,
+      "loss": 0.657,
+      "step": 42415
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8261218986949617e-05,
+      "loss": 0.6501,
+      "step": 42420
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8260589509427457e-05,
+      "loss": 0.6667,
+      "step": 42425
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.825995992883685e-05,
+      "loss": 0.6412,
+      "step": 42430
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8259330245185652e-05,
+      "loss": 0.6534,
+      "step": 42435
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.825870045848172e-05,
+      "loss": 0.6578,
+      "step": 42440
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.825807056873292e-05,
+      "loss": 0.6573,
+      "step": 42445
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8257440575947095e-05,
+      "loss": 0.6463,
+      "step": 42450
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.825681048013212e-05,
+      "loss": 0.6464,
+      "step": 42455
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.825618028129585e-05,
+      "loss": 0.6501,
+      "step": 42460
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8255549979446148e-05,
+      "loss": 0.6443,
+      "step": 42465
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.825491957459088e-05,
+      "loss": 0.6457,
+      "step": 42470
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.825428906673791e-05,
+      "loss": 0.6429,
+      "step": 42475
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.825365845589511e-05,
+      "loss": 0.6416,
+      "step": 42480
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8253027742070342e-05,
+      "loss": 0.6516,
+      "step": 42485
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.825239692527148e-05,
+      "loss": 0.6525,
+      "step": 42490
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8251766005506396e-05,
+      "loss": 0.6356,
+      "step": 42495
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8251134982782952e-05,
+      "loss": 0.6572,
+      "step": 42500
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8250503857109033e-05,
+      "loss": 0.6588,
+      "step": 42505
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8249872628492507e-05,
+      "loss": 0.6519,
+      "step": 42510
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.824924129694125e-05,
+      "loss": 0.655,
+      "step": 42515
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8248609862463145e-05,
+      "loss": 0.6601,
+      "step": 42520
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8247978325066063e-05,
+      "loss": 0.6482,
+      "step": 42525
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.824734668475789e-05,
+      "loss": 0.6546,
+      "step": 42530
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8246714941546507e-05,
+      "loss": 0.6464,
+      "step": 42535
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8246083095439788e-05,
+      "loss": 0.6492,
+      "step": 42540
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8245451146445626e-05,
+      "loss": 0.6523,
+      "step": 42545
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8244819094571897e-05,
+      "loss": 0.6538,
+      "step": 42550
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8244186939826497e-05,
+      "loss": 0.6542,
+      "step": 42555
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.824355468221731e-05,
+      "loss": 0.6461,
+      "step": 42560
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8242922321752222e-05,
+      "loss": 0.6615,
+      "step": 42565
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8242289858439123e-05,
+      "loss": 0.6458,
+      "step": 42570
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8241657292285907e-05,
+      "loss": 0.652,
+      "step": 42575
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8241024623300466e-05,
+      "loss": 0.6486,
+      "step": 42580
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8240391851490696e-05,
+      "loss": 0.6543,
+      "step": 42585
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8239758976864487e-05,
+      "loss": 0.6447,
+      "step": 42590
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8239125999429737e-05,
+      "loss": 0.6581,
+      "step": 42595
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8238492919194347e-05,
+      "loss": 0.6407,
+      "step": 42600
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8237859736166216e-05,
+      "loss": 0.6482,
+      "step": 42605
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.823722645035324e-05,
+      "loss": 0.6371,
+      "step": 42610
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.823659306176332e-05,
+      "loss": 0.6416,
+      "step": 42615
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8235959570404364e-05,
+      "loss": 0.6502,
+      "step": 42620
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8235325976284276e-05,
+      "loss": 0.6403,
+      "step": 42625
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8234692279410956e-05,
+      "loss": 0.6372,
+      "step": 42630
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8234058479792317e-05,
+      "loss": 0.6585,
+      "step": 42635
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.823342457743626e-05,
+      "loss": 0.6574,
+      "step": 42640
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8232790572350705e-05,
+      "loss": 0.6709,
+      "step": 42645
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.823215646454355e-05,
+      "loss": 0.6515,
+      "step": 42650
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8231522254022714e-05,
+      "loss": 0.6533,
+      "step": 42655
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.823088794079611e-05,
+      "loss": 0.6567,
+      "step": 42660
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8230253524871652e-05,
+      "loss": 0.6502,
+      "step": 42665
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8229619006257254e-05,
+      "loss": 0.6429,
+      "step": 42670
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8228984384960832e-05,
+      "loss": 0.6546,
+      "step": 42675
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.822834966099031e-05,
+      "loss": 0.6456,
+      "step": 42680
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.82277148343536e-05,
+      "loss": 0.648,
+      "step": 42685
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8227079905058627e-05,
+      "loss": 0.6536,
+      "step": 42690
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8226444873113315e-05,
+      "loss": 0.6588,
+      "step": 42695
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8225809738525583e-05,
+      "loss": 0.6393,
+      "step": 42700
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8225174501303357e-05,
+      "loss": 0.6435,
+      "step": 42705
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8224539161454564e-05,
+      "loss": 0.6535,
+      "step": 42710
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8223903718987134e-05,
+      "loss": 0.6462,
+      "step": 42715
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8223268173908988e-05,
+      "loss": 0.6606,
+      "step": 42720
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.822263252622806e-05,
+      "loss": 0.6565,
+      "step": 42725
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8221996775952285e-05,
+      "loss": 0.6433,
+      "step": 42730
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.822136092308959e-05,
+      "loss": 0.6534,
+      "step": 42735
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8220724967647907e-05,
+      "loss": 0.6512,
+      "step": 42740
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8220088909635176e-05,
+      "loss": 0.6446,
+      "step": 42745
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8219452749059332e-05,
+      "loss": 0.6479,
+      "step": 42750
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.821881648592831e-05,
+      "loss": 0.6507,
+      "step": 42755
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8218180120250055e-05,
+      "loss": 0.6537,
+      "step": 42760
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.82175436520325e-05,
+      "loss": 0.6501,
+      "step": 42765
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8216907081283586e-05,
+      "loss": 0.6558,
+      "step": 42770
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.821627040801126e-05,
+      "loss": 0.6469,
+      "step": 42775
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8215633632223463e-05,
+      "loss": 0.6478,
+      "step": 42780
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8214996753928143e-05,
+      "loss": 0.6454,
+      "step": 42785
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8214359773133243e-05,
+      "loss": 0.6482,
+      "step": 42790
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8213722689846714e-05,
+      "loss": 0.6512,
+      "step": 42795
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.82130855040765e-05,
+      "loss": 0.6464,
+      "step": 42800
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8212448215830558e-05,
+      "loss": 0.6563,
+      "step": 42805
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8211810825116835e-05,
+      "loss": 0.6604,
+      "step": 42810
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8211173331943287e-05,
+      "loss": 0.6454,
+      "step": 42815
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8210535736317864e-05,
+      "loss": 0.6492,
+      "step": 42820
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8209898038248523e-05,
+      "loss": 0.6471,
+      "step": 42825
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.820926023774322e-05,
+      "loss": 0.6423,
+      "step": 42830
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8208622334809915e-05,
+      "loss": 0.6492,
+      "step": 42835
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8207984329456567e-05,
+      "loss": 0.6541,
+      "step": 42840
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8207346221691135e-05,
+      "loss": 0.656,
+      "step": 42845
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.820670801152158e-05,
+      "loss": 0.6427,
+      "step": 42850
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8206069698955867e-05,
+      "loss": 0.6488,
+      "step": 42855
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8205431284001963e-05,
+      "loss": 0.6444,
+      "step": 42860
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8204792766667827e-05,
+      "loss": 0.6404,
+      "step": 42865
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8204154146961428e-05,
+      "loss": 0.645,
+      "step": 42870
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8203515424890738e-05,
+      "loss": 0.6437,
+      "step": 42875
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8202876600463722e-05,
+      "loss": 0.6467,
+      "step": 42880
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8202237673688354e-05,
+      "loss": 0.6395,
+      "step": 42885
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8201598644572605e-05,
+      "loss": 0.6433,
+      "step": 42890
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8200959513124443e-05,
+      "loss": 0.6569,
+      "step": 42895
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.820032027935185e-05,
+      "loss": 0.6472,
+      "step": 42900
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8199680943262802e-05,
+      "loss": 0.643,
+      "step": 42905
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8199041504865266e-05,
+      "loss": 0.6487,
+      "step": 42910
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.819840196416723e-05,
+      "loss": 0.6457,
+      "step": 42915
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8197762321176675e-05,
+      "loss": 0.6383,
+      "step": 42920
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8197122575901576e-05,
+      "loss": 0.6505,
+      "step": 42925
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8196482728349915e-05,
+      "loss": 0.6445,
+      "step": 42930
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8195842778529677e-05,
+      "loss": 0.6656,
+      "step": 42935
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.819520272644885e-05,
+      "loss": 0.6391,
+      "step": 42940
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8194562572115417e-05,
+      "loss": 0.6517,
+      "step": 42945
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8193922315537363e-05,
+      "loss": 0.6463,
+      "step": 42950
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8193281956722683e-05,
+      "loss": 0.6423,
+      "step": 42955
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8192641495679362e-05,
+      "loss": 0.6584,
+      "step": 42960
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.819200093241539e-05,
+      "loss": 0.6595,
+      "step": 42965
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.8191360266938765e-05,
+      "loss": 0.6437,
+      "step": 42970
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.8190719499257472e-05,
+      "loss": 0.6552,
+      "step": 42975
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.8190078629379514e-05,
+      "loss": 0.6598,
+      "step": 42980
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.8189437657312885e-05,
+      "loss": 0.6431,
+      "step": 42985
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.8188796583065582e-05,
+      "loss": 0.6579,
+      "step": 42990
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.8188155406645603e-05,
+      "loss": 0.6521,
+      "step": 42995
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.8187514128060946e-05,
+      "loss": 0.6576,
+      "step": 43000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.818687274731962e-05,
+      "loss": 0.6567,
+      "step": 43005
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.8186231264429615e-05,
+      "loss": 0.6557,
+      "step": 43010
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.8185589679398948e-05,
+      "loss": 0.659,
+      "step": 43015
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.818494799223562e-05,
+      "loss": 0.6506,
+      "step": 43020
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.8184306202947635e-05,
+      "loss": 0.6515,
+      "step": 43025
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.8183664311542995e-05,
+      "loss": 0.6451,
+      "step": 43030
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.818302231802972e-05,
+      "loss": 0.6421,
+      "step": 43035
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.818238022241582e-05,
+      "loss": 0.6397,
+      "step": 43040
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.81817380247093e-05,
+      "loss": 0.651,
+      "step": 43045
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.8181095724918173e-05,
+      "loss": 0.6611,
+      "step": 43050
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.8180453323050456e-05,
+      "loss": 0.6377,
+      "step": 43055
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.8179810819114164e-05,
+      "loss": 0.6372,
+      "step": 43060
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.8179168213117314e-05,
+      "loss": 0.6574,
+      "step": 43065
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.8178525505067922e-05,
+      "loss": 0.6537,
+      "step": 43070
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.8177882694974008e-05,
+      "loss": 0.6357,
+      "step": 43075
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.817723978284359e-05,
+      "loss": 0.6529,
+      "step": 43080
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.8176596768684697e-05,
+      "loss": 0.6451,
+      "step": 43085
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.8175953652505344e-05,
+      "loss": 0.6402,
+      "step": 43090
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.8175310434313558e-05,
+      "loss": 0.6535,
+      "step": 43095
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.8174667114117368e-05,
+      "loss": 0.6661,
+      "step": 43100
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.8174023691924794e-05,
+      "loss": 0.6456,
+      "step": 43105
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.8173380167743874e-05,
+      "loss": 0.6496,
+      "step": 43110
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.8172736541582626e-05,
