diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,30298 @@
+{
+  "best_metric": 0.8810526315789474,
+  "best_model_checkpoint": "videomae-finetuned-nba-5-class-4-batch-8000-vid-multiclass-4/checkpoint-40000",
+  "epoch": 24.04,
+  "eval_steps": 500,
+  "global_step": 50000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0000000000000004e-08,
+      "loss": 1.6289,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.000000000000001e-08,
+      "loss": 1.6849,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9e-08,
+      "loss": 1.7074,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2000000000000002e-07,
+      "loss": 1.657,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5000000000000002e-07,
+      "loss": 1.6217,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8e-07,
+      "loss": 1.6704,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1e-07,
+      "loss": 1.6519,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4000000000000003e-07,
+      "loss": 1.7045,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7e-07,
+      "loss": 1.6178,
+      "step": 90
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0000000000000004e-07,
+      "loss": 1.6117,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2999999999999996e-07,
+      "loss": 1.614,
+      "step": 110
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6e-07,
+      "loss": 1.6564,
+      "step": 120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8999999999999997e-07,
+      "loss": 1.6711,
+      "step": 130
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2e-07,
+      "loss": 1.6176,
+      "step": 140
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.5e-07,
+      "loss": 1.5739,
+      "step": 150
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.800000000000001e-07,
+      "loss": 1.6065,
+      "step": 160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.100000000000001e-07,
+      "loss": 1.611,
+      "step": 170
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.4e-07,
+      "loss": 1.6273,
+      "step": 180
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.7e-07,
+      "loss": 1.6327,
+      "step": 190
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 1.5965,
+      "step": 200
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.3e-07,
+      "loss": 1.5956,
+      "step": 210
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.599999999999999e-07,
+      "loss": 1.611,
+      "step": 220
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.9e-07,
+      "loss": 1.6605,
+      "step": 230
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.2e-07,
+      "loss": 1.6057,
+      "step": 240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.5e-07,
+      "loss": 1.6235,
+      "step": 250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.799999999999999e-07,
+      "loss": 1.6241,
+      "step": 260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.1e-07,
+      "loss": 1.5955,
+      "step": 270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.4e-07,
+      "loss": 1.5911,
+      "step": 280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.7e-07,
+      "loss": 1.628,
+      "step": 290
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9e-07,
+      "loss": 1.6224,
+      "step": 300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.3e-07,
+      "loss": 1.6257,
+      "step": 310
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.600000000000001e-07,
+      "loss": 1.6915,
+      "step": 320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.9e-07,
+      "loss": 1.63,
+      "step": 330
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0200000000000002e-06,
+      "loss": 1.6068,
+      "step": 340
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0500000000000001e-06,
+      "loss": 1.6432,
+      "step": 350
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.08e-06,
+      "loss": 1.636,
+      "step": 360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.11e-06,
+      "loss": 1.6221,
+      "step": 370
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.14e-06,
+      "loss": 1.6367,
+      "step": 380
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.17e-06,
+      "loss": 1.6197,
+      "step": 390
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 1.6448,
+      "step": 400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.23e-06,
+      "loss": 1.6023,
+      "step": 410
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.26e-06,
+      "loss": 1.5916,
+      "step": 420
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.29e-06,
+      "loss": 1.6068,
+      "step": 430
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3199999999999999e-06,
+      "loss": 1.6125,
+      "step": 440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.35e-06,
+      "loss": 1.5929,
+      "step": 450
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.38e-06,
+      "loss": 1.5875,
+      "step": 460
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.41e-06,
+      "loss": 1.6399,
+      "step": 470
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.44e-06,
+      "loss": 1.6119,
+      "step": 480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4700000000000001e-06,
+      "loss": 1.6221,
+      "step": 490
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5e-06,
+      "loss": 1.5873,
+      "step": 500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.53e-06,
+      "loss": 1.6285,
+      "step": 510
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5599999999999999e-06,
+      "loss": 1.6254,
+      "step": 520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.59e-06,
+      "loss": 1.6067,
+      "step": 530
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.62e-06,
+      "loss": 1.573,
+      "step": 540
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.65e-06,
+      "loss": 1.6053,
+      "step": 550
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.68e-06,
+      "loss": 1.5985,
+      "step": 560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7100000000000001e-06,
+      "loss": 1.5918,
+      "step": 570
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.74e-06,
+      "loss": 1.5986,
+      "step": 580
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.77e-06,
+      "loss": 1.613,
+      "step": 590
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8e-06,
+      "loss": 1.6021,
+      "step": 600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.83e-06,
+      "loss": 1.5799,
+      "step": 610
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.86e-06,
+      "loss": 1.59,
+      "step": 620
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8900000000000001e-06,
+      "loss": 1.6096,
+      "step": 630
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9200000000000003e-06,
+      "loss": 1.5695,
+      "step": 640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.95e-06,
+      "loss": 1.5877,
+      "step": 650
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.98e-06,
+      "loss": 1.6079,
+      "step": 660
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0100000000000002e-06,
+      "loss": 1.5655,
+      "step": 670
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0400000000000004e-06,
+      "loss": 1.6314,
+      "step": 680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.07e-06,
+      "loss": 1.5892,
+      "step": 690
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1000000000000002e-06,
+      "loss": 1.6097,
+      "step": 700
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.13e-06,
+      "loss": 1.6052,
+      "step": 710
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.16e-06,
+      "loss": 1.6101,
+      "step": 720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1899999999999998e-06,
+      "loss": 1.5883,
+      "step": 730
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.22e-06,
+      "loss": 1.6043,
+      "step": 740
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.25e-06,
+      "loss": 1.6058,
+      "step": 750
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.28e-06,
+      "loss": 1.6058,
+      "step": 760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.31e-06,
+      "loss": 1.583,
+      "step": 770
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.34e-06,
+      "loss": 1.5749,
+      "step": 780
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.37e-06,
+      "loss": 1.5734,
+      "step": 790
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 1.5619,
+      "step": 800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.43e-06,
+      "loss": 1.5821,
+      "step": 810
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.46e-06,
+      "loss": 1.5731,
+      "step": 820
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4900000000000003e-06,
+      "loss": 1.5712,
+      "step": 830
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.52e-06,
+      "loss": 1.5662,
+      "step": 840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.55e-06,
+      "loss": 1.5786,
+      "step": 850
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.58e-06,
+      "loss": 1.5821,
+      "step": 860
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.61e-06,
+      "loss": 1.5779,
+      "step": 870
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6399999999999997e-06,
+      "loss": 1.568,
+      "step": 880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.67e-06,
+      "loss": 1.5414,
+      "step": 890
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7e-06,
+      "loss": 1.5311,
+      "step": 900
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.73e-06,
+      "loss": 1.5791,
+      "step": 910
