diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,9632 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9999374491774566,
+  "eval_steps": 500,
+  "global_step": 7993,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5000000000000004e-07,
+      "loss": 1.8341,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.25e-06,
+      "loss": 1.9529,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5e-06,
+      "loss": 1.9198,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.75e-06,
+      "loss": 1.9007,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5e-06,
+      "loss": 1.8667,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.25e-06,
+      "loss": 1.8982,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.5e-06,
+      "loss": 1.9164,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.75e-06,
+      "loss": 1.9576,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1e-05,
+      "loss": 1.9407,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.125e-05,
+      "loss": 1.9003,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.25e-05,
+      "loss": 1.8671,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3750000000000002e-05,
+      "loss": 1.8734,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5e-05,
+      "loss": 1.8261,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6250000000000002e-05,
+      "loss": 1.7735,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.75e-05,
+      "loss": 1.7516,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8750000000000002e-05,
+      "loss": 1.766,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 1.6577,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.125e-05,
+      "loss": 1.6658,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.25e-05,
+      "loss": 1.6037,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.375e-05,
+      "loss": 1.6221,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5e-05,
+      "loss": 1.6102,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.625e-05,
+      "loss": 1.5017,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7500000000000004e-05,
+      "loss": 1.5187,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8749999999999997e-05,
+      "loss": 1.5232,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3e-05,
+      "loss": 1.4907,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.125e-05,
+      "loss": 1.5197,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2500000000000004e-05,
+      "loss": 1.4709,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.375000000000001e-05,
+      "loss": 1.4848,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5e-05,
+      "loss": 1.4829,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.625e-05,
+      "loss": 1.4053,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 1.4657,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.875e-05,
+      "loss": 1.3581,
+      "step": 155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4e-05,
+      "loss": 1.4169,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.125e-05,
+      "loss": 1.4224,
+      "step": 165
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.25e-05,
+      "loss": 1.3826,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.375e-05,
+      "loss": 1.4225,
+      "step": 175
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5e-05,
+      "loss": 1.3698,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6250000000000006e-05,
+      "loss": 1.4793,
+      "step": 185
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.75e-05,
+      "loss": 1.4069,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.875e-05,
+      "loss": 1.4344,
+      "step": 195
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5e-05,
+      "loss": 1.4146,
+      "step": 200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.125e-05,
+      "loss": 1.3875,
+      "step": 205
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.25e-05,
+      "loss": 1.4059,
+      "step": 210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.375e-05,
+      "loss": 1.376,
+      "step": 215
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.500000000000001e-05,
+      "loss": 1.4084,
+      "step": 220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.6250000000000005e-05,
+      "loss": 1.3235,
+      "step": 225
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.7499999999999995e-05,
+      "loss": 1.3423,
+      "step": 230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.8750000000000005e-05,
+      "loss": 1.3499,
+      "step": 235
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6e-05,
+      "loss": 1.364,
+      "step": 240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.125000000000001e-05,
+      "loss": 1.4129,
+      "step": 245
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.25e-05,
+      "loss": 1.3671,
+      "step": 250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.375e-05,
+      "loss": 1.3599,
+      "step": 255
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.500000000000001e-05,
+      "loss": 1.3409,
+      "step": 260
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.625e-05,
+      "loss": 1.3171,
+      "step": 265
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.750000000000001e-05,
+      "loss": 1.3322,
+      "step": 270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.875e-05,
+      "loss": 1.338,
+      "step": 275
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7e-05,
+      "loss": 1.3117,
+      "step": 280
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.125000000000001e-05,
+      "loss": 1.3601,
+      "step": 285
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.25e-05,
+      "loss": 1.2932,
+      "step": 290
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.375e-05,
+      "loss": 1.35,
+      "step": 295
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 1.3544,
+      "step": 300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.625e-05,
+      "loss": 1.3876,
+      "step": 305
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.75e-05,
+      "loss": 1.3225,
+      "step": 310
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.875e-05,
+      "loss": 1.32,
+      "step": 315
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8e-05,
+      "loss": 1.3326,
+      "step": 320
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.125000000000001e-05,
+      "loss": 1.3335,
+      "step": 325
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.25e-05,
+      "loss": 1.2723,
+      "step": 330
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.375e-05,
+      "loss": 1.3459,
+      "step": 335
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.5e-05,
+      "loss": 1.3342,
+      "step": 340
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.625000000000001e-05,
+      "loss": 1.3082,
+      "step": 345
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.75e-05,
+      "loss": 1.2794,
+      "step": 350
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.875e-05,
+      "loss": 1.2994,
+      "step": 355
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9e-05,
+      "loss": 1.2825,
+      "step": 360
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.125e-05,
+      "loss": 1.2668,
+      "step": 365
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.250000000000001e-05,
+      "loss": 1.3181,
+      "step": 370
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.375e-05,
+      "loss": 1.2584,
+      "step": 375
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.5e-05,
+      "loss": 1.2607,
+      "step": 380
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.625000000000001e-05,
+      "loss": 1.322,
+      "step": 385
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.75e-05,
+      "loss": 1.2928,
+      "step": 390
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.875000000000002e-05,
+      "loss": 1.3445,
+      "step": 395
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001,
+      "loss": 1.2258,
+      "step": 400
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010125,
+      "loss": 1.2958,
+      "step": 405
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001025,
+      "loss": 1.2787,
+      "step": 410
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010375000000000001,
+      "loss": 1.3215,
+      "step": 415
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000105,
+      "loss": 1.3149,
+      "step": 420
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010625000000000001,
+      "loss": 1.2583,
+      "step": 425
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001075,
+      "loss": 1.3136,
+      "step": 430
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010875,
+      "loss": 1.2958,
+      "step": 435
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011000000000000002,
+      "loss": 1.3442,
+      "step": 440
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011125000000000001,
+      "loss": 1.3587,
+      "step": 445
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011250000000000001,
+      "loss": 1.3061,
+      "step": 450
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011375,
+      "loss": 1.3344,
+      "step": 455
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011499999999999999,
+      "loss": 1.2874,
+      "step": 460
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011625000000000002,
+      "loss": 1.2575,
+      "step": 465
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011750000000000001,
+      "loss": 1.2969,
+      "step": 470
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011875,
+      "loss": 1.2663,
+      "step": 475
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00012,
+      "loss": 1.2378,
+      "step": 480
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00012124999999999999,
+      "loss": 1.2858,
+      "step": 485
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00012250000000000002,
+      "loss": 1.2478,
+      "step": 490
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00012375,
+      "loss": 1.2748,
+      "step": 495
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000125,
+      "loss": 1.303,
+      "step": 500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00012625,
+      "loss": 1.2927,
+      "step": 505
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001275,
+      "loss": 1.2881,
+      "step": 510
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00012875,
+      "loss": 1.239,
+      "step": 515
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00013000000000000002,
+      "loss": 1.3001,
+      "step": 520
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00013125000000000002,
+      "loss": 1.2719,
+      "step": 525
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001325,
+      "loss": 1.2861,
+      "step": 530
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00013375,
+      "loss": 1.2435,
+      "step": 535
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00013500000000000003,
+      "loss": 1.2344,
+      "step": 540
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00013625,
+      "loss": 1.2604,
+      "step": 545
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001375,
+      "loss": 1.2848,
+      "step": 550
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00013875,
+      "loss": 1.2622,
+      "step": 555
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014,
+      "loss": 1.229,
+      "step": 560
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014125000000000002,
+      "loss": 1.3119,
+      "step": 565
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014250000000000002,
+      "loss": 1.3302,
+      "step": 570
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014375,
+      "loss": 1.2579,
+      "step": 575
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000145,
+      "loss": 1.2405,
+      "step": 580
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014625,
+      "loss": 1.3071,
+      "step": 585
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001475,
+      "loss": 1.2814,
+      "step": 590
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014875,
+      "loss": 1.2853,
+      "step": 595
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015000000000000001,
+      "loss": 1.2761,
+      "step": 600
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015125,
+      "loss": 1.2833,
+      "step": 605
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001525,
+      "loss": 1.2161,
+      "step": 610
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015375000000000002,
+      "loss": 1.2494,
+      "step": 615
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000155,
+      "loss": 1.2007,
+      "step": 620
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015625,
+      "loss": 1.2849,
+      "step": 625
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001575,
+      "loss": 1.245,
+      "step": 630
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015875,
+      "loss": 1.2544,
+      "step": 635
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016,
+      "loss": 1.2569,
+      "step": 640
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016125000000000002,
+      "loss": 1.2728,
+      "step": 645
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016250000000000002,
+      "loss": 1.2815,
+      "step": 650
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016375,
+      "loss": 1.3389,
+      "step": 655
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000165,
+      "loss": 1.2399,
+      "step": 660
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016625000000000003,
+      "loss": 1.2521,
+      "step": 665
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001675,
+      "loss": 1.2654,
+      "step": 670
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016875,
+      "loss": 1.2626,
+      "step": 675
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00017,
+      "loss": 1.2687,
+      "step": 680
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00017125,
+      "loss": 1.2914,
+      "step": 685
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00017250000000000002,
+      "loss": 1.2894,
+      "step": 690
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00017375000000000002,
+      "loss": 1.2913,
+      "step": 695
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000175,
+      "loss": 1.2619,
+      "step": 700
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00017625,
+      "loss": 1.2241,
+      "step": 705
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001775,
+      "loss": 1.235,
+      "step": 710
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00017875,
+      "loss": 1.2523,
+      "step": 715
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00018,
+      "loss": 1.3353,
+      "step": 720
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00018125000000000001,
+      "loss": 1.1952,
+      "step": 725
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001825,
+      "loss": 1.2608,
+      "step": 730
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00018375,
+      "loss": 1.2797,
+      "step": 735
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00018500000000000002,
+      "loss": 1.2814,
+      "step": 740
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00018625,
+      "loss": 1.1934,
+      "step": 745
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001875,
+      "loss": 1.2691,
+      "step": 750
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00018875,
+      "loss": 1.2728,
+      "step": 755
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019,
+      "loss": 1.2872,
+      "step": 760
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019125000000000001,
+      "loss": 1.3103,
+      "step": 765
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019250000000000002,
+      "loss": 1.2455,
+      "step": 770
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019375000000000002,
+      "loss": 1.2633,
+      "step": 775
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000195,
+      "loss": 1.3254,
+      "step": 780
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019625,
+      "loss": 1.2191,
+      "step": 785
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019750000000000003,
+      "loss": 1.2741,
+      "step": 790
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019875,
+      "loss": 1.2912,
+      "step": 795
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 1.2785,
+      "step": 800
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019999976155431637,
+      "loss": 1.2812,
+      "step": 805
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001999990462184026,
+      "loss": 1.2162,
+      "step": 810
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019999785399567007,
+      "loss": 1.3171,
+      "step": 815
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019999618489180437,
+      "loss": 1.2429,
+      "step": 820
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019999403891476531,
+      "loss": 1.2484,
+      "step": 825
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019999141607478688,
+      "loss": 1.2734,
+      "step": 830
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001999883163843772,
+      "loss": 1.2261,
+      "step": 835
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019998473985831844,
+      "loss": 1.2412,
+      "step": 840
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019998068651366663,
+      "loss": 1.2967,
+      "step": 845
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019997615636975198,
+      "loss": 1.2659,
+      "step": 850
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019997114944817824,
+      "loss": 1.2585,
+      "step": 855
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000199965665772823,
+      "loss": 1.3256,
+      "step": 860
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019995970536983748,
+      "loss": 1.2622,
+      "step": 865
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001999532682676463,
+      "loss": 1.2266,
+      "step": 870
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019994635449694745,
+      "loss": 1.245,
+      "step": 875
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001999389640907121,
+      "loss": 1.2527,
+      "step": 880
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001999310970841845,
+      "loss": 1.2426,
+      "step": 885
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019992275351488165,
+      "loss": 1.286,
+      "step": 890
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019991393342259336,
+      "loss": 1.286,
+      "step": 895
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001999046368493819,
+      "loss": 1.2932,
+      "step": 900
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019989486383958178,
+      "loss": 1.2836,
+      "step": 905
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019988461443979966,
+      "loss": 1.2474,
+      "step": 910
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001998738886989141,
+      "loss": 1.2401,
+      "step": 915
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001998626866680751,
+      "loss": 1.2044,
+      "step": 920
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001998510084007043,
+      "loss": 1.2349,
+      "step": 925
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019983885395249434,
+      "loss": 1.2353,
+      "step": 930
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019982622338140865,
+      "loss": 1.3026,
+      "step": 935
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001998131167476814,
+      "loss": 1.2293,
+      "step": 940
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019979953411381697,
+      "loss": 1.2398,
+      "step": 945
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001997854755445898,
+      "loss": 1.2344,
+      "step": 950
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019977094110704396,
+      "loss": 1.2629,
+      "step": 955
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001997559308704929,
+      "loss": 1.2803,
+      "step": 960
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019974044490651924,
+      "loss": 1.2488,
+      "step": 965
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001997244832889741,
+      "loss": 1.27,
+      "step": 970
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019970804609397713,
+      "loss": 1.258,
+      "step": 975
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019969113339991587,
+      "loss": 1.2659,
+      "step": 980
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019967374528744548,
+      "loss": 1.2433,
+      "step": 985
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001996558818394884,
+      "loss": 1.2885,
+      "step": 990
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019963754314123385,
+      "loss": 1.2535,
+      "step": 995
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019961872928013749,
+      "loss": 1.2155,
+      "step": 1000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000199599440345921,
+      "loss": 1.2144,
+      "step": 1005
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019957967643057165,
+      "loss": 1.2125,
+      "step": 1010
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019955943762834188,
+      "loss": 1.3187,
+      "step": 1015
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019953872403574873,
+      "loss": 1.2293,
+      "step": 1020
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001995175357515736,
+      "loss": 1.2467,
+      "step": 1025
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001994958728768615,
+      "loss": 1.241,
+      "step": 1030
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001994737355149209,
+      "loss": 1.2383,
+      "step": 1035
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019945112377132293,
+      "loss": 1.218,
+      "step": 1040
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019942803775390103,
+      "loss": 1.2848,
+      "step": 1045
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019940447757275045,
+      "loss": 1.2456,
+      "step": 1050
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019938044334022763,
+      "loss": 1.2072,
+      "step": 1055
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019935593517094978,
+      "loss": 1.2629,
+      "step": 1060
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001993309531817942,
+      "loss": 1.3064,
+      "step": 1065
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019930549749189793,
+      "loss": 1.2794,
+      "step": 1070
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019927956822265688,
+      "loss": 1.26,
+      "step": 1075
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001992531654977255,
+      "loss": 1.2992,
+      "step": 1080
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019922628944301612,