+      "loss": 0.6468,
+      "step": 43115
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.8172092813449083e-05,
+      "loss": 0.6541,
+      "step": 43120
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.8171448983351284e-05,
+      "loss": 0.65,
+      "step": 43125
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.817080505129726e-05,
+      "loss": 0.637,
+      "step": 43130
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.817016101729504e-05,
+      "loss": 0.6544,
+      "step": 43135
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.8169516881352666e-05,
+      "loss": 0.6615,
+      "step": 43140
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.8168872643478172e-05,
+      "loss": 0.6502,
+      "step": 43145
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.816822830367959e-05,
+      "loss": 0.6516,
+      "step": 43150
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.8167583861964968e-05,
+      "loss": 0.6549,
+      "step": 43155
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.8166939318342348e-05,
+      "loss": 0.6557,
+      "step": 43160
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.8166294672819767e-05,
+      "loss": 0.6415,
+      "step": 43165
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.816564992540527e-05,
+      "loss": 0.649,
+      "step": 43170
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.81650050761069e-05,
+      "loss": 0.662,
+      "step": 43175
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.8164360124932703e-05,
+      "loss": 0.6398,
+      "step": 43180
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.816371507189073e-05,
+      "loss": 0.667,
+      "step": 43185
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.8163069916989027e-05,
+      "loss": 0.6607,
+      "step": 43190
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.8162424660235643e-05,
+      "loss": 0.6438,
+      "step": 43195
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.8161779301638626e-05,
+      "loss": 0.6498,
+      "step": 43200
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.8161133841206036e-05,
+      "loss": 0.6497,
+      "step": 43205
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.816048827894592e-05,
+      "loss": 0.6609,
+      "step": 43210
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.8159842614866336e-05,
+      "loss": 0.6341,
+      "step": 43215
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.815919684897534e-05,
+      "loss": 0.6562,
+      "step": 43220
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.8158550981280987e-05,
+      "loss": 0.6592,
+      "step": 43225
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.8157905011791337e-05,
+      "loss": 0.639,
+      "step": 43230
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.815725894051445e-05,
+      "loss": 0.6541,
+      "step": 43235
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.8156612767458387e-05,
+      "loss": 0.6531,
+      "step": 43240
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.815596649263121e-05,
+      "loss": 0.6502,
+      "step": 43245
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.8155320116040983e-05,
+      "loss": 0.6639,
+      "step": 43250
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.815467363769577e-05,
+      "loss": 0.6449,
+      "step": 43255
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.8154027057603638e-05,
+      "loss": 0.654,
+      "step": 43260
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.8153380375772655e-05,
+      "loss": 0.6675,
+      "step": 43265
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.8152733592210888e-05,
+      "loss": 0.6603,
+      "step": 43270
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.815208670692641e-05,
+      "loss": 0.6419,
+      "step": 43275
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.8151439719927292e-05,
+      "loss": 0.6589,
+      "step": 43280
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.81507926312216e-05,
+      "loss": 0.641,
+      "step": 43285
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.8150145440817417e-05,
+      "loss": 0.6537,
+      "step": 43290
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.814949814872281e-05,
+      "loss": 0.6453,
+      "step": 43295
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.8148850754945865e-05,
+      "loss": 0.6528,
+      "step": 43300
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.814820325949465e-05,
+      "loss": 0.6365,
+      "step": 43305
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.8147555662377246e-05,
+      "loss": 0.6559,
+      "step": 43310
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.814690796360174e-05,
+      "loss": 0.6674,
+      "step": 43315
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.8146260163176205e-05,
+      "loss": 0.6567,
+      "step": 43320
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.814561226110873e-05,
+      "loss": 0.6489,
+      "step": 43325
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.8144964257407388e-05,
+      "loss": 0.646,
+      "step": 43330
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.814431615208028e-05,
+      "loss": 0.6526,
+      "step": 43335
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.814366794513548e-05,
+      "loss": 0.6439,
+      "step": 43340
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.8143019636581084e-05,
+      "loss": 0.6577,
+      "step": 43345
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.8142371226425176e-05,
+      "loss": 0.6414,
+      "step": 43350
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.8141722714675846e-05,
+      "loss": 0.6568,
+      "step": 43355
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.814107410134119e-05,
+      "loss": 0.6413,
+      "step": 43360
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.8140425386429297e-05,
+      "loss": 0.633,
+      "step": 43365
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.8139776569948263e-05,
+      "loss": 0.6467,
+      "step": 43370
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.8139127651906183e-05,
+      "loss": 0.6438,
+      "step": 43375
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.813847863231115e-05,
+      "loss": 0.6557,
+      "step": 43380
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.813782951117127e-05,
+      "loss": 0.6499,
+      "step": 43385
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.8137180288494636e-05,
+      "loss": 0.6549,
+      "step": 43390
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.8136530964289348e-05,
+      "loss": 0.6389,
+      "step": 43395
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.813588153856351e-05,
+      "loss": 0.6541,
+      "step": 43400
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.8135232011325223e-05,
+      "loss": 0.653,
+      "step": 43405
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.8134582382582595e-05,
+      "loss": 0.6313,
+      "step": 43410
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.813393265234373e-05,
+      "loss": 0.6458,
+      "step": 43415
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.813328282061673e-05,
+      "loss": 0.641,
+      "step": 43420
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.8132632887409708e-05,
+      "loss": 0.6505,
+      "step": 43425
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.8131982852730775e-05,
+      "loss": 0.6438,
+      "step": 43430
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.8131332716588038e-05,
+      "loss": 0.6434,
+      "step": 43435
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.813068247898961e-05,
+      "loss": 0.6562,
+      "step": 43440
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.8130032139943604e-05,
+      "loss": 0.6455,
+      "step": 43445
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.812938169945813e-05,
+      "loss": 0.6521,
+      "step": 43450
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.812873115754131e-05,
+      "loss": 0.6446,
+      "step": 43455
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.812808051420126e-05,
+      "loss": 0.6483,
+      "step": 43460
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.81274297694461e-05,
+      "loss": 0.654,
+      "step": 43465
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.8126778923283938e-05,
+      "loss": 0.6418,
+      "step": 43470
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.812612797572291e-05,
+      "loss": 0.6354,
+      "step": 43475
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.8125476926771127e-05,
+      "loss": 0.6569,
+      "step": 43480
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.812482577643672e-05,
+      "loss": 0.6535,
+      "step": 43485
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.8124174524727808e-05,
+      "loss": 0.6525,
+      "step": 43490
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.8123523171652514e-05,
+      "loss": 0.6485,
+      "step": 43495
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.812287171721897e-05,
+      "loss": 0.6311,
+      "step": 43500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.812222016143531e-05,
+      "loss": 0.6469,
+      "step": 43505
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.8121568504309656e-05,
+      "loss": 0.6466,
+      "step": 43510
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.8120916745850134e-05,
+      "loss": 0.6543,
+      "step": 43515
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.8120264886064888e-05,
+      "loss": 0.6395,
+      "step": 43520
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.8119612924962043e-05,
+      "loss": 0.6619,
+      "step": 43525
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.8118960862549735e-05,
+      "loss": 0.6512,
+      "step": 43530
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.81183086988361e-05,
+      "loss": 0.6498,
+      "step": 43535
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.8117656433829283e-05,
+      "loss": 0.648,
+      "step": 43540
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.811700406753741e-05,
+      "loss": 0.6464,
+      "step": 43545
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.8116351599968623e-05,
+      "loss": 0.6423,
+      "step": 43550
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.8115699031131074e-05,
+      "loss": 0.6596,
+      "step": 43555
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.811504636103289e-05,
+      "loss": 0.6514,
+      "step": 43560
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.8114393589682225e-05,
+      "loss": 0.6415,
+      "step": 43565
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.811374071708722e-05,
+      "loss": 0.6452,
+      "step": 43570
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.8113087743256025e-05,
+      "loss": 0.6536,
+      "step": 43575
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.811243466819678e-05,
+      "loss": 0.6413,
+      "step": 43580
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.8111781491917636e-05,
+      "loss": 0.654,
+      "step": 43585
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.8111128214426743e-05,
+      "loss": 0.6465,
+      "step": 43590
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.8110474835732255e-05,
+      "loss": 0.6534,
+      "step": 43595
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.810982135584232e-05,
+      "loss": 0.6508,
+      "step": 43600
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.8109167774765098e-05,
+      "loss": 0.646,
+      "step": 43605
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.8108514092508737e-05,
+      "loss": 0.6524,
+      "step": 43610
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.8107860309081395e-05,
+      "loss": 0.6404,
+      "step": 43615
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.8107206424491227e-05,
+      "loss": 0.6411,
+      "step": 43620
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.81065524387464e-05,
+      "loss": 0.6451,
+      "step": 43625
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.810589835185507e-05,
+      "loss": 0.6511,
+      "step": 43630
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.8105244163825392e-05,
+      "loss": 0.6429,
+      "step": 43635
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.8104589874665533e-05,
+      "loss": 0.6466,
+      "step": 43640
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.810393548438366e-05,
+      "loss": 0.6486,
+      "step": 43645
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.810328099298793e-05,
+      "loss": 0.6502,
+      "step": 43650
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.8102626400486515e-05,
+      "loss": 0.6511,
+      "step": 43655
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.8101971706887583e-05,
+      "loss": 0.6513,
+      "step": 43660
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.8101316912199298e-05,
+      "loss": 0.6509,
+      "step": 43665
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.8100662016429834e-05,
+      "loss": 0.6475,
+      "step": 43670
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.810000701958736e-05,
+      "loss": 0.6299,
+      "step": 43675
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.8099351921680054e-05,
+      "loss": 0.6388,
+      "step": 43680
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.809869672271608e-05,
+      "loss": 0.6499,
+      "step": 43685
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.809804142270362e-05,
+      "loss": 0.6428,
+      "step": 43690
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.809738602165085e-05,
+      "loss": 0.6522,
+      "step": 43695
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.8096730519565945e-05,
+      "loss": 0.6416,
+      "step": 43700
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.8096074916457086e-05,
+      "loss": 0.6422,
+      "step": 43705
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.809541921233245e-05,
+      "loss": 0.6458,
+      "step": 43710
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.8094763407200217e-05,
+      "loss": 0.6437,
+      "step": 43715
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.809410750106858e-05,
+      "loss": 0.6462,
+      "step": 43720
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.809345149394571e-05,
+      "loss": 0.6328,
+      "step": 43725
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.8092795385839798e-05,
+      "loss": 0.6448,
+      "step": 43730
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.809213917675903e-05,
+      "loss": 0.6501,
+      "step": 43735
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.8091482866711595e-05,
+      "loss": 0.6512,
+      "step": 43740
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.8090826455705678e-05,
+      "loss": 0.6456,
+      "step": 43745
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.8090169943749477e-05,
+      "loss": 0.6454,
+      "step": 43750
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.8089513330851173e-05,
+      "loss": 0.6369,
+      "step": 43755
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.8088856617018967e-05,
+      "loss": 0.6393,
+      "step": 43760
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.8088199802261047e-05,
+      "loss": 0.6507,
+      "step": 43765
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.8087542886585612e-05,
+      "loss": 0.6518,
+      "step": 43770
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.8086885870000856e-05,
+      "loss": 0.6491,
+      "step": 43775
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.8086228752514976e-05,
+      "loss": 0.6445,
+      "step": 43780
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.8085571534136174e-05,
+      "loss": 0.6479,
+      "step": 43785
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.808491421487265e-05,
+      "loss": 0.6458,
+      "step": 43790
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.8084256794732605e-05,
+      "loss": 0.6464,
+      "step": 43795
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.8083599273724238e-05,
+      "loss": 0.6549,
+      "step": 43800
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.8082941651855758e-05,
+      "loss": 0.6349,
+      "step": 43805
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.8082283929135368e-05,
+      "loss": 0.6369,