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.76e-06,
+      "loss": 1.5277,
+      "step": 920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.79e-06,
+      "loss": 1.5659,
+      "step": 930
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.82e-06,
+      "loss": 1.5636,
+      "step": 940
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8500000000000002e-06,
+      "loss": 1.5084,
+      "step": 950
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.88e-06,
+      "loss": 1.5494,
+      "step": 960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.91e-06,
+      "loss": 1.5576,
+      "step": 970
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9400000000000002e-06,
+      "loss": 1.5472,
+      "step": 980
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9700000000000004e-06,
+      "loss": 1.5201,
+      "step": 990
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3e-06,
+      "loss": 1.5265,
+      "step": 1000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0300000000000002e-06,
+      "loss": 1.5452,
+      "step": 1010
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.06e-06,
+      "loss": 1.5058,
+      "step": 1020
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.09e-06,
+      "loss": 1.5153,
+      "step": 1030
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1199999999999998e-06,
+      "loss": 1.5483,
+      "step": 1040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.15e-06,
+      "loss": 1.5126,
+      "step": 1050
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.18e-06,
+      "loss": 1.4588,
+      "step": 1060
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.21e-06,
+      "loss": 1.4918,
+      "step": 1070
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.24e-06,
+      "loss": 1.4994,
+      "step": 1080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.27e-06,
+      "loss": 1.4209,
+      "step": 1090
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3e-06,
+      "loss": 1.5078,
+      "step": 1100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3300000000000003e-06,
+      "loss": 1.5444,
+      "step": 1110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.36e-06,
+      "loss": 1.5468,
+      "step": 1120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.39e-06,
+      "loss": 1.4958,
+      "step": 1130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4200000000000003e-06,
+      "loss": 1.4897,
+      "step": 1140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4500000000000004e-06,
+      "loss": 1.4517,
+      "step": 1150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.48e-06,
+      "loss": 1.3506,
+      "step": 1160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5100000000000003e-06,
+      "loss": 1.4706,
+      "step": 1170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.54e-06,
+      "loss": 1.4046,
+      "step": 1180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.57e-06,
+      "loss": 1.498,
+      "step": 1190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6e-06,
+      "loss": 1.4703,
+      "step": 1200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.63e-06,
+      "loss": 1.4965,
+      "step": 1210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.66e-06,
+      "loss": 1.3704,
+      "step": 1220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6900000000000002e-06,
+      "loss": 1.3918,
+      "step": 1230
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.72e-06,
+      "loss": 1.3952,
+      "step": 1240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.75e-06,
+      "loss": 1.3897,
+      "step": 1250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.7800000000000002e-06,
+      "loss": 1.3657,
+      "step": 1260
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.81e-06,
+      "loss": 1.3527,
+      "step": 1270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.8400000000000005e-06,
+      "loss": 1.39,
+      "step": 1280
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.87e-06,
+      "loss": 1.3892,
+      "step": 1290
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9e-06,
+      "loss": 1.4569,
+      "step": 1300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9300000000000005e-06,
+      "loss": 1.3253,
+      "step": 1310
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.96e-06,
+      "loss": 1.5201,
+      "step": 1320
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.99e-06,
+      "loss": 1.4022,
+      "step": 1330
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.0200000000000005e-06,
+      "loss": 1.3751,
+      "step": 1340
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.05e-06,
+      "loss": 1.3118,
+      "step": 1350
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.080000000000001e-06,
+      "loss": 1.3506,
+      "step": 1360
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.1100000000000005e-06,
+      "loss": 1.522,
+      "step": 1370
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.14e-06,
+      "loss": 1.4244,
+      "step": 1380
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.170000000000001e-06,
+      "loss": 1.4063,
+      "step": 1390
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 1.48,
+      "step": 1400
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.229999999999999e-06,
+      "loss": 1.359,
+      "step": 1410
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.26e-06,
+      "loss": 1.323,
+      "step": 1420
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.29e-06,
+      "loss": 1.4078,
+      "step": 1430
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.32e-06,
+      "loss": 1.4117,
+      "step": 1440
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.35e-06,
+      "loss": 1.3358,
+      "step": 1450
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3799999999999996e-06,
+      "loss": 1.3316,
+      "step": 1460
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.41e-06,
+      "loss": 1.2612,
+      "step": 1470
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.44e-06,
+      "loss": 1.3363,
+      "step": 1480
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4699999999999996e-06,
+      "loss": 1.4093,
+      "step": 1490
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.5e-06,
+      "loss": 1.2774,
+      "step": 1500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.53e-06,
+      "loss": 1.2604,
+      "step": 1510
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.56e-06,
+      "loss": 1.2288,
+      "step": 1520
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.59e-06,
+      "loss": 1.3362,
+      "step": 1530
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.62e-06,
+      "loss": 1.4124,
+      "step": 1540
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.65e-06,
+      "loss": 1.2563,
+      "step": 1550
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.68e-06,
+      "loss": 1.2945,
+      "step": 1560
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.71e-06,
+      "loss": 1.4438,
+      "step": 1570
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.74e-06,
+      "loss": 1.2614,
+      "step": 1580
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.77e-06,
+      "loss": 1.313,
+      "step": 1590
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 1.0942,
+      "step": 1600
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.83e-06,
+      "loss": 1.2324,
+      "step": 1610
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.86e-06,
+      "loss": 1.5338,
+      "step": 1620
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.890000000000001e-06,
+      "loss": 1.4801,
+      "step": 1630
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.92e-06,
+      "loss": 1.2899,
+      "step": 1640
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.95e-06,
+      "loss": 1.3281,
+      "step": 1650
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.980000000000001e-06,
+      "loss": 1.2799,
+      "step": 1660
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.01e-06,
+      "loss": 1.2776,
+      "step": 1670
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.04e-06,
+      "loss": 1.1707,
+      "step": 1680
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.070000000000001e-06,
+      "loss": 1.2293,
+      "step": 1690
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.1e-06,
+      "loss": 1.2953,
+      "step": 1700
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.130000000000001e-06,
+      "loss": 1.2206,
+      "step": 1710
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.16e-06,
+      "loss": 1.2309,
+      "step": 1720
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.1899999999999994e-06,
+      "loss": 1.0757,
+      "step": 1730
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.22e-06,
+      "loss": 1.4371,
+      "step": 1740
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.25e-06,
+      "loss": 1.0968,
+      "step": 1750
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.279999999999999e-06,
+      "loss": 1.2251,
+      "step": 1760
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.31e-06,
+      "loss": 1.3717,
+      "step": 1770
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.34e-06,
+      "loss": 1.2957,
+      "step": 1780
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.37e-06,
+      "loss": 1.2934,
+      "step": 1790
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.4e-06,
+      "loss": 1.2453,
+      "step": 1800
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.43e-06,
+      "loss": 1.1924,
+      "step": 1810
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.46e-06,
+      "loss": 1.1958,
+      "step": 1820