+      "loss": 1.2094,
+      "step": 1085
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019919894018669834,
+      "loss": 1.2978,
+      "step": 1090
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001991711178591984,
+      "loss": 1.2521,
+      "step": 1095
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019914282259319853,
+      "loss": 1.2387,
+      "step": 1100
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019911405452363644,
+      "loss": 1.2547,
+      "step": 1105
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001990848137877046,
+      "loss": 1.2979,
+      "step": 1110
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019905510052484955,
+      "loss": 1.2881,
+      "step": 1115
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019902491487677125,
+      "loss": 1.286,
+      "step": 1120
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019899425698742244,
+      "loss": 1.2016,
+      "step": 1125
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019896312700300796,
+      "loss": 1.1912,
+      "step": 1130
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019893152507198406,
+      "loss": 1.219,
+      "step": 1135
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019889945134505754,
+      "loss": 1.2361,
+      "step": 1140
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019886690597518532,
+      "loss": 1.2541,
+      "step": 1145
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001988338891175734,
+      "loss": 1.2967,
+      "step": 1150
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019880040092967635,
+      "loss": 1.2897,
+      "step": 1155
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019876644157119642,
+      "loss": 1.2816,
+      "step": 1160
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001987320112040829,
+      "loss": 1.2125,
+      "step": 1165
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001986971099925312,
+      "loss": 1.2237,
+      "step": 1170
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019866173810298223,
+      "loss": 1.1892,
+      "step": 1175
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019862589570412143,
+      "loss": 1.2418,
+      "step": 1180
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019858958296687816,
+      "loss": 1.2378,
+      "step": 1185
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019855280006442466,
+      "loss": 1.2375,
+      "step": 1190
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019851554717217545,
+      "loss": 1.2362,
+      "step": 1195
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019847782446778638,
+      "loss": 1.2871,
+      "step": 1200
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019843963213115373,
+      "loss": 1.2121,
+      "step": 1205
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019840097034441346,
+      "loss": 1.1942,
+      "step": 1210
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019836183929194034,
+      "loss": 1.2597,
+      "step": 1215
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019832223916034692,
+      "loss": 1.2626,
+      "step": 1220
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001982821701384829,
+      "loss": 1.2218,
+      "step": 1225
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019824163241743387,
+      "loss": 1.2141,
+      "step": 1230
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001982006261905208,
+      "loss": 1.2099,
+      "step": 1235
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019815915165329883,
+      "loss": 1.2983,
+      "step": 1240
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019811720900355645,
+      "loss": 1.2709,
+      "step": 1245
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001980747984413145,
+      "loss": 1.2594,
+      "step": 1250
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019803192016882532,
+      "loss": 1.2682,
+      "step": 1255
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001979885743905717,
+      "loss": 1.222,
+      "step": 1260
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001979447613132659,
+      "loss": 1.2658,
+      "step": 1265
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019790048114584868,
+      "loss": 1.2875,
+      "step": 1270
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001978557340994884,
+      "loss": 1.2597,
+      "step": 1275
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019781052038757982,
+      "loss": 1.2663,
+      "step": 1280
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019776484022574322,
+      "loss": 1.2789,
+      "step": 1285
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019771869383182334,
+      "loss": 1.2344,
+      "step": 1290
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001976720814258884,
+      "loss": 1.2418,
+      "step": 1295
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001976250032302289,
+      "loss": 1.2708,
+      "step": 1300
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019757745946935668,
+      "loss": 1.2924,
+      "step": 1305
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019752945037000388,
+      "loss": 1.2402,
+      "step": 1310
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001974809761611217,
+      "loss": 1.2607,
+      "step": 1315
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001974320370738795,
+      "loss": 1.2524,
+      "step": 1320
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019738263334166353,
+      "loss": 1.2333,
+      "step": 1325
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019733276520007592,
+      "loss": 1.2093,
+      "step": 1330
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001972824328869336,
+      "loss": 1.2345,
+      "step": 1335
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019723163664226692,
+      "loss": 1.2679,
+      "step": 1340
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019718037670831887,
+      "loss": 1.216,
+      "step": 1345
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019712865332954357,
+      "loss": 1.2783,
+      "step": 1350
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019707646675260545,
+      "loss": 1.2859,
+      "step": 1355
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019702381722637772,
+      "loss": 1.2619,
+      "step": 1360
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019697070500194143,
+      "loss": 1.2491,
+      "step": 1365
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019691713033258423,
+      "loss": 1.232,
+      "step": 1370
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019686309347379907,
+      "loss": 1.2277,
+      "step": 1375
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019680859468328306,
+      "loss": 1.2479,
+      "step": 1380
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019675363422093622,
+      "loss": 1.2349,
+      "step": 1385
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001966982123488603,
+      "loss": 1.2478,
+      "step": 1390
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019664232933135734,
+      "loss": 1.2978,
+      "step": 1395
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019658598543492866,
+      "loss": 1.2366,
+      "step": 1400
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019652918092827348,
+      "loss": 1.2014,
+      "step": 1405
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019647191608228757,
+      "loss": 1.2914,
+      "step": 1410
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019641419117006202,
+      "loss": 1.2096,
+      "step": 1415
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019635600646688192,
+      "loss": 1.2096,
+      "step": 1420
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019629736225022516,
+      "loss": 1.2679,
+      "step": 1425
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019623825879976095,
+      "loss": 1.22,
+      "step": 1430
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019617869639734848,
+      "loss": 1.236,
+      "step": 1435
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019611867532703573,
+      "loss": 1.2278,
+      "step": 1440
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019605819587505807,
+      "loss": 1.2824,
+      "step": 1445
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001959972583298367,
+      "loss": 1.1901,
+      "step": 1450
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001959358629819775,
+      "loss": 1.2575,
+      "step": 1455
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019587401012426966,
+      "loss": 1.2171,
+      "step": 1460
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001958117000516841,
+      "loss": 1.25,
+      "step": 1465
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019574893306137215,
+      "loss": 1.3093,
+      "step": 1470
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019568570945266414,
+      "loss": 1.2585,
+      "step": 1475
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001956220295270681,
+      "loss": 1.2159,
+      "step": 1480
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019555789358826798,
+      "loss": 1.2412,
+      "step": 1485
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019549330194212265,
+      "loss": 1.2478,
+      "step": 1490
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000195428254896664,
+      "loss": 1.2373,
+      "step": 1495
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019536275276209585,
+      "loss": 1.3174,
+      "step": 1500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019529679585079216,
+      "loss": 1.2541,
+      "step": 1505
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019523038447729577,
+      "loss": 1.2613,
+      "step": 1510
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001951635189583168,
+      "loss": 1.1991,
+      "step": 1515
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019509619961273115,
+      "loss": 1.2859,
+      "step": 1520
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001950284267615789,
+      "loss": 1.242,
+      "step": 1525
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000194960200728063,
+      "loss": 1.2309,
+      "step": 1530
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001948915218375475,
+      "loss": 1.2167,
+      "step": 1535
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019482239041755607,
+      "loss": 1.2514,
+      "step": 1540
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019475280679777051,
+      "loss": 1.2723,
+      "step": 1545
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001946827713100291,
+      "loss": 1.259,
+      "step": 1550
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019461228428832502,
+      "loss": 1.2243,
+      "step": 1555
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019454134606880478,
+      "loss": 1.2352,
+      "step": 1560
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019446995698976663,
+      "loss": 1.2359,
+      "step": 1565
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019439811739165895,
+      "loss": 1.2812,
+      "step": 1570
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019432582761707857,
+      "loss": 1.2581,
+      "step": 1575
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019425308801076914,
+      "loss": 1.2083,
+      "step": 1580
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019417989891961965,
+      "loss": 1.2224,
+      "step": 1585
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019410626069266246,
+      "loss": 1.2381,
+      "step": 1590
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000194032173681072,
+      "loss": 1.2507,
+      "step": 1595
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019395763823816281,
+      "loss": 1.237,
+      "step": 1600
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019388265471938798,
+      "loss": 1.2314,
+      "step": 1605
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019380722348233742,
+      "loss": 1.1966,
+      "step": 1610
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001937313448867362,
+      "loss": 1.2247,
+      "step": 1615
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019365501929444277,
+      "loss": 1.2897,
+      "step": 1620
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019357824706944733,
+      "loss": 1.2616,
+      "step": 1625
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019350102857786995,
+      "loss": 1.2287,
+      "step": 1630
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000193423364187959,
+      "loss": 1.2547,
+      "step": 1635
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019334525427008924,
+      "loss": 1.2735,
+      "step": 1640
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019326669919676008,
+      "loss": 1.264,
+      "step": 1645
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019318769934259392,
+      "loss": 1.1951,
+      "step": 1650
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019310825508433427,
+      "loss": 1.2887,
+      "step": 1655
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019302836680084389,
+      "loss": 1.2265,
+      "step": 1660
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001929480348731031,
+      "loss": 1.3123,
+      "step": 1665
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000192867259684208,
+      "loss": 1.1787,
+      "step": 1670
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019278604161936843,
+      "loss": 1.3131,
+      "step": 1675
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019270438106590635,
+      "loss": 1.1932,
+      "step": 1680
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019262227841325386,
+      "loss": 1.2613,
+      "step": 1685
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019253973405295146,
+      "loss": 1.3077,
+      "step": 1690
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019245674837864608,
+      "loss": 1.2947,
+      "step": 1695
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019237332178608925,
+      "loss": 1.287,
+      "step": 1700
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019228945467313514,
+      "loss": 1.2234,
+      "step": 1705
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019220514743973878,
+      "loss": 1.231,
+      "step": 1710
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019212040048795413,
+      "loss": 1.1846,
+      "step": 1715
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019203521422193204,
+      "loss": 1.2695,
+      "step": 1720
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019194958904791848,
+      "loss": 1.189,
+      "step": 1725
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019186352537425253,
+      "loss": 1.2454,
+      "step": 1730
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019177702361136437,
+      "loss": 1.2579,
+      "step": 1735
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001916900841717735,
+      "loss": 1.2736,
+      "step": 1740
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019160270747008658,
+      "loss": 1.2265,
+      "step": 1745
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019151489392299552,
+      "loss": 1.2501,
+      "step": 1750
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001914266439492756,
+      "loss": 1.1872,
+      "step": 1755
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019133795796978328,
+      "loss": 1.248,
+      "step": 1760
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001912488364074544,
+      "loss": 1.2398,
+      "step": 1765
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019115927968730193,
+      "loss": 1.2111,
+      "step": 1770
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019106928823641416,
+      "loss": 1.2097,
+      "step": 1775
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019097886248395257,
+      "loss": 1.2963,
+      "step": 1780
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019088800286114977,
+      "loss": 1.1881,
+      "step": 1785
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019079670980130743,
+      "loss": 1.2738,
+      "step": 1790
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019070498373979428,
+      "loss": 1.2581,
+      "step": 1795
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019061282511404404,
+      "loss": 1.2211,
+      "step": 1800
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019052023436355313,
+      "loss": 1.2164,
+      "step": 1805
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019042721192987893,
+      "loss": 1.2331,
+      "step": 1810
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001903337582566374,
+      "loss": 1.2476,
+      "step": 1815
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000190239873789501,
+      "loss": 1.2337,
+      "step": 1820
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019014555897619665,
+      "loss": 1.2142,
+      "step": 1825
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019005081426650355,
+      "loss": 1.2195,
+      "step": 1830
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018995564011225108,
+      "loss": 1.2261,
+      "step": 1835
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018986003696731655,
+      "loss": 1.2049,
+      "step": 1840
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001897640052876231,
+      "loss": 1.2587,
+      "step": 1845
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018966754553113751,
+      "loss": 1.2747,
+      "step": 1850
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018957065815786807,
+      "loss": 1.2517,
+      "step": 1855
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018947334362986223,
+      "loss": 1.1749,
+      "step": 1860
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018937560241120461,
+      "loss": 1.2928,
+      "step": 1865
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001892774349680147,
+      "loss": 1.2554,
+      "step": 1870
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018917884176844444,
+      "loss": 1.3226,
+      "step": 1875
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001890798232826764,
+      "loss": 1.1807,
+      "step": 1880
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018898037998292116,
+      "loss": 1.2396,
+      "step": 1885
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001888805123434152,
+      "loss": 1.2452,
+      "step": 1890
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018878022084041867,
+      "loss": 1.2315,
+      "step": 1895
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018867950595221312,
+      "loss": 1.207,
+      "step": 1900
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018857836815909916,
+      "loss": 1.2217,
+      "step": 1905
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018847680794339418,
+      "loss": 1.2546,
+      "step": 1910
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018837482578943008,
+      "loss": 1.2377,
+      "step": 1915
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018827242218355097,
+      "loss": 1.2281,
+      "step": 1920
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018816959761411078,
+      "loss": 1.2354,
+      "step": 1925
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018806635257147099,
+      "loss": 1.2587,
+      "step": 1930
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018796268754799834,
+      "loss": 1.2403,
+      "step": 1935
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018785860303806233,
+      "loss": 1.2141,
+      "step": 1940
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018775409953803308,
+      "loss": 1.2911,
+      "step": 1945
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018764917754627867,
+      "loss": 1.2265,
+      "step": 1950
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018754383756316306,
+      "loss": 1.3048,
+      "step": 1955
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018743808009104356,
+      "loss": 1.2346,
+      "step": 1960
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001873319056342684,
+      "loss": 1.3056,
+      "step": 1965
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018722531469917434,
+      "loss": 1.2064,
+      "step": 1970
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018711830779408447,
+      "loss": 1.2604,
+      "step": 1975
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001870108854293054,
+      "loss": 1.2248,
+      "step": 1980
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018690304811712512,
+      "loss": 1.2223,
+      "step": 1985
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001867947963718105,
+      "loss": 1.3052,
+      "step": 1990
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018668613070960474,
+      "loss": 1.2639,
+      "step": 1995
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018657705164872504,
+      "loss": 1.2477,
+      "step": 2000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018646755970935994,
+      "loss": 1.2161,
+      "step": 2005
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001863576554136671,
+      "loss": 1.2985,
+      "step": 2010
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018624733928577066,
+      "loss": 1.2563,
+      "step": 2015
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018613661185175862,
+      "loss": 1.202,
+      "step": 2020
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018602547363968063,
+      "loss": 1.2174,
+      "step": 2025
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001859139251795452,
+      "loss": 1.2104,
+      "step": 2030
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018580196700331732,
+      "loss": 1.2556,
+      "step": 2035
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001856895996449158,
+      "loss": 1.2207,
+      "step": 2040
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018557682364021098,
+      "loss": 1.2169,
+      "step": 2045
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001854636395270218,
+      "loss": 1.2957,
+      "step": 2050
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018535004784511362,
+      "loss": 1.2047,
+      "step": 2055
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018523604913619525,
+      "loss": 1.2236,
+      "step": 2060
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018512164394391678,
+      "loss": 1.235,
+      "step": 2065
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018500683281386663,
+      "loss": 1.2838,
+      "step": 2070
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018489161629356925,
+      "loss": 1.1869,
+      "step": 2075
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018477599493248222,
+      "loss": 1.2263,
+      "step": 2080
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018465996928199383,