+      "step": 43810
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.8081626105571274e-05,
+      "loss": 0.6389,
+      "step": 43815
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.808096818117169e-05,
+      "loss": 0.6639,
+      "step": 43820
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.808031015594481e-05,
+      "loss": 0.6439,
+      "step": 43825
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.807965202989886e-05,
+      "loss": 0.649,
+      "step": 43830
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.8078993803042046e-05,
+      "loss": 0.6426,
+      "step": 43835
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.8078335475382576e-05,
+      "loss": 0.6395,
+      "step": 43840
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.807767704692867e-05,
+      "loss": 0.6499,
+      "step": 43845
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.8077018517688542e-05,
+      "loss": 0.6623,
+      "step": 43850
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.8076359887670406e-05,
+      "loss": 0.6509,
+      "step": 43855
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.8075701156882485e-05,
+      "loss": 0.6486,
+      "step": 43860
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.8075042325332992e-05,
+      "loss": 0.6485,
+      "step": 43865
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.8074383393030153e-05,
+      "loss": 0.6533,
+      "step": 43870
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.8073724359982184e-05,
+      "loss": 0.6464,
+      "step": 43875
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.8073065226197312e-05,
+      "loss": 0.6523,
+      "step": 43880
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.8072405991683762e-05,
+      "loss": 0.6458,
+      "step": 43885
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.8071746656449752e-05,
+      "loss": 0.6465,
+      "step": 43890
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.8071087220503515e-05,
+      "loss": 0.6492,
+      "step": 43895
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.807042768385328e-05,
+      "loss": 0.6418,
+      "step": 43900
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.8069768046507272e-05,
+      "loss": 0.6479,
+      "step": 43905
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.8069108308473724e-05,
+      "loss": 0.6402,
+      "step": 43910
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.8068448469760867e-05,
+      "loss": 0.6375,
+      "step": 43915
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.806778853037693e-05,
+      "loss": 0.6465,
+      "step": 43920
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.8067128490330153e-05,
+      "loss": 0.6592,
+      "step": 43925
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.8066468349628773e-05,
+      "loss": 0.6468,
+      "step": 43930
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.8065808108281017e-05,
+      "loss": 0.64,
+      "step": 43935
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.806514776629513e-05,
+      "loss": 0.6548,
+      "step": 43940
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.8064487323679354e-05,
+      "loss": 0.6551,
+      "step": 43945
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.806382678044192e-05,
+      "loss": 0.6454,
+      "step": 43950
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.8063166136591077e-05,
+      "loss": 0.6451,
+      "step": 43955
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.8062505392135067e-05,
+      "loss": 0.6558,
+      "step": 43960
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.806184454708213e-05,
+      "loss": 0.6532,
+      "step": 43965
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.8061183601440517e-05,
+      "loss": 0.6504,
+      "step": 43970
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.806052255521847e-05,
+      "loss": 0.6409,
+      "step": 43975
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.8059861408424244e-05,
+      "loss": 0.6551,
+      "step": 43980
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.805920016106608e-05,
+      "loss": 0.6451,
+      "step": 43985
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.8058538813152233e-05,
+      "loss": 0.6467,
+      "step": 43990
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.8057877364690953e-05,
+      "loss": 0.6482,
+      "step": 43995
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.8057215815690494e-05,
+      "loss": 0.6496,
+      "step": 44000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.805655416615911e-05,
+      "loss": 0.6528,
+      "step": 44005
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.8055892416105053e-05,
+      "loss": 0.6422,
+      "step": 44010
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.805523056553659e-05,
+      "loss": 0.6535,
+      "step": 44015
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.8054568614461966e-05,
+      "loss": 0.6413,
+      "step": 44020
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.805390656288945e-05,
+      "loss": 0.6465,
+      "step": 44025
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.80532444108273e-05,
+      "loss": 0.6427,
+      "step": 44030
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.8052582158283768e-05,
+      "loss": 0.6514,
+      "step": 44035
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.8051919805267136e-05,
+      "loss": 0.6484,
+      "step": 44040
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.805125735178565e-05,
+      "loss": 0.65,
+      "step": 44045
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.8050594797847587e-05,
+      "loss": 0.6445,
+      "step": 44050
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.804993214346121e-05,
+      "loss": 0.651,
+      "step": 44055
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.8049269388634788e-05,
+      "loss": 0.6469,
+      "step": 44060
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.8048606533376588e-05,
+      "loss": 0.6587,
+      "step": 44065
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.8047943577694883e-05,
+      "loss": 0.6443,
+      "step": 44070
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.8047280521597943e-05,
+      "loss": 0.6372,
+      "step": 44075
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.804661736509404e-05,
+      "loss": 0.6643,
+      "step": 44080
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.8045954108191455e-05,
+      "loss": 0.6425,
+      "step": 44085
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.8045290750898456e-05,
+      "loss": 0.6488,
+      "step": 44090
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.8044627293223322e-05,
+      "loss": 0.644,
+      "step": 44095
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.804396373517433e-05,
+      "loss": 0.648,
+      "step": 44100
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.8043300076759763e-05,
+      "loss": 0.6493,
+      "step": 44105
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.8042636317987896e-05,
+      "loss": 0.6481,
+      "step": 44110
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.804197245886702e-05,
+      "loss": 0.6413,
+      "step": 44115
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.8041308499405406e-05,
+      "loss": 0.6577,
+      "step": 44120
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.8040644439611348e-05,
+      "loss": 0.6438,
+      "step": 44125
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.803998027949313e-05,
+      "loss": 0.6366,
+      "step": 44130
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.8039316019059033e-05,
+      "loss": 0.6444,
+      "step": 44135
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.803865165831735e-05,
+      "loss": 0.6447,
+      "step": 44140
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.803798719727637e-05,
+      "loss": 0.6415,
+      "step": 44145
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.8037322635944383e-05,
+      "loss": 0.6519,
+      "step": 44150
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.803665797432968e-05,
+      "loss": 0.6422,
+      "step": 44155
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.803599321244055e-05,
+      "loss": 0.6415,
+      "step": 44160
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.8035328350285298e-05,
+      "loss": 0.6428,
+      "step": 44165
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.8034663387872213e-05,
+      "loss": 0.6494,
+      "step": 44170
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.803399832520959e-05,
+      "loss": 0.6573,
+      "step": 44175
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.8033333162305732e-05,
+      "loss": 0.6418,
+      "step": 44180
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.8032667899168937e-05,
+      "loss": 0.656,
+      "step": 44185
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.80320025358075e-05,
+      "loss": 0.6416,
+      "step": 44190
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.803133707222973e-05,
+      "loss": 0.646,
+      "step": 44195
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.8030671508443928e-05,
+      "loss": 0.6497,
+      "step": 44200
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.80300058444584e-05,
+      "loss": 0.6421,
+      "step": 44205
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.802934008028144e-05,
+      "loss": 0.6443,
+      "step": 44210
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.8028674215921375e-05,
+      "loss": 0.6491,
+      "step": 44215
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.8028008251386493e-05,
+      "loss": 0.6479,
+      "step": 44220
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.8027342186685114e-05,
+      "loss": 0.6553,
+      "step": 44225
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.802667602182555e-05,
+      "loss": 0.6538,
+      "step": 44230
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.8026009756816113e-05,
+      "loss": 0.638,
+      "step": 44235
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.8025343391665107e-05,
+      "loss": 0.6297,
+      "step": 44240
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.8024676926380854e-05,
+      "loss": 0.6542,
+      "step": 44245
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.802401036097167e-05,
+      "loss": 0.6517,
+      "step": 44250
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.8023343695445863e-05,
+      "loss": 0.6556,
+      "step": 44255
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.8022676929811763e-05,
+      "loss": 0.6416,
+      "step": 44260
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.8022010064077685e-05,
+      "loss": 0.642,
+      "step": 44265
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.8021343098251943e-05,
+      "loss": 0.6438,
+      "step": 44270
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.802067603234287e-05,
+      "loss": 0.649,
+      "step": 44275
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.8020008866358782e-05,
+      "loss": 0.6419,
+      "step": 44280
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.8019341600308005e-05,
+      "loss": 0.6557,
+      "step": 44285
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.801867423419886e-05,
+      "loss": 0.6468,
+      "step": 44290
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.8018006768039684e-05,
+      "loss": 0.6479,
+      "step": 44295
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.8017339201838798e-05,
+      "loss": 0.6437,
+      "step": 44300
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.801667153560453e-05,
+      "loss": 0.644,
+      "step": 44305
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.8016003769345216e-05,
+      "loss": 0.6377,
+      "step": 44310
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.8015335903069184e-05,
+      "loss": 0.6488,
+      "step": 44315
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.8014667936784764e-05,
+      "loss": 0.6441,
+      "step": 44320
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.8013999870500298e-05,
+      "loss": 0.6289,
+      "step": 44325
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.8013331704224118e-05,
+      "loss": 0.6422,
+      "step": 44330
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.801266343796456e-05,
+      "loss": 0.641,
+      "step": 44335
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.801199507172996e-05,
+      "loss": 0.6434,
+      "step": 44340
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.8011326605528664e-05,
+      "loss": 0.6464,
+      "step": 44345
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.8010658039369006e-05,
+      "loss": 0.6435,
+      "step": 44350
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.8009989373259334e-05,
+      "loss": 0.6427,
+      "step": 44355
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.8009320607207984e-05,
+      "loss": 0.6376,
+      "step": 44360
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.8008651741223304e-05,
+      "loss": 0.6435,
+      "step": 44365
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.800798277531364e-05,
+      "loss": 0.6314,
+      "step": 44370
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.8007313709487334e-05,
+      "loss": 0.6537,
+      "step": 44375
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.800664454375274e-05,
+      "loss": 0.6524,
+      "step": 44380
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.8005975278118208e-05,
+      "loss": 0.6552,
+      "step": 44385
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.8005305912592087e-05,
+      "loss": 0.6475,
+      "step": 44390
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.8004636447182725e-05,
+      "loss": 0.6398,
+      "step": 44395
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.8003966881898473e-05,
+      "loss": 0.6577,
+      "step": 44400
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.8003297216747694e-05,
+      "loss": 0.6538,
+      "step": 44405
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.800262745173874e-05,
+      "loss": 0.6462,
+      "step": 44410
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.8001957586879965e-05,
+      "loss": 0.6424,
+      "step": 44415
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.8001287622179728e-05,
+      "loss": 0.6559,
+      "step": 44420
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.8000617557646392e-05,
+      "loss": 0.636,
+      "step": 44425
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.799994739328831e-05,
+      "loss": 0.6383,
+      "step": 44430
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7999277129113855e-05,
+      "loss": 0.6423,
+      "step": 44435
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.799860676513138e-05,
+      "loss": 0.6457,
+      "step": 44440
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.799793630134925e-05,
+      "loss": 0.64,
+      "step": 44445
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7997265737775837e-05,
+      "loss": 0.6437,
+      "step": 44450
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.79965950744195e-05,
+      "loss": 0.64,
+      "step": 44455
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7995924311288613e-05,
+      "loss": 0.654,
+      "step": 44460
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7995253448391543e-05,
+      "loss": 0.6351,
+      "step": 44465
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7994582485736664e-05,
+      "loss": 0.6488,
+      "step": 44470
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7993911423332336e-05,
+      "loss": 0.6399,
+      "step": 44475
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7993240261186945e-05,
+      "loss": 0.6395,
+      "step": 44480
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7992568999308858e-05,
+      "loss": 0.6428,
+      "step": 44485
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7991897637706455e-05,
+      "loss": 0.6408,
+      "step": 44490
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.799122617638811e-05,
+      "loss": 0.6441,
+      "step": 44495
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.79905546153622e-05,
+      "loss": 0.6515,
+      "step": 44500
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7989882954637105e-05,
+      "loss": 0.6423,
+      "step": 44505