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.49e-06,
+      "loss": 1.1375,
+      "step": 1830
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.52e-06,
+      "loss": 1.2275,
+      "step": 1840
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.55e-06,
+      "loss": 1.2237,
+      "step": 1850
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.58e-06,
+      "loss": 1.3711,
+      "step": 1860
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.6100000000000005e-06,
+      "loss": 1.1037,
+      "step": 1870
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.64e-06,
+      "loss": 1.2853,
+      "step": 1880
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.67e-06,
+      "loss": 1.2894,
+      "step": 1890
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.7000000000000005e-06,
+      "loss": 1.0479,
+      "step": 1900
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.73e-06,
+      "loss": 1.3781,
+      "step": 1910
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.76e-06,
+      "loss": 1.313,
+      "step": 1920
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.7900000000000005e-06,
+      "loss": 1.2074,
+      "step": 1930
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.82e-06,
+      "loss": 1.1394,
+      "step": 1940
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.850000000000001e-06,
+      "loss": 1.0924,
+      "step": 1950
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.8800000000000005e-06,
+      "loss": 1.1553,
+      "step": 1960
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.91e-06,
+      "loss": 1.3285,
+      "step": 1970
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.940000000000001e-06,
+      "loss": 1.4713,
+      "step": 1980
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9700000000000004e-06,
+      "loss": 0.9996,
+      "step": 1990
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6e-06,
+      "loss": 1.3802,
+      "step": 2000
+    },
+    {
+      "epoch": 0.04,
+      "eval_accuracy": 0.52,
+      "eval_f1": 0.52,
+      "eval_loss": 1.23806631565094,
+      "eval_runtime": 786.4493,
+      "eval_samples_per_second": 6.04,
+      "eval_steps_per_second": 1.511,
+      "step": 2000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.030000000000001e-06,
+      "loss": 1.4493,
+      "step": 2010
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.0600000000000004e-06,
+      "loss": 1.3382,
+      "step": 2020
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.090000000000001e-06,
+      "loss": 1.2994,
+      "step": 2030
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.12e-06,
+      "loss": 1.1561,
+      "step": 2040
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.1499999999999996e-06,
+      "loss": 1.2175,
+      "step": 2050
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.18e-06,
+      "loss": 1.2732,
+      "step": 2060
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.21e-06,
+      "loss": 1.1406,
+      "step": 2070
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.2399999999999995e-06,
+      "loss": 1.2093,
+      "step": 2080
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.27e-06,
+      "loss": 1.0776,
+      "step": 2090
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.3e-06,
+      "loss": 1.3075,
+      "step": 2100
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.3299999999999995e-06,
+      "loss": 1.2189,
+      "step": 2110
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.36e-06,
+      "loss": 1.1371,
+      "step": 2120
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.39e-06,
+      "loss": 1.2426,
+      "step": 2130
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.42e-06,
+      "loss": 1.1542,
+      "step": 2140
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.45e-06,
+      "loss": 1.2668,
+      "step": 2150
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.48e-06,
+      "loss": 1.1043,
+      "step": 2160
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.51e-06,
+      "loss": 1.3451,
+      "step": 2170
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.54e-06,
+      "loss": 1.1112,
+      "step": 2180
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.57e-06,
+      "loss": 0.9727,
+      "step": 2190
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.6e-06,
+      "loss": 1.1381,
+      "step": 2200
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.63e-06,
+      "loss": 1.1848,
+      "step": 2210
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.660000000000001e-06,
+      "loss": 1.2371,
+      "step": 2220
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.69e-06,
+      "loss": 1.4057,
+      "step": 2230
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.72e-06,
+      "loss": 1.2255,
+      "step": 2240
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.750000000000001e-06,
+      "loss": 1.2957,
+      "step": 2250
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 6.78e-06,
+      "loss": 1.1859,
+      "step": 2260
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 6.81e-06,
+      "loss": 1.3443,
+      "step": 2270
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 6.840000000000001e-06,
+      "loss": 1.2337,
+      "step": 2280
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 6.87e-06,
+      "loss": 1.0896,
+      "step": 2290
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 6.900000000000001e-06,
+      "loss": 1.1986,
+      "step": 2300
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 6.9300000000000006e-06,
+      "loss": 1.2643,
+      "step": 2310
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 6.96e-06,
+      "loss": 1.0731,
+      "step": 2320
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 6.990000000000001e-06,
+      "loss": 1.0705,
+      "step": 2330
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.0200000000000006e-06,
+      "loss": 1.141,
+      "step": 2340
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.049999999999999e-06,
+      "loss": 0.9496,
+      "step": 2350
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.08e-06,
+      "loss": 1.2506,
+      "step": 2360
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.11e-06,
+      "loss": 1.3169,
+      "step": 2370
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.14e-06,
+      "loss": 1.2174,
+      "step": 2380
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.17e-06,
+      "loss": 1.3699,
+      "step": 2390
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.2e-06,
+      "loss": 1.2294,
+      "step": 2400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.23e-06,
+      "loss": 1.0979,
+      "step": 2410
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.26e-06,
+      "loss": 1.1005,
+      "step": 2420
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.29e-06,
+      "loss": 1.0107,
+      "step": 2430
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.32e-06,
+      "loss": 1.187,
+      "step": 2440
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.35e-06,
+      "loss": 0.9485,
+      "step": 2450
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.3800000000000005e-06,
+      "loss": 1.2392,
+      "step": 2460
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.41e-06,
+      "loss": 1.1229,
+      "step": 2470
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.44e-06,
+      "loss": 1.0925,
+      "step": 2480
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.4700000000000005e-06,
+      "loss": 1.1106,
+      "step": 2490
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.5e-06,
+      "loss": 1.2757,
+      "step": 2500
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.53e-06,
+      "loss": 1.0142,
+      "step": 2510
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.5600000000000005e-06,
+      "loss": 1.3197,
+      "step": 2520
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.59e-06,
+      "loss": 1.1721,
+      "step": 2530
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.62e-06,
+      "loss": 1.2866,
+      "step": 2540
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.65e-06,
+      "loss": 1.2752,
+      "step": 2550
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.680000000000001e-06,
+      "loss": 1.1496,
+      "step": 2560
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.71e-06,
+      "loss": 1.0845,
+      "step": 2570
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.74e-06,
+      "loss": 1.0634,
+      "step": 2580
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.77e-06,
+      "loss": 1.179,
+      "step": 2590
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.8e-06,
+      "loss": 1.2051,
+      "step": 2600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.830000000000001e-06,
+      "loss": 1.062,
+      "step": 2610
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.860000000000001e-06,
+      "loss": 1.2288,
+      "step": 2620
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.89e-06,
+      "loss": 1.4461,
+      "step": 2630
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.92e-06,
+      "loss": 1.0729,
+      "step": 2640
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.95e-06,
+      "loss": 1.1626,
+      "step": 2650
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.98e-06,
+      "loss": 1.1844,
+      "step": 2660
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.010000000000001e-06,
+      "loss": 1.2165,
+      "step": 2670
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.040000000000001e-06,
+      "loss": 1.0876,
+      "step": 2680
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.07e-06,
+      "loss": 1.2056,
+      "step": 2690