+      "loss": 1.2461,
+      "step": 2085
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001845435398954204,
+      "loss": 1.2032,
+      "step": 2090
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018442670732800364,
+      "loss": 1.2334,
+      "step": 2095
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018430947213690795,
+      "loss": 1.2926,
+      "step": 2100
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018419183488121784,
+      "loss": 1.2523,
+      "step": 2105
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018407379612193526,
+      "loss": 1.189,
+      "step": 2110
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001839553564219768,
+      "loss": 1.1997,
+      "step": 2115
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018383651634617122,
+      "loss": 1.2749,
+      "step": 2120
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001837172764612566,
+      "loss": 1.3316,
+      "step": 2125
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001835976373358776,
+      "loss": 1.2493,
+      "step": 2130
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018347759954058293,
+      "loss": 1.2436,
+      "step": 2135
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001833571636478224,
+      "loss": 1.2612,
+      "step": 2140
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018323633023194447,
+      "loss": 1.2147,
+      "step": 2145
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018311509986919324,
+      "loss": 1.2306,
+      "step": 2150
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018299347313770583,
+      "loss": 1.2275,
+      "step": 2155
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018287145061750962,
+      "loss": 1.2594,
+      "step": 2160
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018274903289051947,
+      "loss": 1.2519,
+      "step": 2165
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000182626220540535,
+      "loss": 1.2363,
+      "step": 2170
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018250301415323765,
+      "loss": 1.2334,
+      "step": 2175
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018237941431618806,
+      "loss": 1.2675,
+      "step": 2180
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001822554216188232,
+      "loss": 1.1873,
+      "step": 2185
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018213103665245357,
+      "loss": 1.2302,
+      "step": 2190
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018200626001026025,
+      "loss": 1.2816,
+      "step": 2195
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018188109228729234,
+      "loss": 1.2215,
+      "step": 2200
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018175553408046385,
+      "loss": 1.2307,
+      "step": 2205
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001816295859885511,
+      "loss": 1.2093,
+      "step": 2210
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018150324861218966,
+      "loss": 1.241,
+      "step": 2215
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001813765225538715,
+      "loss": 1.3257,
+      "step": 2220
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018124940841794231,
+      "loss": 1.207,
+      "step": 2225
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018112190681059842,
+      "loss": 1.2373,
+      "step": 2230
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000180994018339884,
+      "loss": 1.2698,
+      "step": 2235
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018086574361568812,
+      "loss": 1.2235,
+      "step": 2240
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001807370832497418,
+      "loss": 1.2563,
+      "step": 2245
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018060803785561533,
+      "loss": 1.2776,
+      "step": 2250
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000180478608048715,
+      "loss": 1.2367,
+      "step": 2255
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018034879444628037,
+      "loss": 1.2216,
+      "step": 2260
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018021859766738133,
+      "loss": 1.2369,
+      "step": 2265
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001800880183329151,
+      "loss": 1.2362,
+      "step": 2270
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017995705706560318,
+      "loss": 1.2789,
+      "step": 2275
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001798257144899886,
+      "loss": 1.2832,
+      "step": 2280
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017969399123243278,
+      "loss": 1.2632,
+      "step": 2285
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017956188792111251,
+      "loss": 1.2231,
+      "step": 2290
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017942940518601714,
+      "loss": 1.3005,
+      "step": 2295
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017929654365894535,
+      "loss": 1.2639,
+      "step": 2300
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017916330397350233,
+      "loss": 1.2834,
+      "step": 2305
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001790296867650966,
+      "loss": 1.2117,
+      "step": 2310
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001788956926709371,
+      "loss": 1.2173,
+      "step": 2315
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001787613223300301,
+      "loss": 1.2782,
+      "step": 2320
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017862657638317617,
+      "loss": 1.2452,
+      "step": 2325
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001784914554729671,
+      "loss": 1.2861,
+      "step": 2330
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017835596024378284,
+      "loss": 1.2031,
+      "step": 2335
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017822009134178842,
+      "loss": 1.2494,
+      "step": 2340
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017808384941493094,
+      "loss": 1.2082,
+      "step": 2345
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001779472351129363,
+      "loss": 1.1589,
+      "step": 2350
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017781024908730645,
+      "loss": 1.2552,
+      "step": 2355
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017767289199131585,
+      "loss": 1.2197,
+      "step": 2360
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017753516448000865,
+      "loss": 1.175,
+      "step": 2365
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017739706721019543,
+      "loss": 1.2568,
+      "step": 2370
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017725860084045016,
+      "loss": 1.2534,
+      "step": 2375
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000177119766031107,
+      "loss": 1.2676,
+      "step": 2380
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001769805634442572,
+      "loss": 1.2349,
+      "step": 2385
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017684099374374584,
+      "loss": 1.2945,
+      "step": 2390
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017670105759516882,
+      "loss": 1.254,
+      "step": 2395
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017656075566586948,
+      "loss": 1.2296,
+      "step": 2400
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001764200886249357,
+      "loss": 1.2336,
+      "step": 2405
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001762790571431964,
+      "loss": 1.2217,
+      "step": 2410
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017613766189321855,
+      "loss": 1.2477,
+      "step": 2415
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001759959035493039,
+      "loss": 1.2813,
+      "step": 2420
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017585378278748572,
+      "loss": 1.2134,
+      "step": 2425
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001757113002855257,
+      "loss": 1.2578,
+      "step": 2430
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017556845672291055,
+      "loss": 1.2457,
+      "step": 2435
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017542525278084893,
+      "loss": 1.2261,
+      "step": 2440
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017528168914226802,
+      "loss": 1.2406,
+      "step": 2445
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017513776649181052,
+      "loss": 1.2575,
+      "step": 2450
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000174993485515831,
+      "loss": 1.1964,
+      "step": 2455
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001748488469023931,
+      "loss": 1.2288,
+      "step": 2460
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001747038513412658,
+      "loss": 1.2274,
+      "step": 2465
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017455849952392044,
+      "loss": 1.2206,
+      "step": 2470
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017441279214352727,
+      "loss": 1.2528,
+      "step": 2475
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017426672989495222,
+      "loss": 1.2718,
+      "step": 2480
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017412031347475358,
+      "loss": 1.1526,
+      "step": 2485
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017397354358117856,
+      "loss": 1.2691,
+      "step": 2490
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017382642091416016,
+      "loss": 1.2351,
+      "step": 2495
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001736789461753136,
+      "loss": 1.2247,
+      "step": 2500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017353112006793331,
+      "loss": 1.2044,
+      "step": 2505
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017338294329698913,
+      "loss": 1.2284,
+      "step": 2510
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017323441656912333,
+      "loss": 1.2372,
+      "step": 2515
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017308554059264702,
+      "loss": 1.2538,
+      "step": 2520
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017293631607753692,
+      "loss": 1.2139,
+      "step": 2525
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017278674373543185,
+      "loss": 1.1928,
+      "step": 2530
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017263682427962938,
+      "loss": 1.2522,
+      "step": 2535
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017248655842508245,
+      "loss": 1.2457,
+      "step": 2540
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017233594688839595,
+      "loss": 1.2094,
+      "step": 2545
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017218499038782335,
+      "loss": 1.1951,
+      "step": 2550
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001720336896432631,
+      "loss": 1.2621,
+      "step": 2555
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017188204537625543,
+      "loss": 1.2051,
+      "step": 2560
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017173005830997874,
+      "loss": 1.248,
+      "step": 2565
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017157772916924624,
+      "loss": 1.2464,
+      "step": 2570
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017142505868050244,
+      "loss": 1.269,
+      "step": 2575
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017127204757181973,
+      "loss": 1.2079,
+      "step": 2580
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017111869657289485,
+      "loss": 1.2464,
+      "step": 2585
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017096500641504552,
+      "loss": 1.1963,
+      "step": 2590
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017081097783120682,
+      "loss": 1.2332,
+      "step": 2595
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017065661155592777,
+      "loss": 1.1606,
+      "step": 2600
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001705019083253678,
+      "loss": 1.2465,
+      "step": 2605
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017034686887729326,
+      "loss": 1.2307,
+      "step": 2610
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001701914939510739,
+      "loss": 1.2601,
+      "step": 2615
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001700357842876793,
+      "loss": 1.2441,
+      "step": 2620
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016987974062967549,
+      "loss": 1.2618,
+      "step": 2625
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016972336372122113,
+      "loss": 1.1907,
+      "step": 2630
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001695666543080642,
+      "loss": 1.2297,
+      "step": 2635
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001694096131375384,
+      "loss": 1.2401,
+      "step": 2640
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016925224095855947,
+      "loss": 1.2369,
+      "step": 2645
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016909453852162178,
+      "loss": 1.2378,
+      "step": 2650
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016893650657879463,
+      "loss": 1.2851,
+      "step": 2655
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001687781458837187,
+      "loss": 1.31,
+      "step": 2660
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016861945719160248,
+      "loss": 1.2882,
+      "step": 2665
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016846044125921867,
+      "loss": 1.2105,
+      "step": 2670
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016830109884490049,
+      "loss": 1.2422,
+      "step": 2675
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016814143070853813,
+      "loss": 1.2301,
+      "step": 2680
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016798143761157525,
+      "loss": 1.2842,
+      "step": 2685
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016782112031700502,
+      "loss": 1.2508,
+      "step": 2690
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001676604795893668,
+      "loss": 1.202,
+      "step": 2695
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016749951619474243,
+      "loss": 1.2585,
+      "step": 2700
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016733823090075232,
+      "loss": 1.2397,
+      "step": 2705
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001671766244765522,
+      "loss": 1.2387,
+      "step": 2710
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016701469769282913,
+      "loss": 1.2806,
+      "step": 2715
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016685245132179794,
+      "loss": 1.2827,
+      "step": 2720
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016668988613719764,
+      "loss": 1.2304,
+      "step": 2725
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016652700291428744,
+      "loss": 1.2337,
+      "step": 2730
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016636380242984346,
+      "loss": 1.2598,
+      "step": 2735
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001662002854621547,
+      "loss": 1.2078,
+      "step": 2740
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016603645279101952,
+      "loss": 1.2609,
+      "step": 2745
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016587230519774169,
+      "loss": 1.211,
+      "step": 2750
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016570784346512694,
+      "loss": 1.2144,
+      "step": 2755
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016554306837747912,
+      "loss": 1.2476,
+      "step": 2760
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001653779807205963,
+      "loss": 1.2652,
+      "step": 2765
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001652125812817674,
+      "loss": 1.2504,
+      "step": 2770
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016504687084976794,
+      "loss": 1.1935,
+      "step": 2775
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016488085021485675,
+      "loss": 1.2251,
+      "step": 2780
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016471452016877188,
+      "loss": 1.2192,
+      "step": 2785
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016454788150472693,
+      "loss": 1.2401,
+      "step": 2790
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016438093501740735,
+      "loss": 1.2254,
+      "step": 2795
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001642136815029665,
+      "loss": 1.2878,
+      "step": 2800
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016404612175902197,
+      "loss": 1.2561,
+      "step": 2805
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016387825658465167,
+      "loss": 1.2686,
+      "step": 2810
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016371008678039018,
+      "loss": 1.2746,
+      "step": 2815
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016354161314822475,
+      "loss": 1.2332,
+      "step": 2820
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016337283649159158,
+      "loss": 1.2709,
+      "step": 2825
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016320375761537202,
+      "loss": 1.2278,
+      "step": 2830
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016303437732588858,
+      "loss": 1.2682,
+      "step": 2835
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016286469643090126,
+      "loss": 1.2119,
+      "step": 2840
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001626947157396036,
+      "loss": 1.2684,
+      "step": 2845
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001625244360626189,
+      "loss": 1.2188,
+      "step": 2850
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001623538582119961,
+      "loss": 1.2124,
+      "step": 2855
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001621829830012064,
+      "loss": 1.191,
+      "step": 2860
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001620118112451388,
+      "loss": 1.1922,
+      "step": 2865
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016184034376009673,
+      "loss": 1.2112,
+      "step": 2870
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016166858136379376,
+      "loss": 1.2414,
+      "step": 2875
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016149652487534993,
+      "loss": 1.2477,
+      "step": 2880
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016132417511528782,
+      "loss": 1.2556,
+      "step": 2885
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016115153290552855,
+      "loss": 1.2735,
+      "step": 2890
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001609785990693879,
+      "loss": 1.2619,
+      "step": 2895
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016080537443157237,
+      "loss": 1.2559,
+      "step": 2900
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016063185981817537,
+      "loss": 1.3063,
+      "step": 2905
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016045805605667307,
+      "loss": 1.2238,
+      "step": 2910
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001602839639759206,
+      "loss": 1.2689,
+      "step": 2915
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00016010958440614812,
+      "loss": 1.2607,
+      "step": 2920
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001599349181789567,
+      "loss": 1.2254,
+      "step": 2925
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001597599661273145,
+      "loss": 1.1683,
+      "step": 2930
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015958472908555272,
+      "loss": 1.268,
+      "step": 2935
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001594092078893618,
+      "loss": 1.2261,
+      "step": 2940
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015923340337578706,
+      "loss": 1.2434,
+      "step": 2945
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015905731638322506,
+      "loss": 1.1929,
+      "step": 2950
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001588809477514195,
+      "loss": 1.1871,
+      "step": 2955
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015870429832145713,
+      "loss": 1.227,
+      "step": 2960
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015852736893576385,
+      "loss": 1.2298,
+      "step": 2965
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015835016043810066,
+      "loss": 1.2299,
+      "step": 2970
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015817267367355953,
+      "loss": 1.251,
+      "step": 2975
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015799490948855956,
+      "loss": 1.2291,
+      "step": 2980
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015781686873084276,
+      "loss": 1.2134,
+      "step": 2985
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015763855224947018,
+      "loss": 1.2284,
+      "step": 2990
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001574599608948177,
+      "loss": 1.2804,
+      "step": 2995
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015728109551857203,
+      "loss": 1.2652,
+      "step": 3000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015710195697372684,
+      "loss": 1.2871,
+      "step": 3005
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015692254611457824,
+      "loss": 1.2627,
+      "step": 3010
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015674286379672121,
+      "loss": 1.2247,
+      "step": 3015
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001565629108770452,
+      "loss": 1.2207,
+      "step": 3020
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001563826882137301,
+      "loss": 1.2157,
+      "step": 3025
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015620219666624233,
+      "loss": 1.2401,
+      "step": 3030
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015602143709533042,
+      "loss": 1.2599,
+      "step": 3035
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015584041036302118,
+      "loss": 1.1988,
+      "step": 3040
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015565911733261546,
+      "loss": 1.2576,
+      "step": 3045
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015547755886868408,
+      "loss": 1.3267,
+      "step": 3050
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015529573583706374,
+      "loss": 1.2688,
+      "step": 3055
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015511364910485268,
+      "loss": 1.2372,
+      "step": 3060
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015493129954040685,
+      "loss": 1.2339,
+      "step": 3065
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015474868801333558,
+      "loss": 1.2404,
+      "step": 3070
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015456581539449754,
+      "loss": 1.1373,
+      "step": 3075
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015438268255599632,
+      "loss": 1.253,
+      "step": 3080
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015419929037117675,