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7989211194221206e-05,
+      "loss": 0.6323,
+      "step": 44510
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7988539334122882e-05,
+      "loss": 0.6301,
+      "step": 44515
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.798786737435052e-05,
+      "loss": 0.6468,
+      "step": 44520
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7987195314912504e-05,
+      "loss": 0.6514,
+      "step": 44525
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.798652315581722e-05,
+      "loss": 0.6429,
+      "step": 44530
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7985850897073045e-05,
+      "loss": 0.6422,
+      "step": 44535
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.798517853868838e-05,
+      "loss": 0.6455,
+      "step": 44540
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7984506080671608e-05,
+      "loss": 0.6364,
+      "step": 44545
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7983833523031114e-05,
+      "loss": 0.6417,
+      "step": 44550
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.79831608657753e-05,
+      "loss": 0.6526,
+      "step": 44555
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7982488108912553e-05,
+      "loss": 0.6593,
+      "step": 44560
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.798181525245127e-05,
+      "loss": 0.6481,
+      "step": 44565
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.798114229639984e-05,
+      "loss": 0.6392,
+      "step": 44570
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.798046924076667e-05,
+      "loss": 0.6483,
+      "step": 44575
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.797979608556015e-05,
+      "loss": 0.6449,
+      "step": 44580
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7979122830788677e-05,
+      "loss": 0.6459,
+      "step": 44585
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.797844947646066e-05,
+      "loss": 0.6454,
+      "step": 44590
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.797777602258449e-05,
+      "loss": 0.6556,
+      "step": 44595
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7977102469168578e-05,
+      "loss": 0.6449,
+      "step": 44600
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7976428816221327e-05,
+      "loss": 0.6327,
+      "step": 44605
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.797575506375114e-05,
+      "loss": 0.6406,
+      "step": 44610
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7975081211766424e-05,
+      "loss": 0.6371,
+      "step": 44615
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7974407260275587e-05,
+      "loss": 0.6379,
+      "step": 44620
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7973733209287036e-05,
+      "loss": 0.6402,
+      "step": 44625
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7973059058809186e-05,
+      "loss": 0.6496,
+      "step": 44630
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7972384808850444e-05,
+      "loss": 0.6435,
+      "step": 44635
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7971710459419226e-05,
+      "loss": 0.6441,
+      "step": 44640
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7971036010523944e-05,
+      "loss": 0.6482,
+      "step": 44645
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.797036146217301e-05,
+      "loss": 0.6541,
+      "step": 44650
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.796968681437485e-05,
+      "loss": 0.6428,
+      "step": 44655
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.796901206713787e-05,
+      "loss": 0.6406,
+      "step": 44660
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.79683372204705e-05,
+      "loss": 0.6476,
+      "step": 44665
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7967662274381148e-05,
+      "loss": 0.6503,
+      "step": 44670
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7966987228878247e-05,
+      "loss": 0.6439,
+      "step": 44675
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.796631208397021e-05,
+      "loss": 0.6276,
+      "step": 44680
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7965636839665466e-05,
+      "loss": 0.6434,
+      "step": 44685
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7964961495972443e-05,
+      "loss": 0.6465,
+      "step": 44690
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.796428605289956e-05,
+      "loss": 0.6367,
+      "step": 44695
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7963610510455254e-05,
+      "loss": 0.65,
+      "step": 44700
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7962934868647944e-05,
+      "loss": 0.6392,
+      "step": 44705
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.796225912748607e-05,
+      "loss": 0.6328,
+      "step": 44710
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7961583286978054e-05,
+      "loss": 0.6433,
+      "step": 44715
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7960907347132332e-05,
+      "loss": 0.6465,
+      "step": 44720
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7960231307957333e-05,
+      "loss": 0.6363,
+      "step": 44725
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7959555169461504e-05,
+      "loss": 0.6505,
+      "step": 44730
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.795887893165327e-05,
+      "loss": 0.644,
+      "step": 44735
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7958202594541078e-05,
+      "loss": 0.6427,
+      "step": 44740
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7957526158133357e-05,
+      "loss": 0.649,
+      "step": 44745
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7956849622438554e-05,
+      "loss": 0.6264,
+      "step": 44750
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7956172987465107e-05,
+      "loss": 0.6376,
+      "step": 44755
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7955496253221462e-05,
+      "loss": 0.651,
+      "step": 44760
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.795481941971606e-05,
+      "loss": 0.6503,
+      "step": 44765
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.795414248695734e-05,
+      "loss": 0.6267,
+      "step": 44770
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.795346545495376e-05,
+      "loss": 0.6473,
+      "step": 44775
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7952788323713755e-05,
+      "loss": 0.6398,
+      "step": 44780
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.795211109324579e-05,
+      "loss": 0.6366,
+      "step": 44785
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7951433763558297e-05,
+      "loss": 0.6472,
+      "step": 44790
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7950756334659737e-05,
+      "loss": 0.6567,
+      "step": 44795
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7950078806558565e-05,
+      "loss": 0.653,
+      "step": 44800
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7949401179263225e-05,
+      "loss": 0.6337,
+      "step": 44805
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.794872345278218e-05,
+      "loss": 0.639,
+      "step": 44810
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.794804562712388e-05,
+      "loss": 0.644,
+      "step": 44815
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.794736770229679e-05,
+      "loss": 0.6555,
+      "step": 44820
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7946689678309356e-05,
+      "loss": 0.6472,
+      "step": 44825
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7946011555170053e-05,
+      "loss": 0.6397,
+      "step": 44830
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7945333332887332e-05,
+      "loss": 0.6477,
+      "step": 44835
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.794465501146966e-05,
+      "loss": 0.645,
+      "step": 44840
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7943976590925496e-05,
+      "loss": 0.6323,
+      "step": 44845
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.794329807126331e-05,
+      "loss": 0.6519,
+      "step": 44850
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7942619452491562e-05,
+      "loss": 0.644,
+      "step": 44855
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7941940734618727e-05,
+      "loss": 0.6458,
+      "step": 44860
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7941261917653266e-05,
+      "loss": 0.6302,
+      "step": 44865
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7940583001603653e-05,
+      "loss": 0.6498,
+      "step": 44870
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7939903986478354e-05,
+      "loss": 0.6397,
+      "step": 44875
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.793922487228585e-05,
+      "loss": 0.6505,
+      "step": 44880
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7938545659034602e-05,
+      "loss": 0.6453,
+      "step": 44885
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.79378663467331e-05,
+      "loss": 0.6304,
+      "step": 44890
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7937186935389807e-05,
+      "loss": 0.651,
+      "step": 44895
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7936507425013204e-05,
+      "loss": 0.6441,
+      "step": 44900
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7935827815611775e-05,
+      "loss": 0.6479,
+      "step": 44905
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.793514810719399e-05,
+      "loss": 0.6469,
+      "step": 44910
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7934468299768336e-05,
+      "loss": 0.6277,
+      "step": 44915
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.793378839334329e-05,
+      "loss": 0.641,
+      "step": 44920
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7933108387927346e-05,
+      "loss": 0.6344,
+      "step": 44925
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7932428283528976e-05,
+      "loss": 0.6373,
+      "step": 44930
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7931748080156674e-05,
+      "loss": 0.6519,
+      "step": 44935
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7931067777818922e-05,
+      "loss": 0.6494,
+      "step": 44940
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.793038737652421e-05,
+      "loss": 0.6519,
+      "step": 44945
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.792970687628103e-05,
+      "loss": 0.6368,
+      "step": 44950
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.792902627709787e-05,
+      "loss": 0.64,
+      "step": 44955
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.792834557898322e-05,
+      "loss": 0.6429,
+      "step": 44960
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7927664781945573e-05,
+      "loss": 0.6527,
+      "step": 44965
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7926983885993434e-05,
+      "loss": 0.6385,
+      "step": 44970
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7926302891135283e-05,
+      "loss": 0.6374,
+      "step": 44975
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7925621797379626e-05,
+      "loss": 0.651,
+      "step": 44980
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7924940604734962e-05,
+      "loss": 0.6447,
+      "step": 44985
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7924259313209787e-05,
+      "loss": 0.6317,
+      "step": 44990
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7923577922812604e-05,
+      "loss": 0.6408,
+      "step": 44995
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.792289643355191e-05,
+      "loss": 0.6515,
+      "step": 45000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.792221484543621e-05,
+      "loss": 0.6432,
+      "step": 45005
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.792153315847401e-05,
+      "loss": 0.6323,
+      "step": 45010
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7920851372673818e-05,
+      "loss": 0.6349,
+      "step": 45015
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7920169488044138e-05,
+      "loss": 0.6487,
+      "step": 45020
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7919487504593472e-05,
+      "loss": 0.6522,
+      "step": 45025
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7918805422330342e-05,
+      "loss": 0.6495,
+      "step": 45030
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7918123241263247e-05,
+      "loss": 0.6425,
+      "step": 45035
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7917440961400706e-05,
+      "loss": 0.644,
+      "step": 45040
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7916758582751227e-05,
+      "loss": 0.6522,
+      "step": 45045
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7916076105323323e-05,
+      "loss": 0.6382,
+      "step": 45050
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7915393529125516e-05,
+      "loss": 0.6448,
+      "step": 45055
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7914710854166317e-05,
+      "loss": 0.643,
+      "step": 45060
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7914028080454247e-05,
+      "loss": 0.6342,
+      "step": 45065
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7913345207997824e-05,
+      "loss": 0.6442,
+      "step": 45070
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.791266223680557e-05,
+      "loss": 0.6439,
+      "step": 45075
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7911979166886e-05,
+      "loss": 0.6492,
+      "step": 45080
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.791129599824764e-05,
+      "loss": 0.6466,
+      "step": 45085
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7910612730899018e-05,
+      "loss": 0.644,
+      "step": 45090
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.790992936484866e-05,
+      "loss": 0.6418,
+      "step": 45095
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7909245900105085e-05,
+      "loss": 0.6278,
+      "step": 45100
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7908562336676826e-05,
+      "loss": 0.6487,
+      "step": 45105
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.790787867457241e-05,
+      "loss": 0.647,
+      "step": 45110
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7907194913800367e-05,
+      "loss": 0.6445,
+      "step": 45115
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.790651105436923e-05,
+      "loss": 0.6446,
+      "step": 45120
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7905827096287532e-05,
+      "loss": 0.6519,
+      "step": 45125
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.79051430395638e-05,
+      "loss": 0.6492,
+      "step": 45130
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.790445888420658e-05,
+      "loss": 0.6392,
+      "step": 45135
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7903774630224404e-05,
+      "loss": 0.6441,
+      "step": 45140
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7903090277625807e-05,
+      "loss": 0.6353,
+      "step": 45145
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7902405826419323e-05,
+      "loss": 0.6458,
+      "step": 45150
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7901721276613506e-05,
+      "loss": 0.6523,
+      "step": 45155
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7901036628216885e-05,
+      "loss": 0.647,
+      "step": 45160
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7900351881238013e-05,
+      "loss": 0.6425,
+      "step": 45165
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.789966703568542e-05,
+      "loss": 0.6447,
+      "step": 45170
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.789898209156766e-05,
+      "loss": 0.6324,
+      "step": 45175
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.789829704889328e-05,
+      "loss": 0.6464,
+      "step": 45180
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7897611907670823e-05,
+      "loss": 0.649,
+      "step": 45185
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.789692666790884e-05,
+      "loss": 0.6413,
+      "step": 45190
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.789624132961588e-05,
+      "loss": 0.6492,
+      "step": 45195
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7895555892800494e-05,
+      "loss": 0.6451,
+      "step": 45200
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7894870357471236e-05,
+      "loss": 0.635,
+      "step": 45205
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7894184723636657e-05,