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.1e-06,
+      "loss": 1.3396,
+      "step": 2700
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.13e-06,
+      "loss": 1.165,
+      "step": 2710
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.160000000000001e-06,
+      "loss": 0.9852,
+      "step": 2720
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.190000000000001e-06,
+      "loss": 0.7788,
+      "step": 2730
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.220000000000001e-06,
+      "loss": 1.1942,
+      "step": 2740
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.25e-06,
+      "loss": 1.2522,
+      "step": 2750
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.28e-06,
+      "loss": 1.2626,
+      "step": 2760
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.310000000000002e-06,
+      "loss": 1.1067,
+      "step": 2770
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.340000000000001e-06,
+      "loss": 1.0825,
+      "step": 2780
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.370000000000001e-06,
+      "loss": 1.1901,
+      "step": 2790
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 1.2596,
+      "step": 2800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.43e-06,
+      "loss": 1.2528,
+      "step": 2810
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.459999999999999e-06,
+      "loss": 1.1476,
+      "step": 2820
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.49e-06,
+      "loss": 1.2297,
+      "step": 2830
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.52e-06,
+      "loss": 1.0016,
+      "step": 2840
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.55e-06,
+      "loss": 1.2551,
+      "step": 2850
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.58e-06,
+      "loss": 0.9679,
+      "step": 2860
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.609999999999999e-06,
+      "loss": 1.2398,
+      "step": 2870
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.64e-06,
+      "loss": 1.0966,
+      "step": 2880
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.67e-06,
+      "loss": 0.9264,
+      "step": 2890
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.7e-06,
+      "loss": 1.1826,
+      "step": 2900
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.73e-06,
+      "loss": 1.0171,
+      "step": 2910
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.759999999999999e-06,
+      "loss": 1.0463,
+      "step": 2920
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.79e-06,
+      "loss": 1.2805,
+      "step": 2930
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.82e-06,
+      "loss": 1.1559,
+      "step": 2940
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.85e-06,
+      "loss": 1.1426,
+      "step": 2950
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.88e-06,
+      "loss": 1.3127,
+      "step": 2960
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.91e-06,
+      "loss": 1.2136,
+      "step": 2970
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.939999999999999e-06,
+      "loss": 1.2175,
+      "step": 2980
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.97e-06,
+      "loss": 1.3208,
+      "step": 2990
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9e-06,
+      "loss": 1.2648,
+      "step": 3000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.03e-06,
+      "loss": 1.0219,
+      "step": 3010
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.06e-06,
+      "loss": 1.1467,
+      "step": 3020
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.09e-06,
+      "loss": 0.9967,
+      "step": 3030
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.12e-06,
+      "loss": 0.9672,
+      "step": 3040
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.15e-06,
+      "loss": 0.9971,
+      "step": 3050
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.18e-06,
+      "loss": 1.2074,
+      "step": 3060
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.21e-06,
+      "loss": 1.2956,
+      "step": 3070
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.24e-06,
+      "loss": 1.0696,
+      "step": 3080
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.27e-06,
+      "loss": 1.1306,
+      "step": 3090
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.3e-06,
+      "loss": 0.9367,
+      "step": 3100
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.33e-06,
+      "loss": 0.9672,
+      "step": 3110
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.36e-06,
+      "loss": 1.0834,
+      "step": 3120
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.39e-06,
+      "loss": 1.0347,
+      "step": 3130
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.42e-06,
+      "loss": 0.881,
+      "step": 3140
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.450000000000001e-06,
+      "loss": 0.8191,
+      "step": 3150
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.48e-06,
+      "loss": 1.0785,
+      "step": 3160
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.51e-06,
+      "loss": 1.1429,
+      "step": 3170
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.54e-06,
+      "loss": 0.9583,
+      "step": 3180
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.57e-06,
+      "loss": 1.3133,
+      "step": 3190
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 1.1462,
+      "step": 3200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.630000000000001e-06,
+      "loss": 1.1374,
+      "step": 3210
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.66e-06,
+      "loss": 1.0824,
+      "step": 3220
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.69e-06,
+      "loss": 1.2885,
+      "step": 3230
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.72e-06,
+      "loss": 1.2692,
+      "step": 3240
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.75e-06,
+      "loss": 1.4844,
+      "step": 3250
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.780000000000001e-06,
+      "loss": 1.2084,
+      "step": 3260
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.810000000000001e-06,
+      "loss": 1.1797,
+      "step": 3270
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.84e-06,
+      "loss": 1.0209,
+      "step": 3280
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.87e-06,
+      "loss": 1.1393,
+      "step": 3290
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.9e-06,
+      "loss": 1.023,
+      "step": 3300
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.930000000000001e-06,
+      "loss": 1.0185,
+      "step": 3310
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.960000000000001e-06,
+      "loss": 1.0343,
+      "step": 3320
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.990000000000001e-06,
+      "loss": 1.0599,
+      "step": 3330
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.002e-05,
+      "loss": 0.9796,
+      "step": 3340
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.005e-05,
+      "loss": 1.016,
+      "step": 3350
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.008e-05,
+      "loss": 1.3834,
+      "step": 3360
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0110000000000001e-05,
+      "loss": 1.0219,
+      "step": 3370
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0140000000000001e-05,
+      "loss": 0.7479,
+      "step": 3380
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0170000000000001e-05,
+      "loss": 1.2832,
+      "step": 3390
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.02e-05,
+      "loss": 1.2133,
+      "step": 3400
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.023e-05,
+      "loss": 1.3182,
+      "step": 3410
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0260000000000002e-05,
+      "loss": 0.9505,
+      "step": 3420
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0290000000000001e-05,
+      "loss": 0.8651,
+      "step": 3430
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.032e-05,
+      "loss": 0.9899,
+      "step": 3440
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.035e-05,
+      "loss": 1.1628,
+      "step": 3450
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0379999999999999e-05,
+      "loss": 1.2309,
+      "step": 3460
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.041e-05,
+      "loss": 1.0505,
+      "step": 3470
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.044e-05,
+      "loss": 1.042,
+      "step": 3480
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.047e-05,
+      "loss": 1.1551,
+      "step": 3490
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.05e-05,
+      "loss": 1.0644,
+      "step": 3500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0529999999999999e-05,
+      "loss": 1.0462,
+      "step": 3510
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0559999999999999e-05,
+      "loss": 1.0319,
+      "step": 3520
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.059e-05,
+      "loss": 1.0456,
+      "step": 3530
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.062e-05,
+      "loss": 0.9852,
+      "step": 3540
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.065e-05,
+      "loss": 1.0286,
+      "step": 3550
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.068e-05,
+      "loss": 1.0501,
+      "step": 3560
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0709999999999999e-05,
+      "loss": 1.0495,
+      "step": 3570
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.074e-05,
+      "loss": 0.947,
+      "step": 3580
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.077e-05,
+      "loss": 0.9354,
+      "step": 3590
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.08e-05,
+      "loss": 0.9716,