+      "loss": 1.1828,
+      "step": 3085
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015401563971462027,
+      "loss": 1.1713,
+      "step": 3090
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015383173146214102,
+      "loss": 1.2104,
+      "step": 3095
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015364756649078159,
+      "loss": 1.2163,
+      "step": 3100
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015346314567880878,
+      "loss": 1.1944,
+      "step": 3105
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015327846990570958,
+      "loss": 1.2301,
+      "step": 3110
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001530935400521868,
+      "loss": 1.2001,
+      "step": 3115
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001529083570001549,
+      "loss": 1.2168,
+      "step": 3120
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015272292163273594,
+      "loss": 1.238,
+      "step": 3125
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001525372348342551,
+      "loss": 1.2627,
+      "step": 3130
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015235129749023674,
+      "loss": 1.2227,
+      "step": 3135
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015216511048740005,
+      "loss": 1.1709,
+      "step": 3140
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015197867471365467,
+      "loss": 1.2762,
+      "step": 3145
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015179199105809675,
+      "loss": 1.2718,
+      "step": 3150
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001516050604110046,
+      "loss": 1.2289,
+      "step": 3155
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00015141788366383423,
+      "loss": 1.2397,
+      "step": 3160
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00015123046170921543,
+      "loss": 1.1637,
+      "step": 3165
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00015104279544094737,
+      "loss": 1.2437,
+      "step": 3170
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00015085488575399417,
+      "loss": 1.1972,
+      "step": 3175
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00015066673354448105,
+      "loss": 1.2844,
+      "step": 3180
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00015047833970968953,
+      "loss": 1.3154,
+      "step": 3185
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001502897051480536,
+      "loss": 1.2179,
+      "step": 3190
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001501008307591552,
+      "loss": 1.2595,
+      "step": 3195
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014991171744372,
+      "loss": 1.2375,
+      "step": 3200
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000149722366103613,
+      "loss": 1.2673,
+      "step": 3205
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014953277764183452,
+      "loss": 1.2851,
+      "step": 3210
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014934295296251547,
+      "loss": 1.2117,
+      "step": 3215
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014915289297091338,
+      "loss": 1.2198,
+      "step": 3220
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014896259857340795,
+      "loss": 1.1985,
+      "step": 3225
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014877207067749674,
+      "loss": 1.2304,
+      "step": 3230
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014858131019179082,
+      "loss": 1.2464,
+      "step": 3235
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014839031802601048,
+      "loss": 1.2517,
+      "step": 3240
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014819909509098094,
+      "loss": 1.2296,
+      "step": 3245
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014800764229862774,
+      "loss": 1.2287,
+      "step": 3250
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014781596056197282,
+      "loss": 1.2462,
+      "step": 3255
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001476240507951298,
+      "loss": 1.2502,
+      "step": 3260
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001474319139132998,
+      "loss": 1.2512,
+      "step": 3265
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000147239550832767,
+      "loss": 1.2273,
+      "step": 3270
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014704696247089437,
+      "loss": 1.2185,
+      "step": 3275
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014685414974611912,
+      "loss": 1.2294,
+      "step": 3280
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014666111357794853,
+      "loss": 1.2273,
+      "step": 3285
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014646785488695545,
+      "loss": 1.1954,
+      "step": 3290
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014627437459477382,
+      "loss": 1.1856,
+      "step": 3295
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014608067362409453,
+      "loss": 1.1759,
+      "step": 3300
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014588675289866073,
+      "loss": 1.2482,
+      "step": 3305
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014569261334326363,
+      "loss": 1.2644,
+      "step": 3310
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014549825588373804,
+      "loss": 1.251,
+      "step": 3315
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014530368144695784,
+      "loss": 1.2328,
+      "step": 3320
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014510889096083177,
+      "loss": 1.2098,
+      "step": 3325
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014491388535429887,
+      "loss": 1.216,
+      "step": 3330
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014471866555732397,
+      "loss": 1.2758,
+      "step": 3335
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001445232325008935,
+      "loss": 1.2326,
+      "step": 3340
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001443275871170108,
+      "loss": 1.2107,
+      "step": 3345
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014413173033869182,
+      "loss": 1.2861,
+      "step": 3350
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014393566309996064,
+      "loss": 1.2235,
+      "step": 3355
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014373938633584495,
+      "loss": 1.2372,
+      "step": 3360
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014354290098237175,
+      "loss": 1.2765,
+      "step": 3365
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014334620797656274,
+      "loss": 1.1621,
+      "step": 3370
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001431493082564298,
+      "loss": 1.2785,
+      "step": 3375
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014295220276097076,
+      "loss": 1.1814,
+      "step": 3380
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014275489243016472,
+      "loss": 1.2518,
+      "step": 3385
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014255737820496759,
+      "loss": 1.2243,
+      "step": 3390
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014235966102730765,
+      "loss": 1.223,
+      "step": 3395
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001421617418400811,
+      "loss": 1.1933,
+      "step": 3400
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001419636215871474,
+      "loss": 1.2774,
+      "step": 3405
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014176530121332495,
+      "loss": 1.2666,
+      "step": 3410
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014156678166438652,
+      "loss": 1.1821,
+      "step": 3415
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014136806388705464,
+      "loss": 1.178,
+      "step": 3420
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014116914882899732,
+      "loss": 1.2215,
+      "step": 3425
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014097003743882326,
+      "loss": 1.2634,
+      "step": 3430
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014077073066607744,
+      "loss": 1.1969,
+      "step": 3435
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014057122946123672,
+      "loss": 1.2263,
+      "step": 3440
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014037153477570514,
+      "loss": 1.2308,
+      "step": 3445
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014017164756180935,
+      "loss": 1.1962,
+      "step": 3450
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013997156877279426,
+      "loss": 1.2014,
+      "step": 3455
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013977129936281835,
+      "loss": 1.2328,
+      "step": 3460
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001395708402869491,
+      "loss": 1.227,
+      "step": 3465
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013937019250115856,
+      "loss": 1.2449,
+      "step": 3470
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013916935696231872,
+      "loss": 1.2108,
+      "step": 3475
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013896833462819693,
+      "loss": 1.2102,
+      "step": 3480
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001387671264574513,
+      "loss": 1.1968,
+      "step": 3485
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013856573340962627,
+      "loss": 1.2224,
+      "step": 3490
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013836415644514788,
+      "loss": 1.2414,
+      "step": 3495
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013816239652531928,
+      "loss": 1.2562,
+      "step": 3500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001379604546123161,
+      "loss": 1.2191,
+      "step": 3505
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001377583316691819,
+      "loss": 1.3249,
+      "step": 3510
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001375560286598236,
+      "loss": 1.2205,
+      "step": 3515
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013735354654900665,
+      "loss": 1.2754,
+      "step": 3520
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013715088630235087,
+      "loss": 1.2142,
+      "step": 3525
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013694804888632545,
+      "loss": 1.2218,
+      "step": 3530
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013674503526824453,
+      "loss": 1.2239,
+      "step": 3535
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013654184641626252,
+      "loss": 1.1762,
+      "step": 3540
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001363384832993695,
+      "loss": 1.2461,
+      "step": 3545
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001361349468873866,
+      "loss": 1.2013,
+      "step": 3550
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013593123815096146,
+      "loss": 1.2304,
+      "step": 3555
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013572735806156343,
+      "loss": 1.2343,
+      "step": 3560
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013552330759147903,
+      "loss": 1.2204,
+      "step": 3565
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001353190877138074,
+      "loss": 1.2359,
+      "step": 3570
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001351146994024554,
+      "loss": 1.1503,
+      "step": 3575
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013491014363213336,
+      "loss": 1.2075,
+      "step": 3580
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013470542137835004,
+      "loss": 1.2239,
+      "step": 3585
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013450053361740814,
+      "loss": 1.2348,
+      "step": 3590
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001342954813263998,
+      "loss": 1.2395,
+      "step": 3595
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013409026548320165,
+      "loss": 1.2416,
+      "step": 3600
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013388488706647032,
+      "loss": 1.2191,
+      "step": 3605
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013367934705563777,
+      "loss": 1.2614,
+      "step": 3610
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013347364643090655,
+      "loss": 1.2729,
+      "step": 3615
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013326778617324517,
+      "loss": 1.2628,
+      "step": 3620
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001330617672643835,
+      "loss": 1.2214,
+      "step": 3625
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013285559068680784,
+      "loss": 1.2435,
+      "step": 3630
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013264925742375656,
+      "loss": 1.2333,
+      "step": 3635
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013244276845921513,
+      "loss": 1.233,
+      "step": 3640
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001322361247779116,
+      "loss": 1.2332,
+      "step": 3645
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001320293273653119,
+      "loss": 1.2308,
+      "step": 3650
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013182237720761497,
+      "loss": 1.2198,
+      "step": 3655
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013161527529174827,
+      "loss": 1.219,
+      "step": 3660
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000131408022605363,
+      "loss": 1.2969,
+      "step": 3665
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013120062013682928,
+      "loss": 1.1839,
+      "step": 3670
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001309930688752316,
+      "loss": 1.2227,
+      "step": 3675
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000130785369810364,
+      "loss": 1.2469,
+      "step": 3680
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013057752393272538,
+      "loss": 1.2505,
+      "step": 3685
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013036953223351477,
+      "loss": 1.2452,
+      "step": 3690
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013016139570462673,
+      "loss": 1.2318,
+      "step": 3695
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001299531153386463,
+      "loss": 1.2157,
+      "step": 3700
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012974469212884458,
+      "loss": 1.2075,
+      "step": 3705
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012953612706917387,
+      "loss": 1.2625,
+      "step": 3710
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012932742115426296,
+      "loss": 1.2312,
+      "step": 3715
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012911857537941233,
+      "loss": 1.252,
+      "step": 3720
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012890959074058942,
+      "loss": 1.1984,
+      "step": 3725
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012870046823442396,
+      "loss": 1.2742,
+      "step": 3730
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012849120885820314,
+      "loss": 1.2287,
+      "step": 3735
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012828181360986684,
+      "loss": 1.2188,
+      "step": 3740
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012807228348800292,
+      "loss": 1.2021,
+      "step": 3745
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012786261949184244,
+      "loss": 1.2176,
+      "step": 3750
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012765282262125493,
+      "loss": 1.2642,
+      "step": 3755
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001274428938767435,
+      "loss": 1.2589,
+      "step": 3760
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001272328342594403,
+      "loss": 1.2158,
+      "step": 3765
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012702264477110144,
+      "loss": 1.2653,
+      "step": 3770
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012681232641410244,
+      "loss": 1.1943,
+      "step": 3775
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001266018801914334,
+      "loss": 1.2019,
+      "step": 3780
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012639130710669422,
+      "loss": 1.239,
+      "step": 3785
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001261806081640897,
+      "loss": 1.2525,
+      "step": 3790
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012596978436842497,
+      "loss": 1.2084,
+      "step": 3795
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001257588367251005,
+      "loss": 1.2541,
+      "step": 3800
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012554776624010736,
+      "loss": 1.2445,
+      "step": 3805
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012533657392002254,
+      "loss": 1.3241,
+      "step": 3810
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001251252607720039,
+      "loss": 1.214,
+      "step": 3815
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012491382780378563,
+      "loss": 1.1838,
+      "step": 3820
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001247022760236733,
+      "loss": 1.2175,
+      "step": 3825
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012449060644053906,
+      "loss": 1.2644,
+      "step": 3830
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012427882006381697,
+      "loss": 1.2223,
+      "step": 3835
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001240669179034979,
+      "loss": 1.2521,
+      "step": 3840
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000123854900970125,
+      "loss": 1.21,
+      "step": 3845
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001236427702747887,
+      "loss": 1.2301,
+      "step": 3850
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000123430526829122,
+      "loss": 1.191,
+      "step": 3855
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012321817164529547,
+      "loss": 1.2101,
+      "step": 3860
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001230057057360128,
+      "loss": 1.2101,
+      "step": 3865
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012279313011450546,
+      "loss": 1.2114,
+      "step": 3870
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012258044579452828,
+      "loss": 1.1698,
+      "step": 3875
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001223676537903544,
+      "loss": 1.1669,
+      "step": 3880
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012215475511677054,
+      "loss": 1.2586,
+      "step": 3885
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012194175078907211,
+      "loss": 1.2373,
+      "step": 3890
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012172864182305832,
+      "loss": 1.1786,
+      "step": 3895
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012151542923502745,
+      "loss": 1.1985,
+      "step": 3900
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012130211404177197,
+      "loss": 1.2568,
+      "step": 3905
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012108869726057354,
+      "loss": 1.2496,
+      "step": 3910
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012087517990919839,
+      "loss": 1.2345,
+      "step": 3915
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012066156300589237,
+      "loss": 1.262,
+      "step": 3920
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012044784756937603,
+      "loss": 1.2524,
+      "step": 3925
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012023403461883985,
+      "loss": 1.2083,
+      "step": 3930
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012002012517393931,
+      "loss": 1.2374,
+      "step": 3935
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011980612025479012,
+      "loss": 1.1824,
+      "step": 3940
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001195920208819632,
+      "loss": 1.2438,
+      "step": 3945
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011937782807648009,
+      "loss": 1.2397,
+      "step": 3950
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011916354285980767,
+      "loss": 1.2076,
+      "step": 3955
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011894916625385368,
+      "loss": 1.1825,
+      "step": 3960
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011873469928096169,
+      "loss": 1.2031,
+      "step": 3965
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011852014296390612,
+      "loss": 1.301,
+      "step": 3970
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011830549832588755,
+      "loss": 1.2132,
+      "step": 3975
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011809076639052774,
+      "loss": 1.2509,
+      "step": 3980
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011787594818186474,
+      "loss": 1.2458,
+      "step": 3985
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011766104472434803,
+      "loss": 1.2608,
+      "step": 3990
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011744605704283369,
+      "loss": 1.2024,
+      "step": 3995
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011723098616257935,
+      "loss": 1.2155,
+      "step": 4000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011701583310923953,
+      "loss": 1.3011,
+      "step": 4005
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011680059890886055,
+      "loss": 1.1795,
+      "step": 4010
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011658528458787568,
+      "loss": 1.2192,
+      "step": 4015
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011636989117310043,
+      "loss": 1.3378,
+      "step": 4020
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001161544196917273,
+      "loss": 1.1918,
+      "step": 4025
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011593887117132123,
+      "loss": 1.254,
+      "step": 4030
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011572324663981453,
+      "loss": 1.218,
+      "step": 4035
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011550754712550196,
+      "loss": 1.2528,
+      "step": 4040
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011529177365703583,
+      "loss": 1.214,
+      "step": 4045
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011507592726342127,
+      "loss": 1.235,
+      "step": 4050
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011486000897401104,
+      "loss": 1.1889,
+      "step": 4055
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011464401981850081,
+      "loss": 1.2363,
+      "step": 4060
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011442796082692426,
+      "loss": 1.2479,
+      "step": 4065
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011421183302964806,
+      "loss": 1.2125,
+      "step": 4070
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011399563745736701,
+      "loss": 1.2325,
+      "step": 4075
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011377937514109913,
+      "loss": 1.2165,
+      "step": 4080