+      "loss": 0.6435,
+      "step": 45210
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7893498991305317e-05,
+      "loss": 0.6392,
+      "step": 45215
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7892813160485762e-05,
+      "loss": 0.641,
+      "step": 45220
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.789212723118656e-05,
+      "loss": 0.6339,
+      "step": 45225
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7891441203416258e-05,
+      "loss": 0.6448,
+      "step": 45230
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7890755077183426e-05,
+      "loss": 0.6405,
+      "step": 45235
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7890068852496624e-05,
+      "loss": 0.649,
+      "step": 45240
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7889382529364407e-05,
+      "loss": 0.6307,
+      "step": 45245
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7888696107795343e-05,
+      "loss": 0.6358,
+      "step": 45250
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7888009587797995e-05,
+      "loss": 0.6411,
+      "step": 45255
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7887322969380935e-05,
+      "loss": 0.6398,
+      "step": 45260
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7886636252552717e-05,
+      "loss": 0.636,
+      "step": 45265
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7885949437321923e-05,
+      "loss": 0.642,
+      "step": 45270
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7885262523697116e-05,
+      "loss": 0.6323,
+      "step": 45275
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7884575511686867e-05,
+      "loss": 0.644,
+      "step": 45280
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.788388840129975e-05,
+      "loss": 0.6418,
+      "step": 45285
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7883201192544332e-05,
+      "loss": 0.6484,
+      "step": 45290
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7882513885429197e-05,
+      "loss": 0.6381,
+      "step": 45295
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.788182647996291e-05,
+      "loss": 0.648,
+      "step": 45300
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7881138976154055e-05,
+      "loss": 0.6416,
+      "step": 45305
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.788045137401121e-05,
+      "loss": 0.6447,
+      "step": 45310
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.787976367354295e-05,
+      "loss": 0.6391,
+      "step": 45315
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.787907587475786e-05,
+      "loss": 0.642,
+      "step": 45320
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7878387977664522e-05,
+      "loss": 0.6387,
+      "step": 45325
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7877699982271516e-05,
+      "loss": 0.6408,
+      "step": 45330
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7877011888587427e-05,
+      "loss": 0.6254,
+      "step": 45335
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7876323696620838e-05,
+      "loss": 0.6452,
+      "step": 45340
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.787563540638034e-05,
+      "loss": 0.6474,
+      "step": 45345
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.787494701787452e-05,
+      "loss": 0.6569,
+      "step": 45350
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7874258531111963e-05,
+      "loss": 0.648,
+      "step": 45355
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7873569946101266e-05,
+      "loss": 0.6504,
+      "step": 45360
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7872881262851018e-05,
+      "loss": 0.6381,
+      "step": 45365
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.787219248136981e-05,
+      "loss": 0.6484,
+      "step": 45370
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7871503601666233e-05,
+      "loss": 0.6476,
+      "step": 45375
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.787081462374889e-05,
+      "loss": 0.6471,
+      "step": 45380
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7870125547626372e-05,
+      "loss": 0.6398,
+      "step": 45385
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7869436373307278e-05,
+      "loss": 0.6501,
+      "step": 45390
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.786874710080021e-05,
+      "loss": 0.6406,
+      "step": 45395
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7868057730113766e-05,
+      "loss": 0.6387,
+      "step": 45400
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7867368261256542e-05,
+      "loss": 0.6417,
+      "step": 45405
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.786667869423715e-05,
+      "loss": 0.6506,
+      "step": 45410
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7865989029064185e-05,
+      "loss": 0.6264,
+      "step": 45415
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7865299265746257e-05,
+      "loss": 0.6342,
+      "step": 45420
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7864609404291976e-05,
+      "loss": 0.6481,
+      "step": 45425
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.786391944470994e-05,
+      "loss": 0.6485,
+      "step": 45430
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7863229387008765e-05,
+      "loss": 0.6509,
+      "step": 45435
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.786253923119706e-05,
+      "loss": 0.6411,
+      "step": 45440
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.786184897728343e-05,
+      "loss": 0.6407,
+      "step": 45445
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.78611586252765e-05,
+      "loss": 0.6371,
+      "step": 45450
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7860468175184867e-05,
+      "loss": 0.6535,
+      "step": 45455
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.785977762701716e-05,
+      "loss": 0.6382,
+      "step": 45460
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.785908698078199e-05,
+      "loss": 0.6382,
+      "step": 45465
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.785839623648797e-05,
+      "loss": 0.6475,
+      "step": 45470
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7857705394143726e-05,
+      "loss": 0.6384,
+      "step": 45475
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.785701445375787e-05,
+      "loss": 0.6483,
+      "step": 45480
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.785632341533903e-05,
+      "loss": 0.6511,
+      "step": 45485
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7855632278895826e-05,
+      "loss": 0.6393,
+      "step": 45490
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7854941044436883e-05,
+      "loss": 0.6452,
+      "step": 45495
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.785424971197082e-05,
+      "loss": 0.6413,
+      "step": 45500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7853558281506267e-05,
+      "loss": 0.6276,
+      "step": 45505
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.785286675305185e-05,
+      "loss": 0.6379,
+      "step": 45510
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.78521751266162e-05,
+      "loss": 0.6458,
+      "step": 45515
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7851483402207937e-05,
+      "loss": 0.6388,
+      "step": 45520
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7850791579835703e-05,
+      "loss": 0.6574,
+      "step": 45525
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7850099659508128e-05,
+      "loss": 0.6356,
+      "step": 45530
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7849407641233843e-05,
+      "loss": 0.6484,
+      "step": 45535
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7848715525021482e-05,
+      "loss": 0.6522,
+      "step": 45540
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7848023310879678e-05,
+      "loss": 0.6524,
+      "step": 45545
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.784733099881707e-05,
+      "loss": 0.6343,
+      "step": 45550
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7846638588842302e-05,
+      "loss": 0.6301,
+      "step": 45555
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7845946080964004e-05,
+      "loss": 0.642,
+      "step": 45560
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7845253475190825e-05,
+      "loss": 0.646,
+      "step": 45565
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7844560771531397e-05,
+      "loss": 0.6421,
+      "step": 45570
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7843867969994372e-05,
+      "loss": 0.6521,
+      "step": 45575
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.784317507058839e-05,
+      "loss": 0.6465,
+      "step": 45580
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7842482073322097e-05,
+      "loss": 0.6418,
+      "step": 45585
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7841788978204137e-05,
+      "loss": 0.6383,
+      "step": 45590
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.784109578524316e-05,
+      "loss": 0.6536,
+      "step": 45595
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.784040249444782e-05,
+      "loss": 0.6547,
+      "step": 45600
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.783970910582676e-05,
+      "loss": 0.6371,
+      "step": 45605
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.783901561938864e-05,
+      "loss": 0.6372,
+      "step": 45610
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7838322035142098e-05,
+      "loss": 0.6201,
+      "step": 45615
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.78376283530958e-05,
+      "loss": 0.6474,
+      "step": 45620
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.78369345732584e-05,
+      "loss": 0.6394,
+      "step": 45625
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.783624069563855e-05,
+      "loss": 0.6407,
+      "step": 45630
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.783554672024491e-05,
+      "loss": 0.6413,
+      "step": 45635
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.783485264708614e-05,
+      "loss": 0.642,
+      "step": 45640
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7834158476170902e-05,
+      "loss": 0.6434,
+      "step": 45645
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.783346420750785e-05,
+      "loss": 0.6297,
+      "step": 45650
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.783276984110565e-05,
+      "loss": 0.6536,
+      "step": 45655
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7832075376972972e-05,
+      "loss": 0.6382,
+      "step": 45660
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.783138081511847e-05,
+      "loss": 0.6616,
+      "step": 45665
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7830686155550815e-05,
+      "loss": 0.6326,
+      "step": 45670
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.782999139827868e-05,
+      "loss": 0.6332,
+      "step": 45675
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7829296543310725e-05,
+      "loss": 0.6305,
+      "step": 45680
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.782860159065562e-05,
+      "loss": 0.6417,
+      "step": 45685
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.782790654032204e-05,
+      "loss": 0.6475,
+      "step": 45690
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7827211392318662e-05,
+      "loss": 0.6508,
+      "step": 45695
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.782651614665415e-05,
+      "loss": 0.6464,
+      "step": 45700
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.782582080333718e-05,
+      "loss": 0.6295,
+      "step": 45705
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7825125362376434e-05,
+      "loss": 0.6505,
+      "step": 45710
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7824429823780586e-05,
+      "loss": 0.655,
+      "step": 45715
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7823734187558312e-05,
+      "loss": 0.6274,
+      "step": 45720
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7823038453718293e-05,
+      "loss": 0.643,
+      "step": 45725
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.782234262226921e-05,
+      "loss": 0.6443,
+      "step": 45730
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7821646693219748e-05,
+      "loss": 0.6421,
+      "step": 45735
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7820950666578586e-05,
+      "loss": 0.6501,
+      "step": 45740
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.782025454235441e-05,
+      "loss": 0.6423,
+      "step": 45745
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7819558320555902e-05,
+      "loss": 0.6445,
+      "step": 45750
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7818862001191754e-05,
+      "loss": 0.6522,
+      "step": 45755
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.781816558427065e-05,
+      "loss": 0.6437,
+      "step": 45760
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7817469069801287e-05,
+      "loss": 0.6471,
+      "step": 45765
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7816772457792348e-05,
+      "loss": 0.6426,
+      "step": 45770
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7816075748252526e-05,
+      "loss": 0.6449,
+      "step": 45775
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7815378941190514e-05,
+      "loss": 0.6435,
+      "step": 45780
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.781468203661501e-05,
+      "loss": 0.6366,
+      "step": 45785
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7813985034534704e-05,
+      "loss": 0.6293,
+      "step": 45790
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7813287934958293e-05,
+      "loss": 0.6471,
+      "step": 45795
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.781259073789448e-05,
+      "loss": 0.648,
+      "step": 45800
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7811893443351958e-05,
+      "loss": 0.6487,
+      "step": 45805
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.781119605133943e-05,
+      "loss": 0.639,
+      "step": 45810
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.78104985618656e-05,
+      "loss": 0.6521,
+      "step": 45815
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7809800974939168e-05,
+      "loss": 0.6509,
+      "step": 45820
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7809103290568832e-05,
+      "loss": 0.636,
+      "step": 45825
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.780840550876331e-05,
+      "loss": 0.6427,
+      "step": 45830
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7807707629531297e-05,
+      "loss": 0.6372,
+      "step": 45835
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7807009652881505e-05,
+      "loss": 0.656,
+      "step": 45840
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7806311578822644e-05,
+      "loss": 0.6503,
+      "step": 45845
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.780561340736342e-05,
+      "loss": 0.6438,
+      "step": 45850
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.780491513851255e-05,
+      "loss": 0.6365,
+      "step": 45855
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.780421677227874e-05,
+      "loss": 0.6337,
+      "step": 45860
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7803518308670708e-05,
+      "loss": 0.6441,
+      "step": 45865
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7802819747697172e-05,
+      "loss": 0.6404,
+      "step": 45870
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.780212108936684e-05,
+      "loss": 0.6414,
+      "step": 45875
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.780142233368843e-05,
+      "loss": 0.6447,
+      "step": 45880
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7800723480670666e-05,
+      "loss": 0.6408,
+      "step": 45885
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.780002453032226e-05,
+      "loss": 0.6438,
+      "step": 45890
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7799325482651946e-05,
+      "loss": 0.6575,
+      "step": 45895
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.779862633766843e-05,
+      "loss": 0.6377,
+      "step": 45900
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7797927095380448e-05,
+      "loss": 0.6475,
+      "step": 45905
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.779722775579672e-05,