+      "step": 3600
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.083e-05,
+      "loss": 1.0102,
+      "step": 3610
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.086e-05,
+      "loss": 1.2956,
+      "step": 3620
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.089e-05,
+      "loss": 1.311,
+      "step": 3630
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.092e-05,
+      "loss": 1.2648,
+      "step": 3640
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.095e-05,
+      "loss": 1.1586,
+      "step": 3650
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.098e-05,
+      "loss": 1.0152,
+      "step": 3660
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.101e-05,
+      "loss": 1.0581,
+      "step": 3670
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.104e-05,
+      "loss": 1.0796,
+      "step": 3680
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.107e-05,
+      "loss": 1.078,
+      "step": 3690
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.11e-05,
+      "loss": 1.2068,
+      "step": 3700
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.113e-05,
+      "loss": 1.0106,
+      "step": 3710
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.116e-05,
+      "loss": 1.1514,
+      "step": 3720
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.119e-05,
+      "loss": 0.9577,
+      "step": 3730
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.1220000000000001e-05,
+      "loss": 1.2864,
+      "step": 3740
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.125e-05,
+      "loss": 1.1648,
+      "step": 3750
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.128e-05,
+      "loss": 1.1744,
+      "step": 3760
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.131e-05,
+      "loss": 1.0227,
+      "step": 3770
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.134e-05,
+      "loss": 0.8843,
+      "step": 3780
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.137e-05,
+      "loss": 0.9374,
+      "step": 3790
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.1400000000000001e-05,
+      "loss": 1.045,
+      "step": 3800
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.143e-05,
+      "loss": 1.1581,
+      "step": 3810
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.146e-05,
+      "loss": 1.0742,
+      "step": 3820
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.149e-05,
+      "loss": 1.2959,
+      "step": 3830
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.152e-05,
+      "loss": 1.1807,
+      "step": 3840
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.1550000000000001e-05,
+      "loss": 1.0489,
+      "step": 3850
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.1580000000000001e-05,
+      "loss": 1.1499,
+      "step": 3860
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.161e-05,
+      "loss": 1.1995,
+      "step": 3870
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.164e-05,
+      "loss": 0.9318,
+      "step": 3880
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.167e-05,
+      "loss": 1.0526,
+      "step": 3890
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.1700000000000001e-05,
+      "loss": 1.1047,
+      "step": 3900
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.1730000000000001e-05,
+      "loss": 1.0165,
+      "step": 3910
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.1760000000000001e-05,
+      "loss": 1.188,
+      "step": 3920
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.179e-05,
+      "loss": 1.137,
+      "step": 3930
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.182e-05,
+      "loss": 1.1563,
+      "step": 3940
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.185e-05,
+      "loss": 1.1343,
+      "step": 3950
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.1880000000000001e-05,
+      "loss": 1.1096,
+      "step": 3960
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.1910000000000001e-05,
+      "loss": 1.1273,
+      "step": 3970
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.1940000000000001e-05,
+      "loss": 0.976,
+      "step": 3980
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.197e-05,
+      "loss": 0.9634,
+      "step": 3990
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2e-05,
+      "loss": 1.0115,
+      "step": 4000
+    },
+    {
+      "epoch": 1.04,
+      "eval_accuracy": 0.6684210526315789,
+      "eval_f1": 0.6684210526315789,
+      "eval_loss": 1.0522350072860718,
+      "eval_runtime": 771.0264,
+      "eval_samples_per_second": 6.161,
+      "eval_steps_per_second": 1.541,
+      "step": 4000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.2030000000000002e-05,
+      "loss": 1.1145,
+      "step": 4010
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.2060000000000001e-05,
+      "loss": 1.378,
+      "step": 4020
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.2090000000000001e-05,
+      "loss": 1.1337,
+      "step": 4030
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.2120000000000001e-05,
+      "loss": 0.8445,
+      "step": 4040
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.215e-05,
+      "loss": 0.8296,
+      "step": 4050
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.2180000000000002e-05,
+      "loss": 1.2409,
+      "step": 4060
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.221e-05,
+      "loss": 1.0398,
+      "step": 4070
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.224e-05,
+      "loss": 0.9204,
+      "step": 4080
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.227e-05,
+      "loss": 0.9676,
+      "step": 4090
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.2299999999999999e-05,
+      "loss": 1.0203,
+      "step": 4100
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.2329999999999999e-05,
+      "loss": 1.1571,
+      "step": 4110
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.236e-05,
+      "loss": 1.2317,
+      "step": 4120
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.239e-05,
+      "loss": 1.004,
+      "step": 4130
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.242e-05,
+      "loss": 1.3242,
+      "step": 4140
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.245e-05,
+      "loss": 1.0648,
+      "step": 4150
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.2479999999999999e-05,
+      "loss": 0.8635,
+      "step": 4160
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.251e-05,
+      "loss": 0.9419,
+      "step": 4170
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.254e-05,
+      "loss": 1.0629,
+      "step": 4180
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.257e-05,
+      "loss": 1.0586,
+      "step": 4190
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.26e-05,
+      "loss": 0.9666,
+      "step": 4200
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.263e-05,
+      "loss": 0.8782,
+      "step": 4210
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.2659999999999999e-05,
+      "loss": 1.1148,
+      "step": 4220
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.269e-05,
+      "loss": 0.9179,
+      "step": 4230
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.272e-05,
+      "loss": 1.1387,
+      "step": 4240
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.275e-05,
+      "loss": 1.1456,
+      "step": 4250
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.278e-05,
+      "loss": 1.1761,
+      "step": 4260
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.281e-05,
+      "loss": 0.7777,
+      "step": 4270
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.284e-05,
+      "loss": 0.9239,
+      "step": 4280
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.287e-05,
+      "loss": 0.8541,
+      "step": 4290
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.29e-05,
+      "loss": 1.1384,
+      "step": 4300
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.293e-05,
+      "loss": 0.9394,
+      "step": 4310
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.296e-05,
+      "loss": 0.9221,
+      "step": 4320
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.2990000000000001e-05,
+      "loss": 1.0862,
+      "step": 4330
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.302e-05,
+      "loss": 1.1334,
+      "step": 4340
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.305e-05,
+      "loss": 1.1778,
+      "step": 4350
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.308e-05,
+      "loss": 0.9096,
+      "step": 4360
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.311e-05,
+      "loss": 1.006,
+      "step": 4370
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.314e-05,
+      "loss": 1.0986,
+      "step": 4380
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3170000000000001e-05,
+      "loss": 1.0341,
+      "step": 4390
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.32e-05,
+      "loss": 1.1357,
+      "step": 4400
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.323e-05,
+      "loss": 0.9352,
+      "step": 4410
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.326e-05,
+      "loss": 1.1474,
+      "step": 4420
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.329e-05,
+      "loss": 0.8776,
+      "step": 4430
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3320000000000001e-05,
+      "loss": 0.945,
+      "step": 4440
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3350000000000001e-05,
+      "loss": 1.067,
+      "step": 4450
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.338e-05,
+      "loss": 0.6447,
+      "step": 4460
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.341e-05,
+      "loss": 1.123,