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011356304711218076,
+      "loss": 1.2282,
+      "step": 4085
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011334665440226156,
+      "loss": 1.186,
+      "step": 4090
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011313019804329969,
+      "loss": 1.228,
+      "step": 4095
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011291367906755683,
+      "loss": 1.186,
+      "step": 4100
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011269709850759334,
+      "loss": 1.2139,
+      "step": 4105
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011248045739626314,
+      "loss": 1.2892,
+      "step": 4110
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011226375676670901,
+      "loss": 1.2507,
+      "step": 4115
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001120469976523576,
+      "loss": 1.1644,
+      "step": 4120
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011183018108691435,
+      "loss": 1.2377,
+      "step": 4125
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011161330810435877,
+      "loss": 1.1979,
+      "step": 4130
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011139637973893939,
+      "loss": 1.2603,
+      "step": 4135
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011117939702516885,
+      "loss": 1.219,
+      "step": 4140
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011096236099781896,
+      "loss": 1.2541,
+      "step": 4145
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011074527269191585,
+      "loss": 1.2586,
+      "step": 4150
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001105281331427349,
+      "loss": 1.2229,
+      "step": 4155
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001103109433857958,
+      "loss": 1.2413,
+      "step": 4160
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011009370445685786,
+      "loss": 1.2441,
+      "step": 4165
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001098764173919147,
+      "loss": 1.2482,
+      "step": 4170
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010965908322718961,
+      "loss": 1.3298,
+      "step": 4175
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010944170299913048,
+      "loss": 1.2353,
+      "step": 4180
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010922427774440478,
+      "loss": 1.2067,
+      "step": 4185
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010900680849989485,
+      "loss": 1.2329,
+      "step": 4190
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001087892963026927,
+      "loss": 1.2823,
+      "step": 4195
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010857174219009524,
+      "loss": 1.2351,
+      "step": 4200
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010835414719959927,
+      "loss": 1.2786,
+      "step": 4205
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010813651236889648,
+      "loss": 1.2031,
+      "step": 4210
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010791883873586863,
+      "loss": 1.2501,
+      "step": 4215
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010770112733858242,
+      "loss": 1.2632,
+      "step": 4220
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001074833792152848,
+      "loss": 1.2244,
+      "step": 4225
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010726559540439768,
+      "loss": 1.1696,
+      "step": 4230
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010704777694451332,
+      "loss": 1.2806,
+      "step": 4235
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010682992487438911,
+      "loss": 1.2137,
+      "step": 4240
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001066120402329428,
+      "loss": 1.2607,
+      "step": 4245
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001063941240592474,
+      "loss": 1.2313,
+      "step": 4250
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010617617739252636,
+      "loss": 1.1915,
+      "step": 4255
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001059582012721485,
+      "loss": 1.2152,
+      "step": 4260
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010574019673762314,
+      "loss": 1.2638,
+      "step": 4265
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010552216482859507,
+      "loss": 1.2243,
+      "step": 4270
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010530410658483966,
+      "loss": 1.1519,
+      "step": 4275
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010508602304625783,
+      "loss": 1.225,
+      "step": 4280
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010486791525287114,
+      "loss": 1.2854,
+      "step": 4285
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010464978424481687,
+      "loss": 1.2069,
+      "step": 4290
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001044316310623429,
+      "loss": 1.2527,
+      "step": 4295
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000104213456745803,
+      "loss": 1.2293,
+      "step": 4300
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010399526233565158,
+      "loss": 1.2178,
+      "step": 4305
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010377704887243902,
+      "loss": 1.2555,
+      "step": 4310
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010355881739680642,
+      "loss": 1.2392,
+      "step": 4315
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001033405689494809,
+      "loss": 1.2851,
+      "step": 4320
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010312230457127043,
+      "loss": 1.2787,
+      "step": 4325
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010290402530305901,
+      "loss": 1.2171,
+      "step": 4330
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001026857321858016,
+      "loss": 1.2425,
+      "step": 4335
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010246742626051927,
+      "loss": 1.1885,
+      "step": 4340
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010224910856829409,
+      "loss": 1.2036,
+      "step": 4345
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001020307801502643,
+      "loss": 1.2027,
+      "step": 4350
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001018124420476193,
+      "loss": 1.2281,
+      "step": 4355
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010159409530159465,
+      "loss": 1.2505,
+      "step": 4360
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010137574095346712,
+      "loss": 1.236,
+      "step": 4365
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010115738004454971,
+      "loss": 1.2223,
+      "step": 4370
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010093901361618681,
+      "loss": 1.206,
+      "step": 4375
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010072064270974904,
+      "loss": 1.2302,
+      "step": 4380
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010050226836662839,
+      "loss": 1.2105,
+      "step": 4385
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010028389162823326,
+      "loss": 1.2582,
+      "step": 4390
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010006551353598346,
+      "loss": 1.2539,
+      "step": 4395
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.984713513130528e-05,
+      "loss": 1.2426,
+      "step": 4400
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.962875745562645e-05,
+      "loss": 1.2193,
+      "step": 4405
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.941038155037124e-05,
+      "loss": 1.271,
+      "step": 4410
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.919200845695555e-05,
+      "loss": 1.2371,
+      "step": 4415
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.897363921678175e-05,
+      "loss": 1.1791,
+      "step": 4420
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.875527487123397e-05,
+      "loss": 1.2047,
+      "step": 4425
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.853691646167285e-05,
+      "loss": 1.2166,
+      "step": 4430
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.83185650294308e-05,
+      "loss": 1.2113,
+      "step": 4435
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.810022161580699e-05,
+      "loss": 1.216,
+      "step": 4440
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.788188726206226e-05,
+      "loss": 1.2434,
+      "step": 4445
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.766356300941439e-05,
+      "loss": 1.279,
+      "step": 4450
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.744524989903278e-05,
+      "loss": 1.2606,
+      "step": 4455
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.722694897203387e-05,
+      "loss": 1.2013,
+      "step": 4460
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.70086612694759e-05,
+      "loss": 1.2099,
+      "step": 4465
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.679038783235416e-05,
+      "loss": 1.2342,
+      "step": 4470
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.657212970159573e-05,
+      "loss": 1.191,
+      "step": 4475
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.635388791805481e-05,
+      "loss": 1.195,
+      "step": 4480
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.61356635225077e-05,
+      "loss": 1.2266,
+      "step": 4485
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.591745755564761e-05,
+      "loss": 1.2287,
+      "step": 4490
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.569927105808005e-05,
+      "loss": 1.2673,
+      "step": 4495
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.54811050703175e-05,
+      "loss": 1.2062,
+      "step": 4500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.52629606327748e-05,
+      "loss": 1.2824,
+      "step": 4505
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.504483878576386e-05,
+      "loss": 1.2288,
+      "step": 4510
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.482674056948902e-05,
+      "loss": 1.2111,
+      "step": 4515
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.460866702404178e-05,
+      "loss": 1.2076,
+      "step": 4520
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.439061918939615e-05,
+      "loss": 1.1897,
+      "step": 4525
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.417259810540332e-05,
+      "loss": 1.1975,
+      "step": 4530
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.395460481178705e-05,
+      "loss": 1.2569,
+      "step": 4535
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.373664034813858e-05,
+      "loss": 1.2109,
+      "step": 4540
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.351870575391154e-05,
+      "loss": 1.2787,
+      "step": 4545
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.33008020684173e-05,
+      "loss": 1.2448,
+      "step": 4550
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.308293033081964e-05,
+      "loss": 1.2316,
+      "step": 4555
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.286509158013012e-05,
+      "loss": 1.2503,
+      "step": 4560
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.264728685520289e-05,
+      "loss": 1.2337,
+      "step": 4565
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.242951719472995e-05,
+      "loss": 1.2349,
+      "step": 4570
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.221178363723595e-05,
+      "loss": 1.2612,
+      "step": 4575
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.199408722107342e-05,
+      "loss": 1.2396,
+      "step": 4580
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.177642898441784e-05,
+      "loss": 1.1769,
+      "step": 4585
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.155880996526248e-05,
+      "loss": 1.2807,
+      "step": 4590
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.134123120141373e-05,
+      "loss": 1.2658,
+      "step": 4595
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.112369373048587e-05,
+      "loss": 1.222,
+      "step": 4600
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.090619858989636e-05,
+      "loss": 1.212,
+      "step": 4605
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.068874681686068e-05,
+      "loss": 1.2026,
+      "step": 4610
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.047133944838766e-05,
+      "loss": 1.2289,
+      "step": 4615
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.025397752127424e-05,
+      "loss": 1.2567,
+      "step": 4620
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.003666207210064e-05,
+      "loss": 1.1785,
+      "step": 4625
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.981939413722553e-05,
+      "loss": 1.2138,
+      "step": 4630
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.960217475278089e-05,
+      "loss": 1.2448,
+      "step": 4635
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.938500495466729e-05,
+      "loss": 1.182,
+      "step": 4640
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.916788577854861e-05,
+      "loss": 1.2004,
+      "step": 4645
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.895081825984762e-05,
+      "loss": 1.2413,
+      "step": 4650
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.873380343374047e-05,
+      "loss": 1.2421,
+      "step": 4655
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.851684233515217e-05,
+      "loss": 1.236,
+      "step": 4660
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.829993599875149e-05,
+      "loss": 1.2721,
+      "step": 4665
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.808308545894595e-05,
+      "loss": 1.1742,
+      "step": 4670
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.786629174987713e-05,
+      "loss": 1.2497,
+      "step": 4675
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.764955590541546e-05,
+      "loss": 1.2308,
+      "step": 4680
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.743287895915553e-05,
+      "loss": 1.2241,
+      "step": 4685
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.721626194441093e-05,
+      "loss": 1.2789,
+      "step": 4690
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.699970589420957e-05,
+      "loss": 1.2332,
+      "step": 4695
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.678321184128847e-05,
+      "loss": 1.2229,
+      "step": 4700
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.656678081808915e-05,
+      "loss": 1.2342,
+      "step": 4705
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.635041385675248e-05,
+      "loss": 1.2235,
+      "step": 4710
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.613411198911374e-05,
+      "loss": 1.1992,
+      "step": 4715
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.591787624669799e-05,
+      "loss": 1.2564,
+      "step": 4720
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.570170766071468e-05,
+      "loss": 1.2247,
+      "step": 4725
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.548560726205325e-05,
+      "loss": 1.3024,
+      "step": 4730
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.526957608127776e-05,
+      "loss": 1.1935,
+      "step": 4735
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.505361514862235e-05,
+      "loss": 1.2182,
+      "step": 4740
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.483772549398595e-05,
+      "loss": 1.263,
+      "step": 4745
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.46219081469278e-05,
+      "loss": 1.1936,
+      "step": 4750
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.440616413666215e-05,
+      "loss": 1.2138,
+      "step": 4755
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.419049449205351e-05,
+      "loss": 1.252,
+      "step": 4760
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.397490024161191e-05,
+      "loss": 1.2403,
+      "step": 4765
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.37593824134876e-05,
+      "loss": 1.2067,
+      "step": 4770
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.354394203546658e-05,
+      "loss": 1.2379,
+      "step": 4775
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.332858013496536e-05,
+      "loss": 1.1946,
+      "step": 4780
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.31132977390263e-05,
+      "loss": 1.2411,
+      "step": 4785
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.289809587431252e-05,
+      "loss": 1.2003,
+      "step": 4790
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.26829755671032e-05,
+      "loss": 1.219,
+      "step": 4795
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.246793784328843e-05,
+      "loss": 1.2263,
+      "step": 4800
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.225298372836458e-05,
+      "loss": 1.2,
+      "step": 4805
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.203811424742934e-05,
+      "loss": 1.2614,
+      "step": 4810
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.182333042517657e-05,
+      "loss": 1.1986,
+      "step": 4815
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.16086332858919e-05,
+      "loss": 1.1986,
+      "step": 4820
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.139402385344739e-05,
+      "loss": 1.202,
+      "step": 4825
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.117950315129694e-05,
+      "loss": 1.277,
+      "step": 4830
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.096507220247122e-05,
+      "loss": 1.3144,
+      "step": 4835
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.075073202957297e-05,
+      "loss": 1.2113,
+      "step": 4840
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.053648365477192e-05,
+      "loss": 1.2616,
+      "step": 4845
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.032232809980006e-05,
+      "loss": 1.2243,
+      "step": 4850
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.01082663859468e-05,
+      "loss": 1.2503,
+      "step": 4855
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.989429953405391e-05,
+      "loss": 1.2659,
+      "step": 4860
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.968042856451093e-05,
+      "loss": 1.2506,
+      "step": 4865
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.946665449724992e-05,
+      "loss": 1.2295,
+      "step": 4870
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.925297835174107e-05,
+      "loss": 1.211,
+      "step": 4875
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.90394011469874e-05,
+      "loss": 1.2293,
+      "step": 4880
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.882592390152022e-05,
+      "loss": 1.2292,
+      "step": 4885
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.861254763339405e-05,
+      "loss": 1.3234,
+      "step": 4890
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.839927336018187e-05,
+      "loss": 1.2277,
+      "step": 4895
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.818610209897031e-05,
+      "loss": 1.2077,
+      "step": 4900
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.79730348663547e-05,
+      "loss": 1.2091,
+      "step": 4905
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.776007267843432e-05,
+      "loss": 1.2563,
+      "step": 4910
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.754721655080735e-05,
+      "loss": 1.2012,
+      "step": 4915
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.733446749856643e-05,
+      "loss": 1.223,
+      "step": 4920
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.712182653629332e-05,
+      "loss": 1.1959,
+      "step": 4925
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.690929467805445e-05,
+      "loss": 1.2364,
+      "step": 4930
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.669687293739594e-05,
+      "loss": 1.1937,
+      "step": 4935
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.648456232733865e-05,
+      "loss": 1.2006,
+      "step": 4940
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.627236386037363e-05,
+      "loss": 1.2162,
+      "step": 4945
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.606027854845696e-05,
+      "loss": 1.1784,
+      "step": 4950
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.584830740300528e-05,
+      "loss": 1.2595,
+      "step": 4955
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.563645143489063e-05,
+      "loss": 1.2114,
+      "step": 4960
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.542471165443589e-05,
+      "loss": 1.2162,
+      "step": 4965
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.52130890714097e-05,
+      "loss": 1.2254,
+      "step": 4970
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.500158469502197e-05,
+      "loss": 1.2581,
+      "step": 4975
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.479019953391879e-05,
+      "loss": 1.2167,
+      "step": 4980
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.457893459617769e-05,
+      "loss": 1.221,
+      "step": 4985
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.436779088930302e-05,
+      "loss": 1.2078,
+      "step": 4990
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.41567694202208e-05,
+      "loss": 1.2304,
+      "step": 4995
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.394587119527427e-05,
+      "loss": 1.2779,
+      "step": 5000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.37350972202188e-05,
+      "loss": 1.2631,
+      "step": 5005
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.352444850021735e-05,
+      "loss": 1.2075,
+      "step": 5010
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.331392603983538e-05,
+      "loss": 1.1994,
+      "step": 5015
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.310353084303643e-05,
+      "loss": 1.2066,
+      "step": 5020
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.289326391317697e-05,
+      "loss": 1.2265,
+      "step": 5025
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.268312625300186e-05,
+      "loss": 1.2483,
+      "step": 5030
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.24731188646395e-05,
+      "loss": 1.2574,
+      "step": 5035
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.226324274959691e-05,
+      "loss": 1.2036,
+      "step": 5040
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.20534989087552e-05,
+      "loss": 1.2612,
+      "step": 5045
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.184388834236464e-05,
+      "loss": 1.1996,
+      "step": 5050
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.163441205003997e-05,
+      "loss": 1.2041,
+      "step": 5055
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.142507103075548e-05,
+      "loss": 1.233,
+      "step": 5060
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.121586628284048e-05,
+      "loss": 1.1625,
+      "step": 5065
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.100679880397429e-05,
+      "loss": 1.2342,
+      "step": 5070