+      "loss": 0.6476,
+      "step": 45910
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.779652831892597e-05,
+      "loss": 0.6492,
+      "step": 45915
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.779582878477693e-05,
+      "loss": 0.6501,
+      "step": 45920
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.779512915335832e-05,
+      "loss": 0.6494,
+      "step": 45925
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7794429424678877e-05,
+      "loss": 0.6515,
+      "step": 45930
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7793729598747327e-05,
+      "loss": 0.6393,
+      "step": 45935
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7793029675572408e-05,
+      "loss": 0.6515,
+      "step": 45940
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7792329655162843e-05,
+      "loss": 0.6466,
+      "step": 45945
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7791629537527373e-05,
+      "loss": 0.6577,
+      "step": 45950
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7790929322674733e-05,
+      "loss": 0.6427,
+      "step": 45955
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.779022901061366e-05,
+      "loss": 0.6375,
+      "step": 45960
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7789528601352887e-05,
+      "loss": 0.6522,
+      "step": 45965
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.778882809490116e-05,
+      "loss": 0.6332,
+      "step": 45970
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7788127491267214e-05,
+      "loss": 0.6466,
+      "step": 45975
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7787426790459792e-05,
+      "loss": 0.649,
+      "step": 45980
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7786725992487643e-05,
+      "loss": 0.641,
+      "step": 45985
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7786025097359502e-05,
+      "loss": 0.649,
+      "step": 45990
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7785324105084117e-05,
+      "loss": 0.6498,
+      "step": 45995
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7784623015670237e-05,
+      "loss": 0.6518,
+      "step": 46000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7783921829126605e-05,
+      "loss": 0.6459,
+      "step": 46005
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.778322054546197e-05,
+      "loss": 0.6461,
+      "step": 46010
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.778251916468509e-05,
+      "loss": 0.6365,
+      "step": 46015
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.778181768680471e-05,
+      "loss": 0.6411,
+      "step": 46020
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.778111611182958e-05,
+      "loss": 0.634,
+      "step": 46025
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7780414439768456e-05,
+      "loss": 0.6516,
+      "step": 46030
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7779712670630097e-05,
+      "loss": 0.644,
+      "step": 46035
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7779010804423254e-05,
+      "loss": 0.6423,
+      "step": 46040
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7778308841156687e-05,
+      "loss": 0.6384,
+      "step": 46045
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7777606780839152e-05,
+      "loss": 0.6475,
+      "step": 46050
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7776904623479407e-05,
+      "loss": 0.6339,
+      "step": 46055
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.777620236908622e-05,
+      "loss": 0.6475,
+      "step": 46060
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7775500017668344e-05,
+      "loss": 0.649,
+      "step": 46065
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7774797569234553e-05,
+      "loss": 0.6426,
+      "step": 46070
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7774095023793606e-05,
+      "loss": 0.6351,
+      "step": 46075
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7773392381354267e-05,
+      "loss": 0.6463,
+      "step": 46080
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.77726896419253e-05,
+      "loss": 0.6386,
+      "step": 46085
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7771986805515485e-05,
+      "loss": 0.6439,
+      "step": 46090
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7771283872133583e-05,
+      "loss": 0.635,
+      "step": 46095
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7770580841788364e-05,
+      "loss": 0.6469,
+      "step": 46100
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.77698777144886e-05,
+      "loss": 0.6461,
+      "step": 46105
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7769174490243068e-05,
+      "loss": 0.6381,
+      "step": 46110
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7768471169060538e-05,
+      "loss": 0.653,
+      "step": 46115
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7767767750949784e-05,
+      "loss": 0.6432,
+      "step": 46120
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7767064235919594e-05,
+      "loss": 0.6457,
+      "step": 46125
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.776636062397873e-05,
+      "loss": 0.6404,
+      "step": 46130
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7765656915135983e-05,
+      "loss": 0.6439,
+      "step": 46135
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.776495310940013e-05,
+      "loss": 0.6407,
+      "step": 46140
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.776424920677995e-05,
+      "loss": 0.6396,
+      "step": 46145
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7763545207284226e-05,
+      "loss": 0.6384,
+      "step": 46150
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7762841110921744e-05,
+      "loss": 0.6376,
+      "step": 46155
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7762136917701288e-05,
+      "loss": 0.6374,
+      "step": 46160
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7761432627631644e-05,
+      "loss": 0.6495,
+      "step": 46165
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.77607282407216e-05,
+      "loss": 0.6433,
+      "step": 46170
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7760023756979944e-05,
+      "loss": 0.6347,
+      "step": 46175
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.775931917641547e-05,
+      "loss": 0.6443,
+      "step": 46180
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.775861449903696e-05,
+      "loss": 0.6492,
+      "step": 46185
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7757909724853218e-05,
+      "loss": 0.6487,
+      "step": 46190
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.775720485387303e-05,
+      "loss": 0.6335,
+      "step": 46195
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.775649988610519e-05,
+      "loss": 0.641,
+      "step": 46200
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.77557948215585e-05,
+      "loss": 0.6408,
+      "step": 46205
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.775508966024175e-05,
+      "loss": 0.6443,
+      "step": 46210
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7754384402163746e-05,
+      "loss": 0.6361,
+      "step": 46215
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.775367904733328e-05,
+      "loss": 0.6361,
+      "step": 46220
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.775297359575916e-05,
+      "loss": 0.6408,
+      "step": 46225
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.775226804745018e-05,
+      "loss": 0.6332,
+      "step": 46230
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7751562402415152e-05,
+      "loss": 0.6468,
+      "step": 46235
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7750856660662872e-05,
+      "loss": 0.6491,
+      "step": 46240
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.775015082220215e-05,
+      "loss": 0.6386,
+      "step": 46245
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7749444887041797e-05,
+      "loss": 0.6388,
+      "step": 46250
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7748738855190615e-05,
+      "loss": 0.6404,
+      "step": 46255
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7748032726657413e-05,
+      "loss": 0.6382,
+      "step": 46260
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7747326501451002e-05,
+      "loss": 0.6416,
+      "step": 46265
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.77466201795802e-05,
+      "loss": 0.6318,
+      "step": 46270
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7745913761053812e-05,
+      "loss": 0.6504,
+      "step": 46275
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7745207245880654e-05,
+      "loss": 0.6376,
+      "step": 46280
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7744500634069544e-05,
+      "loss": 0.6435,
+      "step": 46285
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7743793925629296e-05,
+      "loss": 0.6381,
+      "step": 46290
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7743087120568727e-05,
+      "loss": 0.6318,
+      "step": 46295
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7742380218896658e-05,
+      "loss": 0.6455,
+      "step": 46300
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7741673220621907e-05,
+      "loss": 0.6334,
+      "step": 46305
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.77409661257533e-05,
+      "loss": 0.649,
+      "step": 46310
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7740258934299654e-05,
+      "loss": 0.6313,
+      "step": 46315
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7739551646269796e-05,
+      "loss": 0.6525,
+      "step": 46320
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7738844261672547e-05,
+      "loss": 0.6331,
+      "step": 46325
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7738136780516738e-05,
+      "loss": 0.6501,
+      "step": 46330
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7737429202811194e-05,
+      "loss": 0.647,
+      "step": 46335
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7736721528564747e-05,
+      "loss": 0.6412,
+      "step": 46340
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.773601375778622e-05,
+      "loss": 0.6429,
+      "step": 46345
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.773530589048445e-05,
+      "loss": 0.6437,
+      "step": 46350
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7734597926668264e-05,
+      "loss": 0.6601,
+      "step": 46355
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7733889866346497e-05,
+      "loss": 0.6404,
+      "step": 46360
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7733181709527986e-05,
+      "loss": 0.6493,
+      "step": 46365
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7732473456221567e-05,
+      "loss": 0.6451,
+      "step": 46370
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7731765106436073e-05,
+      "loss": 0.643,
+      "step": 46375
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.773105666018035e-05,
+      "loss": 0.6355,
+      "step": 46380
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7730348117463222e-05,
+      "loss": 0.6476,
+      "step": 46385
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.772963947829355e-05,
+      "loss": 0.6502,
+      "step": 46390
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7728930742680156e-05,
+      "loss": 0.6404,
+      "step": 46395
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.77282219106319e-05,
+      "loss": 0.6263,
+      "step": 46400
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.772751298215761e-05,
+      "loss": 0.6372,
+      "step": 46405
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7726803957266146e-05,
+      "loss": 0.6344,
+      "step": 46410
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.772609483596635e-05,
+      "loss": 0.6516,
+      "step": 46415
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7725385618267065e-05,
+      "loss": 0.6443,
+      "step": 46420
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7724676304177145e-05,
+      "loss": 0.6449,
+      "step": 46425
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7723966893705437e-05,
+      "loss": 0.6396,
+      "step": 46430
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.772325738686079e-05,
+      "loss": 0.6398,
+      "step": 46435
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7722547783652064e-05,
+      "loss": 0.6482,
+      "step": 46440
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7721838084088113e-05,
+      "loss": 0.6441,
+      "step": 46445
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7721128288177782e-05,
+      "loss": 0.6411,
+      "step": 46450
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.772041839592994e-05,
+      "loss": 0.6543,
+      "step": 46455
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.771970840735343e-05,
+      "loss": 0.6445,
+      "step": 46460
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.771899832245712e-05,
+      "loss": 0.632,
+      "step": 46465
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7718288141249873e-05,
+      "loss": 0.6477,
+      "step": 46470
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.771757786374054e-05,
+      "loss": 0.6378,
+      "step": 46475
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7716867489937995e-05,
+      "loss": 0.6329,
+      "step": 46480
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.771615701985109e-05,
+      "loss": 0.6369,
+      "step": 46485
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7715446453488696e-05,
+      "loss": 0.6305,
+      "step": 46490
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7714735790859678e-05,
+      "loss": 0.6465,
+      "step": 46495
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7714025031972904e-05,
+      "loss": 0.6351,
+      "step": 46500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7713314176837237e-05,
+      "loss": 0.6329,
+      "step": 46505
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.771260322546155e-05,
+      "loss": 0.64,
+      "step": 46510
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7711892177854717e-05,
+      "loss": 0.6454,
+      "step": 46515
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7711181034025607e-05,
+      "loss": 0.6342,
+      "step": 46520
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.771046979398309e-05,
+      "loss": 0.6338,
+      "step": 46525
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7709758457736042e-05,
+      "loss": 0.6362,
+      "step": 46530
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7709047025293344e-05,
+      "loss": 0.6375,
+      "step": 46535
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7708335496663864e-05,
+      "loss": 0.6377,
+      "step": 46540
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7707623871856484e-05,
+      "loss": 0.6341,
+      "step": 46545
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7706912150880083e-05,
+      "loss": 0.6408,
+      "step": 46550
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.770620033374354e-05,
+      "loss": 0.6476,
+      "step": 46555
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.770548842045574e-05,
+      "loss": 0.6463,
+      "step": 46560
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.770477641102556e-05,
+      "loss": 0.641,
+      "step": 46565
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.770406430546189e-05,
+      "loss": 0.642,
+      "step": 46570
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.770335210377361e-05,
+      "loss": 0.6389,
+      "step": 46575
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7702639805969612e-05,
+      "loss": 0.6431,
+      "step": 46580
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.770192741205878e-05,
+      "loss": 0.6407,
+      "step": 46585
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7701214922049998e-05,
+      "loss": 0.6425,
+      "step": 46590
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7700502335952162e-05,
+      "loss": 0.6429,
+      "step": 46595
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7699789653774157e-05,
+      "loss": 0.6356,
+      "step": 46600
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.769907687552488e-05,
+      "loss": 0.6414,
+      "step": 46605
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.769836400121323e-05,