+      "step": 4470
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.344e-05,
+      "loss": 0.9915,
+      "step": 4480
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3470000000000001e-05,
+      "loss": 1.0084,
+      "step": 4490
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3500000000000001e-05,
+      "loss": 1.0558,
+      "step": 4500
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3530000000000001e-05,
+      "loss": 0.9298,
+      "step": 4510
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.356e-05,
+      "loss": 0.948,
+      "step": 4520
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.359e-05,
+      "loss": 1.0212,
+      "step": 4530
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.362e-05,
+      "loss": 1.1543,
+      "step": 4540
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3650000000000001e-05,
+      "loss": 1.1628,
+      "step": 4550
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3680000000000001e-05,
+      "loss": 1.231,
+      "step": 4560
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3710000000000001e-05,
+      "loss": 1.251,
+      "step": 4570
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.374e-05,
+      "loss": 0.821,
+      "step": 4580
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.377e-05,
+      "loss": 0.786,
+      "step": 4590
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3800000000000002e-05,
+      "loss": 1.1307,
+      "step": 4600
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3830000000000001e-05,
+      "loss": 1.1182,
+      "step": 4610
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3860000000000001e-05,
+      "loss": 1.0388,
+      "step": 4620
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.389e-05,
+      "loss": 1.0585,
+      "step": 4630
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.392e-05,
+      "loss": 0.7796,
+      "step": 4640
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.395e-05,
+      "loss": 1.1808,
+      "step": 4650
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3980000000000002e-05,
+      "loss": 0.9879,
+      "step": 4660
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.4010000000000001e-05,
+      "loss": 1.0262,
+      "step": 4670
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.4040000000000001e-05,
+      "loss": 0.9567,
+      "step": 4680
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.4069999999999999e-05,
+      "loss": 1.2993,
+      "step": 4690
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.4099999999999999e-05,
+      "loss": 1.024,
+      "step": 4700
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.413e-05,
+      "loss": 1.0662,
+      "step": 4710
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.416e-05,
+      "loss": 1.2413,
+      "step": 4720
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.419e-05,
+      "loss": 1.0022,
+      "step": 4730
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.422e-05,
+      "loss": 0.807,
+      "step": 4740
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.4249999999999999e-05,
+      "loss": 1.2506,
+      "step": 4750
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.428e-05,
+      "loss": 1.1809,
+      "step": 4760
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.431e-05,
+      "loss": 1.0797,
+      "step": 4770
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.434e-05,
+      "loss": 0.9592,
+      "step": 4780
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.437e-05,
+      "loss": 1.0962,
+      "step": 4790
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.44e-05,
+      "loss": 1.1471,
+      "step": 4800
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.4429999999999999e-05,
+      "loss": 0.9697,
+      "step": 4810
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.446e-05,
+      "loss": 1.0342,
+      "step": 4820
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.449e-05,
+      "loss": 0.8653,
+      "step": 4830
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.452e-05,
+      "loss": 1.1813,
+      "step": 4840
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.455e-05,
+      "loss": 1.097,
+      "step": 4850
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.458e-05,
+      "loss": 1.3181,
+      "step": 4860
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.461e-05,
+      "loss": 1.0778,
+      "step": 4870
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.464e-05,
+      "loss": 1.096,
+      "step": 4880
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.467e-05,
+      "loss": 0.8033,
+      "step": 4890
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.47e-05,
+      "loss": 1.0194,
+      "step": 4900
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.473e-05,
+      "loss": 1.0501,
+      "step": 4910
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.4760000000000001e-05,
+      "loss": 0.7855,
+      "step": 4920
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.479e-05,
+      "loss": 1.151,
+      "step": 4930
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.482e-05,
+      "loss": 1.072,
+      "step": 4940
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.485e-05,
+      "loss": 1.0244,
+      "step": 4950
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.488e-05,
+      "loss": 0.9692,
+      "step": 4960
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.491e-05,
+      "loss": 1.0249,
+      "step": 4970
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.4940000000000001e-05,
+      "loss": 0.964,
+      "step": 4980
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.497e-05,
+      "loss": 0.9706,
+      "step": 4990
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.5e-05,
+      "loss": 1.2112,
+      "step": 5000
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.4999998172295556e-05,
+      "loss": 0.8264,
+      "step": 5010
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.4999992689183113e-05,
+      "loss": 0.952,
+      "step": 5020
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.4999983550665345e-05,
+      "loss": 1.3463,
+      "step": 5030
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.4999970756746704e-05,
+      "loss": 0.9856,
+      "step": 5040
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.4999954307433428e-05,
+      "loss": 0.904,
+      "step": 5050
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.4999934202733533e-05,
+      "loss": 0.9741,
+      "step": 5060
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.4999910442656817e-05,
+      "loss": 0.8908,
+      "step": 5070
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.4999883027214862e-05,
+      "loss": 1.2765,
+      "step": 5080
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.4999851956421028e-05,
+      "loss": 0.9875,
+      "step": 5090
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.499981723029046e-05,
+      "loss": 0.7411,
+      "step": 5100
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.4999778848840085e-05,
+      "loss": 0.9148,
+      "step": 5110
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.4999736812088606e-05,
+      "loss": 0.7769,
+      "step": 5120
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.4999691120056512e-05,
+      "loss": 0.8871,
+      "step": 5130
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.4999641772766074e-05,
+      "loss": 0.9924,
+      "step": 5140
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.4999588770241342e-05,
+      "loss": 1.062,
+      "step": 5150
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.499953211250815e-05,
+      "loss": 1.0356,
+      "step": 5160
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.499947179959411e-05,
+      "loss": 0.9518,
+      "step": 5170
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.4999407831528622e-05,
+      "loss": 1.1531,
+      "step": 5180
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.4999340208342858e-05,
+      "loss": 0.9666,
+      "step": 5190
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.4999268930069782e-05,
+      "loss": 1.3574,
+      "step": 5200
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.4999193996744131e-05,
+      "loss": 0.9965,
+      "step": 5210
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.4999115408402427e-05,
+      "loss": 1.0885,
+      "step": 5220
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.4999033165082974e-05,
+      "loss": 0.9979,
+      "step": 5230
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.4998947266825853e-05,
+      "loss": 1.0737,
+      "step": 5240
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.4998857713672935e-05,
+      "loss": 0.9925,
+      "step": 5250
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.4998764505667862e-05,
+      "loss": 1.1046,
+      "step": 5260
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.4998667642856068e-05,
+      "loss": 0.9812,
+      "step": 5270
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.4998567125284757e-05,
+      "loss": 1.0785,
+      "step": 5280
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.4998462953002925e-05,
+      "loss": 1.0161,
+      "step": 5290
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.4998355126061342e-05,
+      "loss": 0.8528,
+      "step": 5300
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.499824364451256e-05,
+      "loss": 1.0053,
+      "step": 5310
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.4998128508410916e-05,
+      "loss": 1.0668,
+      "step": 5320
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.4998009717812525e-05,
+      "loss": 0.9532,
+      "step": 5330