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.07978695911817e-05,
+      "loss": 1.2235,
+      "step": 5075
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.058907964082812e-05,
+      "loss": 1.2313,
+      "step": 5080
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.038042994861472e-05,
+      "loss": 1.1874,
+      "step": 5085
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.017192150957397e-05,
+      "loss": 1.2139,
+      "step": 5090
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.996355531806452e-05,
+      "loss": 1.2313,
+      "step": 5095
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.97553323677668e-05,
+      "loss": 1.2437,
+      "step": 5100
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.954725365167807e-05,
+      "loss": 1.2373,
+      "step": 5105
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.93393201621078e-05,
+      "loss": 1.227,
+      "step": 5110
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.913153289067278e-05,
+      "loss": 1.2191,
+      "step": 5115
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.892389282829265e-05,
+      "loss": 1.2265,
+      "step": 5120
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.871640096518491e-05,
+      "loss": 1.2349,
+      "step": 5125
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.85090582908603e-05,
+      "loss": 1.2107,
+      "step": 5130
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.830186579411822e-05,
+      "loss": 1.2257,
+      "step": 5135
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.809482446304172e-05,
+      "loss": 1.2675,
+      "step": 5140
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.788793528499307e-05,
+      "loss": 1.2171,
+      "step": 5145
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.76811992466089e-05,
+      "loss": 1.2494,
+      "step": 5150
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.747461733379552e-05,
+      "loss": 1.2435,
+      "step": 5155
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.726819053172423e-05,
+      "loss": 1.2341,
+      "step": 5160
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.70619198248267e-05,
+      "loss": 1.1986,
+      "step": 5165
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.685580619679005e-05,
+      "loss": 1.248,
+      "step": 5170
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.664985063055238e-05,
+      "loss": 1.2279,
+      "step": 5175
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.644405410829806e-05,
+      "loss": 1.2744,
+      "step": 5180
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.623841761145284e-05,
+      "loss": 1.2651,
+      "step": 5185
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.603294212067955e-05,
+      "loss": 1.2297,
+      "step": 5190
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.582762861587294e-05,
+      "loss": 1.2235,
+      "step": 5195
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.562247807615549e-05,
+      "loss": 1.2332,
+      "step": 5200
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.541749147987234e-05,
+      "loss": 1.2057,
+      "step": 5205
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.521266980458696e-05,
+      "loss": 1.254,
+      "step": 5210
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.500801402707617e-05,
+      "loss": 1.2466,
+      "step": 5215
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.480352512332569e-05,
+      "loss": 1.1983,
+      "step": 5220
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.459920406852552e-05,
+      "loss": 1.2202,
+      "step": 5225
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.439505183706505e-05,
+      "loss": 1.274,
+      "step": 5230
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.419106940252872e-05,
+      "loss": 1.2028,
+      "step": 5235
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.39872577376911e-05,
+      "loss": 1.2583,
+      "step": 5240
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.378361781451245e-05,
+      "loss": 1.2148,
+      "step": 5245
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.358015060413398e-05,
+      "loss": 1.1955,
+      "step": 5250
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.337685707687327e-05,
+      "loss": 1.2277,
+      "step": 5255
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.31737382022196e-05,
+      "loss": 1.251,
+      "step": 5260
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.297079494882927e-05,
+      "loss": 1.2749,
+      "step": 5265
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.276802828452123e-05,
+      "loss": 1.2174,
+      "step": 5270
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.256543917627214e-05,
+      "loss": 1.2224,
+      "step": 5275
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.236302859021205e-05,
+      "loss": 1.1853,
+      "step": 5280
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.216079749161946e-05,
+      "loss": 1.2123,
+      "step": 5285
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.195874684491709e-05,
+      "loss": 1.2359,
+      "step": 5290
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.175687761366702e-05,
+      "loss": 1.2675,
+      "step": 5295
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.155519076056619e-05,
+      "loss": 1.2272,
+      "step": 5300
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.13536872474418e-05,
+      "loss": 1.262,
+      "step": 5305
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.115236803524666e-05,
+      "loss": 1.1697,
+      "step": 5310
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.0951234084054765e-05,
+      "loss": 1.2045,
+      "step": 5315
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.075028635305653e-05,
+      "loss": 1.2479,
+      "step": 5320
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.054952580055439e-05,
+      "loss": 1.171,
+      "step": 5325
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.034895338395801e-05,
+      "loss": 1.1941,
+      "step": 5330
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.014857005978003e-05,
+      "loss": 1.1904,
+      "step": 5335
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.994837678363112e-05,
+      "loss": 1.1995,
+      "step": 5340
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.97483745102158e-05,
+      "loss": 1.2285,
+      "step": 5345
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.9548564193327636e-05,
+      "loss": 1.2381,
+      "step": 5350
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.9348946785844727e-05,
+      "loss": 1.2373,
+      "step": 5355
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.914952323972536e-05,
+      "loss": 1.2439,
+      "step": 5360
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.8950294506003104e-05,
+      "loss": 1.2676,
+      "step": 5365
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.8751261534782655e-05,
+      "loss": 1.2708,
+      "step": 5370
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.8552425275235035e-05,
+      "loss": 1.2583,
+      "step": 5375
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.835378667559323e-05,
+      "loss": 1.2292,
+      "step": 5380
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.8155346683147584e-05,
+      "loss": 1.196,
+      "step": 5385
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.7957106244241263e-05,
+      "loss": 1.1872,
+      "step": 5390
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.775906630426581e-05,
+      "loss": 1.2128,
+      "step": 5395
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.75612278076566e-05,
+      "loss": 1.2186,
+      "step": 5400
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.7363591697888365e-05,
+      "loss": 1.2266,
+      "step": 5405
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.716615891747057e-05,
+      "loss": 1.2569,
+      "step": 5410
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.696893040794327e-05,
+      "loss": 1.2047,
+      "step": 5415
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.6771907109872015e-05,
+      "loss": 1.2267,
+      "step": 5420
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.657508996284402e-05,
+      "loss": 1.2933,
+      "step": 5425
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.637847990546323e-05,
+      "loss": 1.2206,
+      "step": 5430
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.618207787534606e-05,
+      "loss": 1.2631,
+      "step": 5435
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.59858848091168e-05,
+      "loss": 1.2402,
+      "step": 5440
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.5789901642403274e-05,
+      "loss": 1.1605,
+      "step": 5445
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.5594129309832275e-05,
+      "loss": 1.276,
+      "step": 5450
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.539856874502512e-05,
+      "loss": 1.2453,
+      "step": 5455
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.520322088059334e-05,
+      "loss": 1.2295,
+      "step": 5460
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.500808664813399e-05,
+      "loss": 1.2992,
+      "step": 5465
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.481316697822538e-05,
+      "loss": 1.2052,
+      "step": 5470
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.461846280042261e-05,
+      "loss": 1.2013,
+      "step": 5475
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.442397504325307e-05,
+      "loss": 1.2624,
+      "step": 5480
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.422970463421204e-05,
+      "loss": 1.2076,
+      "step": 5485
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.403565249975848e-05,
+      "loss": 1.2712,
+      "step": 5490
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.3841819565310116e-05,
+      "loss": 1.2209,
+      "step": 5495
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.3648206755239494e-05,
+      "loss": 1.2387,
+      "step": 5500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.345481499286948e-05,
+      "loss": 1.1784,
+      "step": 5505
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.326164520046864e-05,
+      "loss": 1.241,
+      "step": 5510
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.306869829924705e-05,
+      "loss": 1.2578,
+      "step": 5515
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.28759752093518e-05,
+      "loss": 1.2305,
+      "step": 5520
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.26834768498627e-05,
+      "loss": 1.1786,
+      "step": 5525
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.2491204138787745e-05,
+      "loss": 1.2353,
+      "step": 5530
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.2299157993058986e-05,
+      "loss": 1.2378,
+      "step": 5535
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.210733932852792e-05,
+      "loss": 1.1858,
+      "step": 5540
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.191574905996106e-05,
+      "loss": 1.2209,
+      "step": 5545
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.172438810103599e-05,
+      "loss": 1.1974,
+      "step": 5550
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.153325736433657e-05,
+      "loss": 1.2649,
+      "step": 5555
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.134235776134878e-05,
+      "loss": 1.2664,
+      "step": 5560
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.115169020245629e-05,
+      "loss": 1.2476,
+      "step": 5565
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.0961255596936366e-05,
+      "loss": 1.2234,
+      "step": 5570
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.0771054852955036e-05,
+      "loss": 1.2379,
+      "step": 5575
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.058108887756335e-05,
+      "loss": 1.2377,
+      "step": 5580
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.039135857669265e-05,
+      "loss": 1.2931,
+      "step": 5585
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.020186485515024e-05,
+      "loss": 1.2326,
+      "step": 5590
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.001260861661542e-05,
+      "loss": 1.195,
+      "step": 5595
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.982359076363483e-05,
+      "loss": 1.1842,
+      "step": 5600
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.963481219761831e-05,
+      "loss": 1.1949,
+      "step": 5605
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.9446273818834476e-05,
+      "loss": 1.1958,
+      "step": 5610
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.9257976526406714e-05,
+      "loss": 1.2433,
+      "step": 5615
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.9069921218308404e-05,
+      "loss": 1.201,
+      "step": 5620
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.888210879135917e-05,
+      "loss": 1.1987,
+      "step": 5625
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.869454014122025e-05,
+      "loss": 1.266,
+      "step": 5630
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.850721616239034e-05,
+      "loss": 1.2765,
+      "step": 5635
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.83201377482013e-05,
+      "loss": 1.234,
+      "step": 5640
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8133305790813964e-05,
+      "loss": 1.2165,
+      "step": 5645
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7946721181213806e-05,
+      "loss": 1.1788,
+      "step": 5650
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.776038480920667e-05,
+      "loss": 1.2054,
+      "step": 5655
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.757429756341475e-05,
+      "loss": 1.2462,
+      "step": 5660
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.73884603312719e-05,
+      "loss": 1.2817,
+      "step": 5665
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.720287399901996e-05,
+      "loss": 1.1914,
+      "step": 5670
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.701753945170407e-05,
+      "loss": 1.2526,
+      "step": 5675
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.683245757316872e-05,
+      "loss": 1.2235,
+      "step": 5680
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.664762924605339e-05,
+      "loss": 1.2289,
+      "step": 5685
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.646305535178841e-05,
+      "loss": 1.2411,
+      "step": 5690
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.627873677059076e-05,
+      "loss": 1.2382,
+      "step": 5695
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6094674381459804e-05,
+      "loss": 1.2224,
+      "step": 5700
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.591086906217326e-05,
+      "loss": 1.2199,
+      "step": 5705
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.5727321689282774e-05,
+      "loss": 1.2055,
+      "step": 5710
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.5544033138109955e-05,
+      "loss": 1.1955,
+      "step": 5715
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.536100428274205e-05,
+      "loss": 1.3198,
+      "step": 5720
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.517823599602788e-05,
+      "loss": 1.2095,
+      "step": 5725
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.499572914957364e-05,
+      "loss": 1.2287,
+      "step": 5730
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.481348461373868e-05,
+      "loss": 1.2672,
+      "step": 5735
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.463150325763156e-05,
+      "loss": 1.2442,
+      "step": 5740
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.4449785949105515e-05,
+      "loss": 1.249,
+      "step": 5745
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.426833355475482e-05,
+      "loss": 1.175,
+      "step": 5750
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.408714693991023e-05,
+      "loss": 1.2178,
+      "step": 5755
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.390622696863509e-05,
+      "loss": 1.2485,
+      "step": 5760
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.37255745037211e-05,
+      "loss": 1.2375,
+      "step": 5765
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.354519040668428e-05,
+      "loss": 1.2551,
+      "step": 5770
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.336507553776082e-05,
+      "loss": 1.1726,
+      "step": 5775
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.318523075590295e-05,
+      "loss": 1.1907,
+      "step": 5780
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.3005656918775005e-05,
+      "loss": 1.2666,
+      "step": 5785
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.2826354882748974e-05,
+      "loss": 1.2873,
+      "step": 5790
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.264732550290089e-05,
+      "loss": 1.1883,
+      "step": 5795
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.2468569633006394e-05,
+      "loss": 1.3305,
+      "step": 5800
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.229008812553678e-05,
+      "loss": 1.2694,
+      "step": 5805
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.211188183165494e-05,
+      "loss": 1.1939,
+      "step": 5810
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.193395160121133e-05,
+      "loss": 1.2648,
+      "step": 5815
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.175629828273985e-05,
+      "loss": 1.2239,
+      "step": 5820
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.15789227234538e-05,
+      "loss": 1.1807,
+      "step": 5825
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.1401825769242006e-05,
+      "loss": 1.2455,
+      "step": 5830
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.122500826466443e-05,
+      "loss": 1.2771,
+      "step": 5835
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.10484710529486e-05,
+      "loss": 1.3095,
+      "step": 5840
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.087221497598518e-05,
+      "loss": 1.2266,
+      "step": 5845
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.06962408743242e-05,
+      "loss": 1.2451,
+      "step": 5850
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.052054958717095e-05,
+      "loss": 1.2398,
+      "step": 5855
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.0345141952382026e-05,
+      "loss": 1.1949,
+      "step": 5860
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.017001880646129e-05,
+      "loss": 1.2503,
+      "step": 5865
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.999518098455586e-05,
+      "loss": 1.2361,
+      "step": 5870
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.98206293204523e-05,
+      "loss": 1.2436,
+      "step": 5875
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.964636464657238e-05,
+      "loss": 1.1848,
+      "step": 5880
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.947238779396929e-05,
+      "loss": 1.2107,
+      "step": 5885
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.929869959232363e-05,
+      "loss": 1.1979,
+      "step": 5890
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.912530086993944e-05,
+      "loss": 1.2726,
+      "step": 5895
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.895219245374024e-05,
+      "loss": 1.2462,
+      "step": 5900
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.877937516926511e-05,
+      "loss": 1.2945,
+      "step": 5905
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.860684984066485e-05,
+      "loss": 1.2304,
+      "step": 5910
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.843461729069773e-05,
+      "loss": 1.1819,
+      "step": 5915
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.8262678340726e-05,
+      "loss": 1.2471,
+      "step": 5920
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.8091033810711655e-05,
+      "loss": 1.2409,
+      "step": 5925
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.791968451921263e-05,
+      "loss": 1.265,
+      "step": 5930
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.774863128337887e-05,
+      "loss": 1.1473,
+      "step": 5935
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7577874918948586e-05,
+      "loss": 1.2493,
+      "step": 5940
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.740741624024405e-05,
+      "loss": 1.2343,
+      "step": 5945
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.723725606016799e-05,
+      "loss": 1.2109,
+      "step": 5950
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.706739519019969e-05,
+      "loss": 1.1899,
+      "step": 5955
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.689783444039088e-05,
+      "loss": 1.2098,
+      "step": 5960
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.67285746193622e-05,
+      "loss": 1.1938,
+      "step": 5965
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.655961653429911e-05,
+      "loss": 1.2459,
+      "step": 5970
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.639096099094813e-05,
+      "loss": 1.1726,
+      "step": 5975
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.6222608793612954e-05,
+      "loss": 1.2611,
+      "step": 5980
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.605456074515078e-05,
+      "loss": 1.2847,
+      "step": 5985
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.5886817646968105e-05,
+      "loss": 1.2001,
+      "step": 5990
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.571938029901739e-05,
+      "loss": 1.2373,
+      "step": 5995
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.555224949979288e-05,
+      "loss": 1.2169,
+      "step": 6000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.538542604632684e-05,
+      "loss": 1.2389,
+      "step": 6005
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.521891073418602e-05,
+      "loss": 1.2217,
+      "step": 6010
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.505270435746752e-05,
+      "loss": 1.2399,
+      "step": 6015
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.488680770879523e-05,
+      "loss": 1.275,
+      "step": 6020
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.4721221579315885e-05,
+      "loss": 1.2671,
+      "step": 6025
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.4555946758695554e-05,
+      "loss": 1.2226,
+      "step": 6030
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.439098403511546e-05,
+      "loss": 1.1972,
+      "step": 6035
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.422633419526868e-05,
+      "loss": 1.2337,
+      "step": 6040
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.4061998024356054e-05,
+      "loss": 1.2255,
+      "step": 6045
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.3897976306082614e-05,
+      "loss": 1.2221,
+      "step": 6050
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.3734269822653755e-05,