+      "loss": 0.6329,
+      "step": 46610
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7697651030848094e-05,
+      "loss": 0.6289,
+      "step": 46615
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7696937964438366e-05,
+      "loss": 0.6487,
+      "step": 46620
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7696224801992947e-05,
+      "loss": 0.6345,
+      "step": 46625
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7695511543520734e-05,
+      "loss": 0.63,
+      "step": 46630
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.769479818903063e-05,
+      "loss": 0.6263,
+      "step": 46635
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7694084738531527e-05,
+      "loss": 0.6357,
+      "step": 46640
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7693371192032337e-05,
+      "loss": 0.6389,
+      "step": 46645
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7692657549541955e-05,
+      "loss": 0.6316,
+      "step": 46650
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.769194381106929e-05,
+      "loss": 0.6503,
+      "step": 46655
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.769122997662325e-05,
+      "loss": 0.6301,
+      "step": 46660
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7690516046212735e-05,
+      "loss": 0.6453,
+      "step": 46665
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7689802019846653e-05,
+      "loss": 0.6224,
+      "step": 46670
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7689087897533916e-05,
+      "loss": 0.6464,
+      "step": 46675
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7688373679283437e-05,
+      "loss": 0.648,
+      "step": 46680
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7687659365104117e-05,
+      "loss": 0.6389,
+      "step": 46685
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.768694495500488e-05,
+      "loss": 0.6474,
+      "step": 46690
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7686230448994637e-05,
+      "loss": 0.6282,
+      "step": 46695
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.76855158470823e-05,
+      "loss": 0.6415,
+      "step": 46700
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7684801149276784e-05,
+      "loss": 0.6399,
+      "step": 46705
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.768408635558701e-05,
+      "loss": 0.639,
+      "step": 46710
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7683371466021896e-05,
+      "loss": 0.6358,
+      "step": 46715
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7682656480590362e-05,
+      "loss": 0.6503,
+      "step": 46720
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.768194139930133e-05,
+      "loss": 0.6535,
+      "step": 46725
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7681226222163715e-05,
+      "loss": 0.6489,
+      "step": 46730
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7680510949186447e-05,
+      "loss": 0.6293,
+      "step": 46735
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7679795580378454e-05,
+      "loss": 0.6367,
+      "step": 46740
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7679080115748655e-05,
+      "loss": 0.6391,
+      "step": 46745
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.767836455530598e-05,
+      "loss": 0.6367,
+      "step": 46750
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.767764889905935e-05,
+      "loss": 0.6505,
+      "step": 46755
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7676933147017707e-05,
+      "loss": 0.6434,
+      "step": 46760
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7676217299189972e-05,
+      "loss": 0.6451,
+      "step": 46765
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7675501355585075e-05,
+      "loss": 0.6372,
+      "step": 46770
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7674785316211963e-05,
+      "loss": 0.6376,
+      "step": 46775
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7674069181079553e-05,
+      "loss": 0.6517,
+      "step": 46780
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.767335295019679e-05,
+      "loss": 0.6545,
+      "step": 46785
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.767263662357261e-05,
+      "loss": 0.6418,
+      "step": 46790
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.767192020121595e-05,
+      "loss": 0.6361,
+      "step": 46795
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7671203683135747e-05,
+      "loss": 0.6343,
+      "step": 46800
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7670487069340944e-05,
+      "loss": 0.6325,
+      "step": 46805
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7669770359840476e-05,
+      "loss": 0.6355,
+      "step": 46810
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7669053554643298e-05,
+      "loss": 0.6309,
+      "step": 46815
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7668336653758338e-05,
+      "loss": 0.6292,
+      "step": 46820
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.766761965719455e-05,
+      "loss": 0.6424,
+      "step": 46825
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.766690256496088e-05,
+      "loss": 0.6503,
+      "step": 46830
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.766618537706627e-05,
+      "loss": 0.645,
+      "step": 46835
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.766546809351968e-05,
+      "loss": 0.6473,
+      "step": 46840
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7664750714330048e-05,
+      "loss": 0.6496,
+      "step": 46845
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7664033239506327e-05,
+      "loss": 0.64,
+      "step": 46850
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.766331566905747e-05,
+      "loss": 0.6366,
+      "step": 46855
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7662598002992436e-05,
+      "loss": 0.6442,
+      "step": 46860
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.766188024132017e-05,
+      "loss": 0.6417,
+      "step": 46865
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7661162384049633e-05,
+      "loss": 0.6487,
+      "step": 46870
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.766044443118978e-05,
+      "loss": 0.635,
+      "step": 46875
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.765972638274957e-05,
+      "loss": 0.6478,
+      "step": 46880
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7659008238737968e-05,
+      "loss": 0.6435,
+      "step": 46885
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7658289999163924e-05,
+      "loss": 0.6458,
+      "step": 46890
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.76575716640364e-05,
+      "loss": 0.6474,
+      "step": 46895
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.765685323336437e-05,
+      "loss": 0.6411,
+      "step": 46900
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.765613470715678e-05,
+      "loss": 0.6433,
+      "step": 46905
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7655416085422613e-05,
+      "loss": 0.6414,
+      "step": 46910
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.765469736817083e-05,
+      "loss": 0.6382,
+      "step": 46915
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7653978555410394e-05,
+      "loss": 0.6424,
+      "step": 46920
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7653259647150275e-05,
+      "loss": 0.6454,
+      "step": 46925
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7652540643399447e-05,
+      "loss": 0.6456,
+      "step": 46930
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7651821544166878e-05,
+      "loss": 0.6354,
+      "step": 46935
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.765110234946154e-05,
+      "loss": 0.6396,
+      "step": 46940
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7650383059292408e-05,
+      "loss": 0.635,
+      "step": 46945
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7649663673668454e-05,
+      "loss": 0.637,
+      "step": 46950
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7648944192598657e-05,
+      "loss": 0.6429,
+      "step": 46955
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.764822461609199e-05,
+      "loss": 0.6303,
+      "step": 46960
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7647504944157436e-05,
+      "loss": 0.6515,
+      "step": 46965
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7646785176803973e-05,
+      "loss": 0.6453,
+      "step": 46970
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.764606531404058e-05,
+      "loss": 0.6325,
+      "step": 46975
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7645345355876243e-05,
+      "loss": 0.6261,
+      "step": 46980
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.764462530231994e-05,
+      "loss": 0.638,
+      "step": 46985
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7643905153380658e-05,
+      "loss": 0.6352,
+      "step": 46990
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7643184909067377e-05,
+      "loss": 0.6445,
+      "step": 46995
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.764246456938909e-05,
+      "loss": 0.6414,
+      "step": 47000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7641744134354785e-05,
+      "loss": 0.6452,
+      "step": 47005
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7641023603973447e-05,
+      "loss": 0.6408,
+      "step": 47010
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.764030297825407e-05,
+      "loss": 0.6428,
+      "step": 47015
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7639582257205644e-05,
+      "loss": 0.6427,
+      "step": 47020
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7638861440837157e-05,
+      "loss": 0.6479,
+      "step": 47025
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.763814052915761e-05,
+      "loss": 0.6354,
+      "step": 47030
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.763741952217599e-05,
+      "loss": 0.649,
+      "step": 47035
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7636698419901304e-05,
+      "loss": 0.631,
+      "step": 47040
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.763597722234254e-05,
+      "loss": 0.6434,
+      "step": 47045
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.76352559295087e-05,
+      "loss": 0.6438,
+      "step": 47050
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.763453454140878e-05,
+      "loss": 0.6294,
+      "step": 47055
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7633813058051785e-05,
+      "loss": 0.6308,
+      "step": 47060
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7633091479446717e-05,
+      "loss": 0.6403,
+      "step": 47065
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.763236980560258e-05,
+      "loss": 0.6404,
+      "step": 47070
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7631648036528374e-05,
+      "loss": 0.6402,
+      "step": 47075
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7630926172233107e-05,
+      "loss": 0.6368,
+      "step": 47080
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7630204212725783e-05,
+      "loss": 0.6431,
+      "step": 47085
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7629482158015418e-05,
+      "loss": 0.6404,
+      "step": 47090
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7628760008111015e-05,
+      "loss": 0.641,
+      "step": 47095
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7628037763021586e-05,
+      "loss": 0.6405,
+      "step": 47100
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.762731542275614e-05,
+      "loss": 0.6419,
+      "step": 47105
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.762659298732369e-05,
+      "loss": 0.6465,
+      "step": 47110
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7625870456733255e-05,
+      "loss": 0.6294,
+      "step": 47115
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7625147830993847e-05,
+      "loss": 0.6382,
+      "step": 47120
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.762442511011448e-05,
+      "loss": 0.647,
+      "step": 47125
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7623702294104175e-05,
+      "loss": 0.6384,
+      "step": 47130
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.762297938297195e-05,
+      "loss": 0.6419,
+      "step": 47135
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.762225637672682e-05,
+      "loss": 0.6399,
+      "step": 47140
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7621533275377815e-05,
+      "loss": 0.64,
+      "step": 47145
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7620810078933946e-05,
+      "loss": 0.6411,
+      "step": 47150
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7620086787404253e-05,
+      "loss": 0.6461,
+      "step": 47155
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7619363400797744e-05,
+      "loss": 0.6418,
+      "step": 47160
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7618639919123448e-05,
+      "loss": 0.6465,
+      "step": 47165
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.76179163423904e-05,
+      "loss": 0.6447,
+      "step": 47170
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.761719267060762e-05,
+      "loss": 0.6531,
+      "step": 47175
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.761646890378414e-05,
+      "loss": 0.6517,
+      "step": 47180
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7615745041928996e-05,
+      "loss": 0.643,
+      "step": 47185
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7615021085051213e-05,
+      "loss": 0.6313,
+      "step": 47190
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7614297033159826e-05,
+      "loss": 0.6374,
+      "step": 47195
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7613572886263865e-05,
+      "loss": 0.6446,
+      "step": 47200
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7612848644372373e-05,
+      "loss": 0.64,
+      "step": 47205
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.761212430749438e-05,
+      "loss": 0.6433,
+      "step": 47210
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.761139987563893e-05,
+      "loss": 0.6428,
+      "step": 47215
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7610675348815055e-05,
+      "loss": 0.6475,
+      "step": 47220
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.76099507270318e-05,
+      "loss": 0.6359,
+      "step": 47225
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.76092260102982e-05,
+      "loss": 0.6436,
+      "step": 47230
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.760850119862331e-05,
+      "loss": 0.6267,
+      "step": 47235
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.760777629201616e-05,
+      "loss": 0.6354,
+      "step": 47240
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7607051290485798e-05,
+      "loss": 0.6528,
+      "step": 47245
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7606326194041274e-05,
+      "loss": 0.646,
+      "step": 47250
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7605601002691633e-05,
+      "loss": 0.6394,
+      "step": 47255
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7604875716445925e-05,
+      "loss": 0.6438,
+      "step": 47260
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7604150335313195e-05,
+      "loss": 0.6266,
+      "step": 47265
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7603424859302498e-05,
+      "loss": 0.6447,
+      "step": 47270
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7602699288422885e-05,
+      "loss": 0.643,
+      "step": 47275
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7601973622683404e-05,
+      "loss": 0.6426,
+      "step": 47280
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.760124786209312e-05,
+      "loss": 0.6392,
+      "step": 47285
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.760052200666108e-05,
+      "loss": 0.6363,
+      "step": 47290
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.759979605639634e-05,
+      "loss": 0.6425,
+      "step": 47295
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7599070011307965e-05,
+      "loss": 0.6379,
+      "step": 47300
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7598343871405004e-05,
+      "loss": 0.6439,
+      "step": 47305
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7597617636696525e-05,