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.4997887272775285e-05,
+      "loss": 0.8229,
+      "step": 5340
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.4997761173358875e-05,
+      "loss": 1.0487,
+      "step": 5350
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.499763141962475e-05,
+      "loss": 0.8786,
+      "step": 5360
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.4997498011636154e-05,
+      "loss": 0.9382,
+      "step": 5370
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.499736094945811e-05,
+      "loss": 1.2307,
+      "step": 5380
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.4997220233157415e-05,
+      "loss": 1.0116,
+      "step": 5390
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.4997075862802657e-05,
+      "loss": 1.0605,
+      "step": 5400
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.49969278384642e-05,
+      "loss": 0.9716,
+      "step": 5410
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.4996776160214188e-05,
+      "loss": 1.1633,
+      "step": 5420
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.4996620828126546e-05,
+      "loss": 0.7762,
+      "step": 5430
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.4996461842276982e-05,
+      "loss": 1.1011,
+      "step": 5440
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.4996299202742987e-05,
+      "loss": 1.0375,
+      "step": 5450
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.4996132909603825e-05,
+      "loss": 1.0724,
+      "step": 5460
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.4995962962940547e-05,
+      "loss": 1.5755,
+      "step": 5470
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.4995789362835983e-05,
+      "loss": 0.9854,
+      "step": 5480
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.4995612109374742e-05,
+      "loss": 0.8795,
+      "step": 5490
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.4995431202643219e-05,
+      "loss": 0.8795,
+      "step": 5500
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.499524664272958e-05,
+      "loss": 0.9471,
+      "step": 5510
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.4995058429723783e-05,
+      "loss": 1.116,
+      "step": 5520
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.499486656371756e-05,
+      "loss": 1.0457,
+      "step": 5530
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.4994671044804419e-05,
+      "loss": 1.1661,
+      "step": 5540
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.4994471873079658e-05,
+      "loss": 0.9537,
+      "step": 5550
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.499426904864035e-05,
+      "loss": 0.9942,
+      "step": 5560
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.4994062571585351e-05,
+      "loss": 1.101,
+      "step": 5570
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.4993852442015293e-05,
+      "loss": 1.3061,
+      "step": 5580
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.499363866003259e-05,
+      "loss": 1.0937,
+      "step": 5590
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.4993421225741438e-05,
+      "loss": 0.8348,
+      "step": 5600
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.4993200139247813e-05,
+      "loss": 1.1291,
+      "step": 5610
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.4992975400659466e-05,
+      "loss": 0.9377,
+      "step": 5620
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.4992747010085936e-05,
+      "loss": 0.9313,
+      "step": 5630
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.4992514967638537e-05,
+      "loss": 1.1957,
+      "step": 5640
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.4992279273430361e-05,
+      "loss": 0.9746,
+      "step": 5650
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.4992039927576285e-05,
+      "loss": 1.1093,
+      "step": 5660
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.4991796930192962e-05,
+      "loss": 0.9546,
+      "step": 5670
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.4991550281398828e-05,
+      "loss": 0.9618,
+      "step": 5680
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.4991299981314094e-05,
+      "loss": 0.9739,
+      "step": 5690
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.4991046030060756e-05,
+      "loss": 0.9386,
+      "step": 5700
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.4990788427762585e-05,
+      "loss": 1.2641,
+      "step": 5710
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.4990527174545132e-05,
+      "loss": 1.0383,
+      "step": 5720
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.499026227053573e-05,
+      "loss": 0.8838,
+      "step": 5730
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.498999371586349e-05,
+      "loss": 1.0993,
+      "step": 5740
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.4989721510659303e-05,
+      "loss": 1.2501,
+      "step": 5750
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.498944565505584e-05,
+      "loss": 1.2304,
+      "step": 5760
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.4989166149187544e-05,
+      "loss": 0.9101,
+      "step": 5770
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.498888299319065e-05,
+      "loss": 1.1794,
+      "step": 5780
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.4988596187203158e-05,
+      "loss": 0.9571,
+      "step": 5790
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.4988305731364858e-05,
+      "loss": 1.2548,
+      "step": 5800
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.4988011625817314e-05,
+      "loss": 1.0551,
+      "step": 5810
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.4987713870703869e-05,
+      "loss": 1.1907,
+      "step": 5820
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.4987412466169642e-05,
+      "loss": 1.0176,
+      "step": 5830
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.4987107412361541e-05,
+      "loss": 1.0289,
+      "step": 5840
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.4986798709428242e-05,
+      "loss": 1.3242,
+      "step": 5850
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.49864863575202e-05,
+      "loss": 1.2081,
+      "step": 5860
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.4986170356789654e-05,
+      "loss": 0.8864,
+      "step": 5870
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.4985850707390621e-05,
+      "loss": 0.7922,
+      "step": 5880
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.4985527409478893e-05,
+      "loss": 0.9615,
+      "step": 5890
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.4985200463212038e-05,
+      "loss": 0.8607,
+      "step": 5900
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.498486986874941e-05,
+      "loss": 0.8566,
+      "step": 5910
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.4984535626252133e-05,
+      "loss": 1.2487,
+      "step": 5920
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.4984197735883119e-05,
+      "loss": 1.1537,
+      "step": 5930
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.4983856197807045e-05,
+      "loss": 1.1069,
+      "step": 5940
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.4983511012190374e-05,
+      "loss": 0.6814,
+      "step": 5950
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.498316217920135e-05,
+      "loss": 1.1646,
+      "step": 5960
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.4982809699009982e-05,
+      "loss": 1.0384,
+      "step": 5970
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.4982453571788074e-05,
+      "loss": 1.0221,
+      "step": 5980
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.4982093797709188e-05,
+      "loss": 1.0354,
+      "step": 5990
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.4981730376948682e-05,
+      "loss": 0.9749,
+      "step": 6000
+    },
+    {
+      "epoch": 2.04,
+      "eval_accuracy": 0.7536842105263157,
+      "eval_f1": 0.7536842105263157,
+      "eval_loss": 0.9298247694969177,
+      "eval_runtime": 753.1546,
+      "eval_samples_per_second": 6.307,
+      "eval_steps_per_second": 1.577,
+      "step": 6000
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4981363309683678e-05,
+      "loss": 1.0715,
+      "step": 6010
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4980992596093081e-05,
+      "loss": 0.7671,
+      "step": 6020
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4980618236357574e-05,
+      "loss": 1.0827,
+      "step": 6030
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4980240230659615e-05,
+      "loss": 0.8094,
+      "step": 6040
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4979858579183435e-05,
+      "loss": 1.0124,
+      "step": 6050
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4979473282115054e-05,
+      "loss": 1.0736,
+      "step": 6060
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4979084339642255e-05,
+      "loss": 0.9094,
+      "step": 6070
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4978691751954603e-05,
+      "loss": 0.8073,
+      "step": 6080
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.497829551924345e-05,
+      "loss": 1.0181,
+      "step": 6090
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4977895641701902e-05,
+      "loss": 0.9183,
+      "step": 6100
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4977492119524863e-05,
+      "loss": 0.7493,
+      "step": 6110
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4977084952909003e-05,
+      "loss": 1.0122,
+      "step": 6120
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4976674142052769e-05,
+      "loss": 0.9298,
+      "step": 6130
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4976259687156385e-05,