+      "loss": 1.1575,
+      "step": 6055
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.3570879354771554e-05,
+      "loss": 1.1745,
+      "step": 6060
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.340780568163108e-05,
+      "loss": 1.2404,
+      "step": 6065
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.3245049580916546e-05,
+      "loss": 1.2305,
+      "step": 6070
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.3082611828797816e-05,
+      "loss": 1.2651,
+      "step": 6075
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.292049319992647e-05,
+      "loss": 1.2278,
+      "step": 6080
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.275869446743227e-05,
+      "loss": 1.2184,
+      "step": 6085
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.259721640291938e-05,
+      "loss": 1.1887,
+      "step": 6090
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.243605977646278e-05,
+      "loss": 1.2237,
+      "step": 6095
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.227522535660449e-05,
+      "loss": 1.2071,
+      "step": 6100
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.211471391034992e-05,
+      "loss": 1.2189,
+      "step": 6105
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.195452620316445e-05,
+      "loss": 1.231,
+      "step": 6110
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.179466299896926e-05,
+      "loss": 1.2322,
+      "step": 6115
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.163512506013827e-05,
+      "loss": 1.1895,
+      "step": 6120
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.1475913147494115e-05,
+      "loss": 1.2462,
+      "step": 6125
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.131702802030466e-05,
+      "loss": 1.2451,
+      "step": 6130
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.1158470436279364e-05,
+      "loss": 1.2167,
+      "step": 6135
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.1000241151565643e-05,
+      "loss": 1.2649,
+      "step": 6140
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.0842340920745316e-05,
+      "loss": 1.2931,
+      "step": 6145
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.068477049683091e-05,
+      "loss": 1.2572,
+      "step": 6150
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.052753063126227e-05,
+      "loss": 1.2199,
+      "step": 6155
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.03706220739026e-05,
+      "loss": 1.2612,
+      "step": 6160
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.021404557303539e-05,
+      "loss": 1.2165,
+      "step": 6165
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.0057801875360404e-05,
+      "loss": 1.2521,
+      "step": 6170
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.990189172599035e-05,
+      "loss": 1.1771,
+      "step": 6175
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.9746315868447273e-05,
+      "loss": 1.212,
+      "step": 6180
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.9591075044659e-05,
+      "loss": 1.2577,
+      "step": 6185
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.9436169994955643e-05,
+      "loss": 1.2378,
+      "step": 6190
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.9281601458065945e-05,
+      "loss": 1.1987,
+      "step": 6195
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.912737017111403e-05,
+      "loss": 1.2142,
+      "step": 6200
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8973476869615466e-05,
+      "loss": 1.2517,
+      "step": 6205
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8819922287474222e-05,
+      "loss": 1.1987,
+      "step": 6210
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.866670715697879e-05,
+      "loss": 1.2858,
+      "step": 6215
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.851383220879892e-05,
+      "loss": 1.2256,
+      "step": 6220
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.836129817198203e-05,
+      "loss": 1.2316,
+      "step": 6225
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8209105773949783e-05,
+      "loss": 1.2449,
+      "step": 6230
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8057255740494593e-05,
+      "loss": 1.2279,
+      "step": 6235
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.790574879577612e-05,
+      "loss": 1.2536,
+      "step": 6240
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7754585662317968e-05,
+      "loss": 1.1736,
+      "step": 6245
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7603767061004026e-05,
+      "loss": 1.2252,
+      "step": 6250
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7453293711075212e-05,
+      "loss": 1.2421,
+      "step": 6255
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7303166330125918e-05,
+      "loss": 1.2152,
+      "step": 6260
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.715338563410068e-05,
+      "loss": 1.2339,
+      "step": 6265
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7003952337290694e-05,
+      "loss": 1.2379,
+      "step": 6270
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.685486715233043e-05,
+      "loss": 1.2184,
+      "step": 6275
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.670613079019436e-05,
+      "loss": 1.2403,
+      "step": 6280
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.6557743960193214e-05,
+      "loss": 1.2331,
+      "step": 6285
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.6409707369971094e-05,
+      "loss": 1.2242,
+      "step": 6290
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.6262021725501674e-05,
+      "loss": 1.1681,
+      "step": 6295
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.611468773108503e-05,
+      "loss": 1.2578,
+      "step": 6300
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.5967706089344258e-05,
+      "loss": 1.239,
+      "step": 6305
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.582107750122219e-05,
+      "loss": 1.2168,
+      "step": 6310
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.5674802665977814e-05,
+      "loss": 1.2664,
+      "step": 6315
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.5528882281183208e-05,
+      "loss": 1.1996,
+      "step": 6320
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.5383317042720166e-05,
+      "loss": 1.1681,
+      "step": 6325
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.5238107644776642e-05,
+      "loss": 1.1917,
+      "step": 6330
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.509325477984379e-05,
+      "loss": 1.2144,
+      "step": 6335
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4948759138712384e-05,
+      "loss": 1.2315,
+      "step": 6340
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4804621410469687e-05,
+      "loss": 1.2261,
+      "step": 6345
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4660842282496033e-05,
+      "loss": 1.1728,
+      "step": 6350
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.4517422440461758e-05,
+      "loss": 1.1741,
+      "step": 6355
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.4374362568323638e-05,
+      "loss": 1.1539,
+      "step": 6360
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.423166334832183e-05,
+      "loss": 1.1881,
+      "step": 6365
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.4089325460976707e-05,
+      "loss": 1.1905,
+      "step": 6370
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.394734958508523e-05,
+      "loss": 1.1682,
+      "step": 6375
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.380573639771819e-05,
+      "loss": 1.2683,
+      "step": 6380
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3664486574216627e-05,
+      "loss": 1.2757,
+      "step": 6385
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3523600788188748e-05,
+      "loss": 1.169,
+      "step": 6390
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3383079711506682e-05,
+      "loss": 1.2452,
+      "step": 6395
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.324292401430339e-05,
+      "loss": 1.1997,
+      "step": 6400
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3103134364969213e-05,
+      "loss": 1.2237,
+      "step": 6405
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2963711430148926e-05,
+      "loss": 1.1717,
+      "step": 6410
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.282465587473852e-05,
+      "loss": 1.2149,
+      "step": 6415
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2685968361881903e-05,
+      "loss": 1.2932,
+      "step": 6420
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2547649552967863e-05,
+      "loss": 1.157,
+      "step": 6425
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2409700107626842e-05,
+      "loss": 1.2471,
+      "step": 6430
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.227212068372785e-05,
+      "loss": 1.295,
+      "step": 6435
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.2134911937375246e-05,
+      "loss": 1.2177,
+      "step": 6440
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1998074522905765e-05,
+      "loss": 1.219,
+      "step": 6445
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1861609092885205e-05,
+      "loss": 1.3656,
+      "step": 6450
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1725516298105352e-05,
+      "loss": 1.2052,
+      "step": 6455
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1589796787581074e-05,
+      "loss": 1.2334,
+      "step": 6460
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1454451208546978e-05,
+      "loss": 1.2087,
+      "step": 6465
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1319480206454435e-05,
+      "loss": 1.2806,
+      "step": 6470
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.118488442496851e-05,
+      "loss": 1.223,
+      "step": 6475
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1050664505964868e-05,
+      "loss": 1.2361,
+      "step": 6480
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0916821089526683e-05,
+      "loss": 1.1873,
+      "step": 6485
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0783354813941714e-05,
+      "loss": 1.2525,
+      "step": 6490
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0650266315699074e-05,
+      "loss": 1.1974,
+      "step": 6495
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0517556229486322e-05,
+      "loss": 1.2938,
+      "step": 6500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.038522518818641e-05,
+      "loss": 1.2446,
+      "step": 6505
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.025327382287464e-05,
+      "loss": 1.2366,
+      "step": 6510
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.0121702762815697e-05,
+      "loss": 1.183,
+      "step": 6515
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9990512635460567e-05,
+      "loss": 1.2404,
+      "step": 6520
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.985970406644373e-05,
+      "loss": 1.176,
+      "step": 6525
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9729277679579837e-05,
+      "loss": 1.2065,
+      "step": 6530
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9599234096861175e-05,
+      "loss": 1.2067,
+      "step": 6535
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.946957393845432e-05,
+      "loss": 1.276,
+      "step": 6540
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9340297822697372e-05,
+      "loss": 1.2085,
+      "step": 6545
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9211406366096974e-05,
+      "loss": 1.244,
+      "step": 6550
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.908290018332536e-05,
+      "loss": 1.2415,
+      "step": 6555
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.89547798872174e-05,
+      "loss": 1.2359,
+      "step": 6560
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8827046088767732e-05,
+      "loss": 1.2542,
+      "step": 6565
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8699699397127857e-05,
+      "loss": 1.2644,
+      "step": 6570
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8572740419603085e-05,
+      "loss": 1.1877,
+      "step": 6575
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8446169761649866e-05,
+      "loss": 1.2479,
+      "step": 6580
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.831998802687275e-05,
+      "loss": 1.2276,
+      "step": 6585
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.81941958170215e-05,
+      "loss": 1.223,
+      "step": 6590
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.8068793731988332e-05,
+      "loss": 1.2541,
+      "step": 6595
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7943782369804952e-05,
+      "loss": 1.2549,
+      "step": 6600
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.781916232663975e-05,
+      "loss": 1.2333,
+      "step": 6605
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7694934196794944e-05,
+      "loss": 1.2914,
+      "step": 6610
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7571098572703804e-05,
+      "loss": 1.2144,
+      "step": 6615
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7447656044927697e-05,
+      "loss": 1.2248,
+      "step": 6620
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7324607202153397e-05,
+      "loss": 1.2142,
+      "step": 6625
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.72019526311902e-05,
+      "loss": 1.2317,
+      "step": 6630
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7079692916967172e-05,
+      "loss": 1.2049,
+      "step": 6635
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.695782864253034e-05,
+      "loss": 1.1773,
+      "step": 6640
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6836360389039906e-05,
+      "loss": 1.1928,
+      "step": 6645
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6715288735767475e-05,
+      "loss": 1.1881,
+      "step": 6650
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6594614260093288e-05,
+      "loss": 1.1763,
+      "step": 6655
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.647433753750357e-05,
+      "loss": 1.2282,
+      "step": 6660
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6354459141587576e-05,
+      "loss": 1.2112,
+      "step": 6665
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6234979644035053e-05,
+      "loss": 1.2554,
+      "step": 6670
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6115899614633388e-05,
+      "loss": 1.2243,
+      "step": 6675
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.599721962126497e-05,
+      "loss": 1.247,
+      "step": 6680
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5878940229904448e-05,
+      "loss": 1.2524,
+      "step": 6685
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5761062004616e-05,
+      "loss": 1.2116,
+      "step": 6690
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5643585507550785e-05,
+      "loss": 1.205,
+      "step": 6695
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.552651129894397e-05,
+      "loss": 1.2597,
+      "step": 6700
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5409839937112414e-05,
+      "loss": 1.2377,
+      "step": 6705
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5293571978451772e-05,
+      "loss": 1.2583,
+      "step": 6710
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5177707977433876e-05,
+      "loss": 1.2339,
+      "step": 6715
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5062248486604125e-05,
+      "loss": 1.238,
+      "step": 6720
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.494719405657895e-05,
+      "loss": 1.136,
+      "step": 6725
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4832545236042916e-05,
+      "loss": 1.2502,
+      "step": 6730
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4718302571746345e-05,
+      "loss": 1.213,
+      "step": 6735
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4604466608502709e-05,
+      "loss": 1.2483,
+      "step": 6740
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4491037889185822e-05,
+      "loss": 1.2053,
+      "step": 6745
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4378016954727492e-05,
+      "loss": 1.1983,
+      "step": 6750
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4265404344114785e-05,
+      "loss": 1.217,
+      "step": 6755
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4153200594387527e-05,
+      "loss": 1.2548,
+      "step": 6760
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4041406240635691e-05,
+      "loss": 1.2645,
+      "step": 6765
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3930021815996963e-05,
+      "loss": 1.2537,
+      "step": 6770
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3819047851653988e-05,
+      "loss": 1.2306,
+      "step": 6775
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3708484876832006e-05,
+      "loss": 1.2329,
+      "step": 6780
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3598333418796349e-05,
+      "loss": 1.2378,
+      "step": 6785
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3488594002849809e-05,
+      "loss": 1.2038,
+      "step": 6790
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3379267152330165e-05,
+      "loss": 1.2732,
+      "step": 6795
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3270353388607726e-05,
+      "loss": 1.1857,
+      "step": 6800
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3161853231082855e-05,
+      "loss": 1.2093,
+      "step": 6805
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3053767197183387e-05,
+      "loss": 1.2439,
+      "step": 6810
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2946095802362369e-05,
+      "loss": 1.2162,
+      "step": 6815
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2838839560095317e-05,
+      "loss": 1.1833,
+      "step": 6820
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2731998981877991e-05,
+      "loss": 1.28,
+      "step": 6825
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2625574577223931e-05,
+      "loss": 1.1957,
+      "step": 6830
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2519566853661913e-05,
+      "loss": 1.2774,
+      "step": 6835
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2413976316733611e-05,
+      "loss": 1.1973,
+      "step": 6840
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2308803469991192e-05,
+      "loss": 1.2473,
+      "step": 6845
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2204048814994883e-05,
+      "loss": 1.2438,
+      "step": 6850
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2099712851310563e-05,
+      "loss": 1.2745,
+      "step": 6855
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1995796076507482e-05,
+      "loss": 1.2045,
+      "step": 6860
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1892298986155758e-05,
+      "loss": 1.2446,
+      "step": 6865
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1789222073824035e-05,
+      "loss": 1.1972,
+      "step": 6870
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1686565831077256e-05,
+      "loss": 1.1953,
+      "step": 6875
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.158433074747417e-05,
+      "loss": 1.2117,
+      "step": 6880
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1482517310565066e-05,
+      "loss": 1.2374,
+      "step": 6885
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1381126005889408e-05,
+      "loss": 1.1644,
+      "step": 6890
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1280157316973639e-05,
+      "loss": 1.2214,
+      "step": 6895
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1179611725328632e-05,
+      "loss": 1.1822,
+      "step": 6900
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.10794897104477e-05,
+      "loss": 1.2511,
+      "step": 6905
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0979791749804102e-05,
+      "loss": 1.2504,
+      "step": 6910
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.088051831884872e-05,
+      "loss": 1.1955,
+      "step": 6915
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0781669891008061e-05,
+      "loss": 1.2156,
+      "step": 6920
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.06832469376817e-05,
+      "loss": 1.2507,
+      "step": 6925
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0585249928240226e-05,
+      "loss": 1.2264,
+      "step": 6930
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0487679330022893e-05,
+      "loss": 1.2213,
+      "step": 6935
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0390535608335506e-05,
+      "loss": 1.163,
+      "step": 6940
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0293819226448031e-05,
+      "loss": 1.2332,
+      "step": 6945
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0197530645592579e-05,
+      "loss": 1.2193,
+      "step": 6950
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0101670324961098e-05,
+      "loss": 1.245,
+      "step": 6955
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.000623872170311e-05,
+      "loss": 1.1803,
+      "step": 6960
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.911236290923765e-06,
+      "loss": 1.2356,
+      "step": 6965
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.81666348568142e-06,
+      "loss": 1.1854,
+      "step": 6970
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.722520756985632e-06,
+      "loss": 1.257,
+      "step": 6975
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.628808553794922e-06,
+      "loss": 1.1904,
+      "step": 6980
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.535527323014748e-06,
+      "loss": 1.2469,
+      "step": 6985
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.442677509495179e-06,
+      "loss": 1.2247,
+      "step": 6990
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.350259556029017e-06,
+      "loss": 1.194,
+      "step": 6995
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.25827390334949e-06,
+      "loss": 1.2252,
+      "step": 7000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.16672099012822e-06,
+      "loss": 1.2245,
+      "step": 7005
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.075601252973153e-06,
+      "loss": 1.2702,
+      "step": 7010
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.984915126426452e-06,
+      "loss": 1.2296,
+      "step": 7015
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.894663042962426e-06,
+      "loss": 1.204,
+      "step": 7020
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.804845432985454e-06,
+      "loss": 1.2106,
+      "step": 7025
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.715462724827995e-06,
+      "loss": 1.2468,
+      "step": 7030
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.626515344748454e-06,
+      "loss": 1.2369,
+      "step": 7035