+      "loss": 0.6333,
+      "step": 47310
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.759689130719159e-05,
+      "loss": 0.6405,
+      "step": 47315
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7596164882899254e-05,
+      "loss": 0.637,
+      "step": 47320
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7595438363828584e-05,
+      "loss": 0.636,
+      "step": 47325
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.759471174998865e-05,
+      "loss": 0.6462,
+      "step": 47330
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7593985041388514e-05,
+      "loss": 0.6361,
+      "step": 47335
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.7593258238037242e-05,
+      "loss": 0.6436,
+      "step": 47340
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.75925313399439e-05,
+      "loss": 0.6403,
+      "step": 47345
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.759180434711757e-05,
+      "loss": 0.6409,
+      "step": 47350
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7591077259567306e-05,
+      "loss": 0.6335,
+      "step": 47355
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.759035007730219e-05,
+      "loss": 0.6402,
+      "step": 47360
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7589622800331293e-05,
+      "loss": 0.6396,
+      "step": 47365
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.758889542866369e-05,
+      "loss": 0.6414,
+      "step": 47370
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7588167962308458e-05,
+      "loss": 0.6362,
+      "step": 47375
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.758744040127467e-05,
+      "loss": 0.6399,
+      "step": 47380
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7586712745571402e-05,
+      "loss": 0.6339,
+      "step": 47385
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.758598499520774e-05,
+      "loss": 0.645,
+      "step": 47390
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.758525715019276e-05,
+      "loss": 0.6414,
+      "step": 47395
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7584529210535545e-05,
+      "loss": 0.6391,
+      "step": 47400
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7583801176245176e-05,
+      "loss": 0.6284,
+      "step": 47405
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7583073047330738e-05,
+      "loss": 0.6293,
+      "step": 47410
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7582344823801317e-05,
+      "loss": 0.6415,
+      "step": 47415
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7581616505665996e-05,
+      "loss": 0.6448,
+      "step": 47420
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7580888092933867e-05,
+      "loss": 0.6408,
+      "step": 47425
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7580159585614015e-05,
+      "loss": 0.6423,
+      "step": 47430
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7579430983715525e-05,
+      "loss": 0.6498,
+      "step": 47435
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7578702287247494e-05,
+      "loss": 0.6301,
+      "step": 47440
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.757797349621902e-05,
+      "loss": 0.6447,
+      "step": 47445
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7577244610639183e-05,
+      "loss": 0.6441,
+      "step": 47450
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7576515630517087e-05,
+      "loss": 0.6372,
+      "step": 47455
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.757578655586182e-05,
+      "loss": 0.6608,
+      "step": 47460
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7575057386682487e-05,
+      "loss": 0.6465,
+      "step": 47465
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7574328122988183e-05,
+      "loss": 0.6449,
+      "step": 47470
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7573598764788e-05,
+      "loss": 0.6414,
+      "step": 47475
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7572869312091052e-05,
+      "loss": 0.641,
+      "step": 47480
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.757213976490643e-05,
+      "loss": 0.6423,
+      "step": 47485
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7571410123243236e-05,
+      "loss": 0.649,
+      "step": 47490
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.757068038711058e-05,
+      "loss": 0.6338,
+      "step": 47495
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7569950556517566e-05,
+      "loss": 0.6373,
+      "step": 47500
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7569220631473297e-05,
+      "loss": 0.6465,
+      "step": 47505
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7568490611986884e-05,
+      "loss": 0.6449,
+      "step": 47510
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.756776049806743e-05,
+      "loss": 0.6358,
+      "step": 47515
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.756703028972405e-05,
+      "loss": 0.6384,
+      "step": 47520
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7566299986965855e-05,
+      "loss": 0.6479,
+      "step": 47525
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7565569589801952e-05,
+      "loss": 0.6416,
+      "step": 47530
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7564839098241458e-05,
+      "loss": 0.6297,
+      "step": 47535
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7564108512293487e-05,
+      "loss": 0.636,
+      "step": 47540
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7563377831967157e-05,
+      "loss": 0.6231,
+      "step": 47545
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.756264705727158e-05,
+      "loss": 0.6417,
+      "step": 47550
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7561916188215877e-05,
+      "loss": 0.6494,
+      "step": 47555
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7561185224809164e-05,
+      "loss": 0.6361,
+      "step": 47560
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.756045416706057e-05,
+      "loss": 0.6361,
+      "step": 47565
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7559723014979205e-05,
+      "loss": 0.6474,
+      "step": 47570
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7558991768574197e-05,
+      "loss": 0.6353,
+      "step": 47575
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7558260427854673e-05,
+      "loss": 0.643,
+      "step": 47580
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.755752899282975e-05,
+      "loss": 0.6402,
+      "step": 47585
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7556797463508565e-05,
+      "loss": 0.6482,
+      "step": 47590
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7556065839900232e-05,
+      "loss": 0.6282,
+      "step": 47595
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7555334122013894e-05,
+      "loss": 0.6511,
+      "step": 47600
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7554602309858674e-05,
+      "loss": 0.6411,
+      "step": 47605
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7553870403443696e-05,
+      "loss": 0.6463,
+      "step": 47610
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7553138402778102e-05,
+      "loss": 0.64,
+      "step": 47615
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7552406307871023e-05,
+      "loss": 0.6329,
+      "step": 47620
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7551674118731592e-05,
+      "loss": 0.6362,
+      "step": 47625
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7550941835368945e-05,
+      "loss": 0.6346,
+      "step": 47630
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.755020945779222e-05,
+      "loss": 0.6466,
+      "step": 47635
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7549476986010552e-05,
+      "loss": 0.6412,
+      "step": 47640
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7548744420033085e-05,
+      "loss": 0.6342,
+      "step": 47645
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.754801175986895e-05,
+      "loss": 0.6336,
+      "step": 47650
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.7547279005527303e-05,
+      "loss": 0.6369,
+      "step": 47655
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7546546157017274e-05,
+      "loss": 0.6216,
+      "step": 47660
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7545813214348012e-05,
+      "loss": 0.6504,
+      "step": 47665
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.754508017752866e-05,
+      "loss": 0.6328,
+      "step": 47670
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7544347046568363e-05,
+      "loss": 0.6455,
+      "step": 47675
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7543613821476277e-05,
+      "loss": 0.6384,
+      "step": 47680
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7542880502261542e-05,
+      "loss": 0.6401,
+      "step": 47685
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7542147088933307e-05,
+      "loss": 0.6328,
+      "step": 47690
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.754141358150073e-05,
+      "loss": 0.6446,
+      "step": 47695
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7540679979972958e-05,
+      "loss": 0.6389,
+      "step": 47700
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7539946284359146e-05,
+      "loss": 0.6301,
+      "step": 47705
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7539212494668447e-05,
+      "loss": 0.6381,
+      "step": 47710
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7538478610910016e-05,
+      "loss": 0.6372,
+      "step": 47715
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7537744633093015e-05,
+      "loss": 0.6446,
+      "step": 47720
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7537010561226595e-05,
+      "loss": 0.6295,
+      "step": 47725
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.753627639531992e-05,
+      "loss": 0.6468,
+      "step": 47730
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7535542135382147e-05,
+      "loss": 0.6398,
+      "step": 47735
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.753480778142244e-05,
+      "loss": 0.6383,
+      "step": 47740
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7534073333449958e-05,
+      "loss": 0.628,
+      "step": 47745
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7533338791473872e-05,
+      "loss": 0.6262,
+      "step": 47750
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7532604155503336e-05,
+      "loss": 0.6446,
+      "step": 47755
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.753186942554753e-05,
+      "loss": 0.6362,
+      "step": 47760
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.753113460161561e-05,
+      "loss": 0.6392,
+      "step": 47765
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7530399683716746e-05,
+      "loss": 0.6405,
+      "step": 47770
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.752966467186011e-05,
+      "loss": 0.6291,
+      "step": 47775
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7528929566054877e-05,
+      "loss": 0.6312,
+      "step": 47780
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7528194366310213e-05,
+      "loss": 0.6398,
+      "step": 47785
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.752745907263529e-05,
+      "loss": 0.6313,
+      "step": 47790
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7526723685039288e-05,
+      "loss": 0.6545,
+      "step": 47795
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.752598820353138e-05,
+      "loss": 0.6311,
+      "step": 47800
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7525252628120743e-05,
+      "loss": 0.6299,
+      "step": 47805
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7524516958816555e-05,
+      "loss": 0.6342,
+      "step": 47810
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.752378119562799e-05,
+      "loss": 0.6451,
+      "step": 47815
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7523045338564237e-05,
+      "loss": 0.6328,
+      "step": 47820
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7522309387634472e-05,
+      "loss": 0.6409,
+      "step": 47825
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.752157334284788e-05,
+      "loss": 0.6479,
+      "step": 47830
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.752083720421364e-05,
+      "loss": 0.6365,
+      "step": 47835
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7520100971740942e-05,
+      "loss": 0.6376,
+      "step": 47840
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7519364645438972e-05,
+      "loss": 0.6494,
+      "step": 47845
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7518628225316915e-05,
+      "loss": 0.6491,
+      "step": 47850
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7517891711383957e-05,
+      "loss": 0.6334,
+      "step": 47855
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7517155103649293e-05,
+      "loss": 0.6371,
+      "step": 47860
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7516418402122112e-05,
+      "loss": 0.6411,
+      "step": 47865
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7515681606811602e-05,
+      "loss": 0.6454,
+      "step": 47870
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7514944717726962e-05,
+      "loss": 0.6293,
+      "step": 47875
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7514207734877384e-05,
+      "loss": 0.6296,
+      "step": 47880
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.751347065827206e-05,
+      "loss": 0.6476,
+      "step": 47885
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7512733487920193e-05,
+      "loss": 0.6356,
+      "step": 47890
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7511996223830974e-05,
+      "loss": 0.647,
+      "step": 47895
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.751125886601361e-05,
+      "loss": 0.6364,
+      "step": 47900
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7510521414477292e-05,
+      "loss": 0.6435,
+      "step": 47905
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7509783869231228e-05,
+      "loss": 0.6491,
+      "step": 47910
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7509046230284616e-05,
+      "loss": 0.6381,
+      "step": 47915
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7508308497646664e-05,
+      "loss": 0.6437,
+      "step": 47920
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7507570671326573e-05,
+      "loss": 0.6424,
+      "step": 47925
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7506832751333553e-05,
+      "loss": 0.6289,
+      "step": 47930
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7506094737676808e-05,
+      "loss": 0.6252,
+      "step": 47935
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7505356630365544e-05,
+      "loss": 0.6401,
+      "step": 47940
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.750461842940897e-05,
+      "loss": 0.6468,
+      "step": 47945
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.750388013481631e-05,
+      "loss": 0.6371,
+      "step": 47950
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7503141746596758e-05,
+      "loss": 0.6351,
+      "step": 47955
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.750240326475954e-05,
+      "loss": 0.6279,
+      "step": 47960
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.750166468931386e-05,
+      "loss": 0.6368,
+      "step": 47965
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.750092602026894e-05,
+      "loss": 0.6238,
+      "step": 47970
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.7500187257634e-05,
+      "loss": 0.6393,
+      "step": 47975
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.7499448401418243e-05,
+      "loss": 0.6476,
+      "step": 47980
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.7498709451630904e-05,
+      "loss": 0.6275,
+      "step": 47985
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.7497970408281192e-05,
+      "loss": 0.6402,
+      "step": 47990
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.7497231271378335e-05,
+      "loss": 0.6296,
+      "step": 47995
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.749649204093155e-05,
+      "loss": 0.6321,
+      "step": 48000
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 156250,
+  "num_train_epochs": 5,
+  "save_steps": 2000,
+  "total_flos": 3.310222499070345e+18,
+  "trial_name": null,
+  "trial_params": null
+}