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.5380037169292e-06,
+      "loss": 1.2286,
+      "step": 7040
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.44992826347456e-06,
+      "loss": 1.2015,
+      "step": 7045
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.362289404408751e-06,
+      "loss": 1.206,
+      "step": 7050
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.275087557673934e-06,
+      "loss": 1.2186,
+      "step": 7055
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.188323139128174e-06,
+      "loss": 1.2939,
+      "step": 7060
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.101996562543546e-06,
+      "loss": 1.2394,
+      "step": 7065
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.01610823960397e-06,
+      "loss": 1.1787,
+      "step": 7070
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.930658579903483e-06,
+      "loss": 1.2321,
+      "step": 7075
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.845647990944138e-06,
+      "loss": 1.2155,
+      "step": 7080
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.761076878134076e-06,
+      "loss": 1.2654,
+      "step": 7085
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.676945644785638e-06,
+      "loss": 1.2361,
+      "step": 7090
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.593254692113449e-06,
+      "loss": 1.2317,
+      "step": 7095
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.510004419232386e-06,
+      "loss": 1.2133,
+      "step": 7100
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.427195223155825e-06,
+      "loss": 1.2085,
+      "step": 7105
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.3448274987937046e-06,
+      "loss": 1.2363,
+      "step": 7110
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.262901638950548e-06,
+      "loss": 1.2551,
+      "step": 7115
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.181418034323739e-06,
+      "loss": 1.2386,
+      "step": 7120
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.100377073501552e-06,
+      "loss": 1.2171,
+      "step": 7125
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.019779142961325e-06,
+      "loss": 1.247,
+      "step": 7130
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.939624627067609e-06,
+      "loss": 1.2677,
+      "step": 7135
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.859913908070426e-06,
+      "loss": 1.2385,
+      "step": 7140
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.780647366103276e-06,
+      "loss": 1.2148,
+      "step": 7145
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.701825379181437e-06,
+      "loss": 1.2399,
+      "step": 7150
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.62344832320021e-06,
+      "loss": 1.2374,
+      "step": 7155
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.545516571932964e-06,
+      "loss": 1.2319,
+      "step": 7160
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.4680304970295095e-06,
+      "loss": 1.1835,
+      "step": 7165
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.390990468014268e-06,
+      "loss": 1.244,
+      "step": 7170
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.314396852284454e-06,
+      "loss": 1.26,
+      "step": 7175
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.238250015108427e-06,
+      "loss": 1.2531,
+      "step": 7180
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.162550319623894e-06,
+      "loss": 1.2704,
+      "step": 7185
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.087298126836139e-06,
+      "loss": 1.223,
+      "step": 7190
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.01249379561637e-06,
+      "loss": 1.2113,
+      "step": 7195
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.938137682700018e-06,
+      "loss": 1.2136,
+      "step": 7200
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.8642301426849455e-06,
+      "loss": 1.2261,
+      "step": 7205
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.790771528029837e-06,
+      "loss": 1.1693,
+      "step": 7210
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.7177621890524894e-06,
+      "loss": 1.2483,
+      "step": 7215
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.645202473928113e-06,
+      "loss": 1.1596,
+      "step": 7220
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.573092728687734e-06,
+      "loss": 1.2331,
+      "step": 7225
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.501433297216529e-06,
+      "loss": 1.1981,
+      "step": 7230
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.430224521252136e-06,
+      "loss": 1.2312,
+      "step": 7235
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.359466740383012e-06,
+      "loss": 1.2067,
+      "step": 7240
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.2891602920469575e-06,
+      "loss": 1.2017,
+      "step": 7245
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.2193055115293375e-06,
+      "loss": 1.2237,
+      "step": 7250
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.149902731961565e-06,
+      "loss": 1.2578,
+      "step": 7255
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.080952284319496e-06,
+      "loss": 1.2231,
+      "step": 7260
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.012454497421903e-06,
+      "loss": 1.1905,
+      "step": 7265
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.944409697928775e-06,
+      "loss": 1.2489,
+      "step": 7270
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.876818210339906e-06,
+      "loss": 1.2039,
+      "step": 7275
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.809680356993273e-06,
+      "loss": 1.2106,
+      "step": 7280
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.742996458063476e-06,
+      "loss": 1.2391,
+      "step": 7285
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.676766831560297e-06,
+      "loss": 1.2515,
+      "step": 7290
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.610991793327102e-06,
+      "loss": 1.256,
+      "step": 7295
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.545671657039363e-06,
+      "loss": 1.2145,
+      "step": 7300
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.480806734203169e-06,
+      "loss": 1.1947,
+      "step": 7305
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.416397334153766e-06,
+      "loss": 1.2157,
+      "step": 7310
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.35244376405397e-06,
+      "loss": 1.2253,
+      "step": 7315
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.288946328892885e-06,
+      "loss": 1.2069,
+      "step": 7320
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.225905331484292e-06,
+      "loss": 1.2239,
+      "step": 7325
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.163321072465221e-06,
+      "loss": 1.1986,
+      "step": 7330
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.101193850294649e-06,
+      "loss": 1.2351,
+      "step": 7335
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.039523961251912e-06,
+      "loss": 1.217,
+      "step": 7340
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.978311699435399e-06,
+      "loss": 1.2028,
+      "step": 7345
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.917557356761081e-06,
+      "loss": 1.2869,
+      "step": 7350
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.857261222961217e-06,
+      "loss": 1.2012,
+      "step": 7355
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7974235855828156e-06,
+      "loss": 1.2492,
+      "step": 7360
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7380447299864342e-06,
+      "loss": 1.252,
+      "step": 7365
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6791249393446916e-06,
+      "loss": 1.2083,
+      "step": 7370
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6206644946409906e-06,
+      "loss": 1.2232,
+      "step": 7375
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.562663674668154e-06,
+      "loss": 1.2421,
+      "step": 7380
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5051227560270683e-06,
+      "loss": 1.2394,
+      "step": 7385
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4480420131254186e-06,
+      "loss": 1.2384,
+      "step": 7390
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.391421718176324e-06,
+      "loss": 1.191,
+      "step": 7395
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.335262141197104e-06,
+      "loss": 1.254,
+      "step": 7400
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.279563550007936e-06,
+      "loss": 1.2812,
+      "step": 7405
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.224326210230566e-06,
+      "loss": 1.1998,
+      "step": 7410
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.1695503852871234e-06,
+      "loss": 1.294,
+      "step": 7415
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.1152363363987856e-06,
+      "loss": 1.194,
+      "step": 7420
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.0613843225845486e-06,
+      "loss": 1.2232,
+      "step": 7425
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.0079946006600247e-06,
+      "loss": 1.1924,
+      "step": 7430
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.955067425236202e-06,
+      "loss": 1.2307,
+      "step": 7435
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9026030487181887e-06,
+      "loss": 1.2746,
+      "step": 7440
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8506017213040802e-06,
+      "loss": 1.2516,
+      "step": 7445
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7990636909837163e-06,
+      "loss": 1.2431,
+      "step": 7450
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7479892035375154e-06,
+      "loss": 1.2416,
+      "step": 7455
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6973785025352972e-06,
+      "loss": 1.2047,
+      "step": 7460
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6472318293351286e-06,
+      "loss": 1.2163,
+      "step": 7465
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5975494230821575e-06,
+      "loss": 1.213,
+      "step": 7470
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.548331520707492e-06,
+      "loss": 1.2552,
+      "step": 7475
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.4995783569270793e-06,
+      "loss": 1.2363,
+      "step": 7480
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.4512901642405163e-06,
+      "loss": 1.2183,
+      "step": 7485
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.4034671729300408e-06,
+      "loss": 1.2366,
+      "step": 7490
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.356109611059387e-06,
+      "loss": 1.2298,
+      "step": 7495
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.3092177044726437e-06,
+      "loss": 1.2116,
+      "step": 7500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2627916767932854e-06,
+      "loss": 1.2323,
+      "step": 7505
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.216831749423054e-06,
+      "loss": 1.2388,
+      "step": 7510
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1713381415408352e-06,
+      "loss": 1.2411,
+      "step": 7515
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1263110701017164e-06,
+      "loss": 1.1264,
+      "step": 7520
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0817507498359644e-06,
+      "loss": 1.2822,
+      "step": 7525
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0376573932478384e-06,
+      "loss": 1.2128,
+      "step": 7530
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9940312106147886e-06,
+      "loss": 1.2713,
+      "step": 7535
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9508724099863042e-06,
+      "loss": 1.298,
+      "step": 7540
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9081811971829787e-06,
+      "loss": 1.2,
+      "step": 7545
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8659577757955016e-06,
+      "loss": 1.1591,
+      "step": 7550
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8242023471837788e-06,
+      "loss": 1.2002,
+      "step": 7555
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7829151104757802e-06,
+      "loss": 1.1894,
+      "step": 7560
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7420962625668059e-06,
+      "loss": 1.2567,
+      "step": 7565
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7017459981184202e-06,
+      "loss": 1.2795,
+      "step": 7570
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6618645095575425e-06,
+      "loss": 1.2537,
+      "step": 7575
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6224519870755572e-06,
+      "loss": 1.2301,
+      "step": 7580
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5835086186273718e-06,
+      "loss": 1.2443,
+      "step": 7585
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5450345899305496e-06,
+      "loss": 1.2132,
+      "step": 7590
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5070300844644114e-06,
+      "loss": 1.2202,
+      "step": 7595
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4694952834691688e-06,
+      "loss": 1.2344,
+      "step": 7600
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4324303659450588e-06,
+      "loss": 1.219,
+      "step": 7605
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.395835508651444e-06,
+      "loss": 1.2289,
+      "step": 7610
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.359710886106058e-06,
+      "loss": 1.2259,
+      "step": 7615
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3240566705841062e-06,
+      "loss": 1.2471,
+      "step": 7620
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2888730321174657e-06,
+      "loss": 1.2772,
+      "step": 7625
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2541601384938651e-06,
+      "loss": 1.1938,
+      "step": 7630
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.2199181552561056e-06,
+      "loss": 1.2427,
+      "step": 7635
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1861472457012301e-06,
+      "loss": 1.2376,
+      "step": 7640
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1528475708798114e-06,
+      "loss": 1.2138,
+      "step": 7645
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1200192895951311e-06,
+      "loss": 1.2573,
+      "step": 7650
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0876625584024026e-06,
+      "loss": 1.1977,
+      "step": 7655
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.055777531608093e-06,
+      "loss": 1.165,
+      "step": 7660
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0243643612691478e-06,
+      "loss": 1.179,
+      "step": 7665
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.93423197192267e-07,
+      "loss": 1.1903,
+      "step": 7670
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.629541869331848e-07,
+      "loss": 1.207,
+      "step": 7675
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.329574757959813e-07,
+      "loss": 1.2175,
+      "step": 7680
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.034332068323825e-07,
+      "loss": 1.2146,
+      "step": 7685
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.743815208410833e-07,
+      "loss": 1.2578,
+      "step": 7690
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.458025563670702e-07,
+      "loss": 1.1793,
+      "step": 7695
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.176964497009332e-07,
+      "loss": 1.2312,
+      "step": 7700
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.900633348782882e-07,
+      "loss": 1.222,
+      "step": 7705
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.629033436790777e-07,
+      "loss": 1.1906,
+      "step": 7710
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.362166056269493e-07,
+      "loss": 1.2747,
+      "step": 7715
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.100032479886443e-07,
+      "loss": 1.2164,
+      "step": 7720
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.842633957734213e-07,
+      "loss": 1.2646,
+      "step": 7725
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.589971717323895e-07,
+      "loss": 1.2018,
+      "step": 7730
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.342046963580095e-07,
+      "loss": 1.2557,
+      "step": 7735
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.098860878834489e-07,
+      "loss": 1.2476,
+      "step": 7740
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.86041462282061e-07,
+      "loss": 1.1789,
+      "step": 7745
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.626709332667845e-07,
+      "loss": 1.2342,
+      "step": 7750
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.397746122896785e-07,
+      "loss": 1.209,
+      "step": 7755
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.173526085413105e-07,
+      "loss": 1.1891,
+      "step": 7760
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.954050289502799e-07,
+      "loss": 1.121,
+      "step": 7765
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7393197818270674e-07,
+      "loss": 1.216,
+      "step": 7770
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.529335586417105e-07,
+      "loss": 1.2014,
+      "step": 7775
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3240987046694327e-07,
+      "loss": 1.2685,
+      "step": 7780
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1236101153409036e-07,
+      "loss": 1.28,
+      "step": 7785
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.927870774544484e-07,
+      "loss": 1.263,
+      "step": 7790
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.7368816157441476e-07,
+      "loss": 1.2468,
+      "step": 7795
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.550643549750543e-07,
+      "loss": 1.2802,
+      "step": 7800
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3691574647171096e-07,
+      "loss": 1.2266,
+      "step": 7805
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1924242261351935e-07,
+      "loss": 1.2196,
+      "step": 7810
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.020444676830492e-07,
+      "loss": 1.2579,
+      "step": 7815
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.853219636958504e-07,
+      "loss": 1.2365,
+      "step": 7820
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6907499040011995e-07,
+      "loss": 1.1901,
+      "step": 7825
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.533036252762466e-07,
+      "loss": 1.2284,
+      "step": 7830
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3800794353651123e-07,
+      "loss": 1.2083,
+      "step": 7835
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2318801812472035e-07,
+      "loss": 1.2617,
+      "step": 7840
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0884391971579542e-07,
+      "loss": 1.2174,
+      "step": 7845
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.949757167155286e-07,
+      "loss": 1.1911,
+      "step": 7850
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8158347526014973e-07,
+      "loss": 1.1934,
+      "step": 7855
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6866725921613758e-07,
+      "loss": 1.1893,
+      "step": 7860
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5622713017978685e-07,
+      "loss": 1.267,
+      "step": 7865
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.442631474770084e-07,
+      "loss": 1.2208,
+      "step": 7870
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.327753681630073e-07,
+      "loss": 1.184,
+      "step": 7875
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2176384702200505e-07,
+      "loss": 1.2172,
+      "step": 7880
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1122863656698457e-07,
+      "loss": 1.2494,
+      "step": 7885
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.01169787039479e-07,
+      "loss": 1.234,
+      "step": 7890
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.158734640924981e-08,
+      "loss": 1.2643,
+      "step": 7895
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.248136037415366e-08,
+      "loss": 1.224,
+      "step": 7900
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.385187235983138e-08,
+      "loss": 1.1957,
+      "step": 7905
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.569892351956375e-08,
+      "loss": 1.1747,
+      "step": 7910
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.8022552734082744e-08,
+      "loss": 1.2167,
+      "step": 7915
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.0822796611327275e-08,
+      "loss": 1.2388,
+      "step": 7920
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.4099689486298835e-08,
+      "loss": 1.1948,
+      "step": 7925
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.785326342092832e-08,
+      "loss": 1.2333,
+      "step": 7930
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.208354820388726e-08,
+      "loss": 1.2133,
+      "step": 7935
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6790571350432393e-08,
+      "loss": 1.2177,
+      "step": 7940
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1974358102339054e-08,
+      "loss": 1.2099,
+      "step": 7945
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7634931427679134e-08,
+      "loss": 1.2405,
+      "step": 7950
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3772312020832179e-08,
+      "loss": 1.2382,
+      "step": 7955
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0386518302285542e-08,
+      "loss": 1.222,
+      "step": 7960
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.47756641860109e-09,
+      "loss": 1.2124,
+      "step": 7965
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.0454702423152754e-09,
+      "loss": 1.1807,
+      "step": 7970
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.0902413718836288e-09,
+      "loss": 1.2576,
+      "step": 7975
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6118891316141416e-09,
+      "loss": 1.2759,
+      "step": 7980
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.104205716672695e-10,
+      "loss": 1.248,
+      "step": 7985
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.584046794490875e-11,
+      "loss": 1.2101,
+      "step": 7990
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 1.2118308544158936,
+      "eval_runtime": 1435.8146,
+      "eval_samples_per_second": 19.715,
+      "eval_steps_per_second": 0.616,
+      "step": 7993
+    },
+    {
+      "epoch": 1.0,
+      "step": 7993,
+      "total_flos": 3.445896033337344e+16,
+      "train_loss": 1.2476971375693162,
+      "train_runtime": 45278.0566,
+      "train_samples_per_second": 5.649,
+      "train_steps_per_second": 0.177
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 7993,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 100,
+  "total_flos": 3.445896033337344e+16,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}