+      "loss": 0.9499,
+      "step": 6140
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4975841588421854e-05,
+      "loss": 1.0176,
+      "step": 6150
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.497541984605295e-05,
+      "loss": 1.1361,
+      "step": 6160
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4974994460255223e-05,
+      "loss": 0.7681,
+      "step": 6170
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4974565431236006e-05,
+      "loss": 0.8797,
+      "step": 6180
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.49741327592044e-05,
+      "loss": 0.9948,
+      "step": 6190
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4973696444371283e-05,
+      "loss": 1.2721,
+      "step": 6200
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.497325648694931e-05,
+      "loss": 0.9341,
+      "step": 6210
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4972812887152913e-05,
+      "loss": 0.9105,
+      "step": 6220
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4972365645198294e-05,
+      "loss": 1.1488,
+      "step": 6230
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4971914761303436e-05,
+      "loss": 1.1213,
+      "step": 6240
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4971460235688093e-05,
+      "loss": 0.9571,
+      "step": 6250
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4971002068573793e-05,
+      "loss": 0.8745,
+      "step": 6260
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4970540260183847e-05,
+      "loss": 1.1222,
+      "step": 6270
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.497007481074333e-05,
+      "loss": 0.9804,
+      "step": 6280
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4969605720479096e-05,
+      "loss": 0.933,
+      "step": 6290
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4969132989619776e-05,
+      "loss": 1.0085,
+      "step": 6300
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4968656618395776e-05,
+      "loss": 0.8859,
+      "step": 6310
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.496817660703927e-05,
+      "loss": 0.7726,
+      "step": 6320
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4967692955784207e-05,
+      "loss": 1.1581,
+      "step": 6330
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4967205664866318e-05,
+      "loss": 0.9762,
+      "step": 6340
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4966714734523101e-05,
+      "loss": 0.7109,
+      "step": 6350
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4966220164993826e-05,
+      "loss": 0.9304,
+      "step": 6360
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4965721956519547e-05,
+      "loss": 0.7953,
+      "step": 6370
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.496522010934308e-05,
+      "loss": 1.2858,
+      "step": 6380
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4964714623709019e-05,
+      "loss": 1.1374,
+      "step": 6390
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4964205499863734e-05,
+      "loss": 1.2635,
+      "step": 6400
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4963692738055364e-05,
+      "loss": 1.2273,
+      "step": 6410
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4963176338533823e-05,
+      "loss": 0.9485,
+      "step": 6420
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.49626563015508e-05,
+      "loss": 0.8397,
+      "step": 6430
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4962132627359753e-05,
+      "loss": 0.9709,
+      "step": 6440
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4961605316215913e-05,
+      "loss": 0.7586,
+      "step": 6450
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4961074368376286e-05,
+      "loss": 1.2284,
+      "step": 6460
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.496053978409965e-05,
+      "loss": 1.1168,
+      "step": 6470
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4960001563646557e-05,
+      "loss": 0.9851,
+      "step": 6480
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4959459707279325e-05,
+      "loss": 1.106,
+      "step": 6490
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.495891421526205e-05,
+      "loss": 1.0387,
+      "step": 6500
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.49583650878606e-05,
+      "loss": 0.8843,
+      "step": 6510
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.495781232534261e-05,
+      "loss": 1.2072,
+      "step": 6520
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4957255927977493e-05,
+      "loss": 0.7769,
+      "step": 6530
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4956695896036427e-05,
+      "loss": 0.7896,
+      "step": 6540
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4956132229792366e-05,
+      "loss": 0.9106,
+      "step": 6550
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4955564929520036e-05,
+      "loss": 1.1088,
+      "step": 6560
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4954993995495928e-05,
+      "loss": 1.0579,
+      "step": 6570
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4954419427998312e-05,
+      "loss": 1.0675,
+      "step": 6580
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4953841227307225e-05,
+      "loss": 1.046,
+      "step": 6590
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4953259393704474e-05,
+      "loss": 0.7583,
+      "step": 6600
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4952673927473636e-05,
+      "loss": 0.9493,
+      "step": 6610
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4952084828900064e-05,
+      "loss": 0.9263,
+      "step": 6620
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4951492098270874e-05,
+      "loss": 0.9671,
+      "step": 6630
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4950895735874958e-05,
+      "loss": 1.1237,
+      "step": 6640
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4950295742002972e-05,
+      "loss": 1.056,
+      "step": 6650
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4949692116947354e-05,
+      "loss": 1.439,
+      "step": 6660
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4949084861002293e-05,
+      "loss": 0.9316,
+      "step": 6670
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4948473974463767e-05,
+      "loss": 0.6299,
+      "step": 6680
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4947859457629508e-05,
+      "loss": 0.7169,
+      "step": 6690
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4947241310799028e-05,
+      "loss": 0.6922,
+      "step": 6700
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4946619534273603e-05,
+      "loss": 0.8926,
+      "step": 6710
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.494599412835628e-05,
+      "loss": 1.0797,
+      "step": 6720
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4945365093351874e-05,
+      "loss": 0.8354,
+      "step": 6730
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4944732429566967e-05,
+      "loss": 0.7327,
+      "step": 6740
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4944096137309916e-05,
+      "loss": 0.8729,
+      "step": 6750
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4943456216890838e-05,
+      "loss": 1.0992,
+      "step": 6760
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4942812668621623e-05,
+      "loss": 0.6533,
+      "step": 6770
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4942165492815934e-05,
+      "loss": 1.1681,
+      "step": 6780
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.494151468978919e-05,
+      "loss": 0.8496,
+      "step": 6790
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4940860259858585e-05,
+      "loss": 1.1699,
+      "step": 6800
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4940202203343083e-05,
+      "loss": 0.9825,
+      "step": 6810
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4939540520563411e-05,
+      "loss": 0.9348,
+      "step": 6820
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4938875211842066e-05,
+      "loss": 0.949,
+      "step": 6830
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4938206277503313e-05,
+      "loss": 1.1664,
+      "step": 6840
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4937533717873178e-05,
+      "loss": 0.961,
+      "step": 6850
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4936857533279463e-05,
+      "loss": 1.007,
+      "step": 6860
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4936177724051729e-05,
+      "loss": 0.9197,
+      "step": 6870
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.493549429052131e-05,
+      "loss": 0.8877,
+      "step": 6880
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.49348072330213e-05,
+      "loss": 0.9528,
+      "step": 6890
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4934116551886563e-05,
+      "loss": 0.8209,
+      "step": 6900
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.493342224745373e-05,
+      "loss": 0.8172,
+      "step": 6910
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4932724320061195e-05,
+      "loss": 1.2846,
+      "step": 6920
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.493202277004912e-05,
+      "loss": 1.0496,
+      "step": 6930
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4931317597759435e-05,
+      "loss": 1.0846,
+      "step": 6940
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4930608803535828e-05,
+      "loss": 1.0522,
+      "step": 6950
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4929896387723756e-05,
+      "loss": 0.9628,
+      "step": 6960
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4929180350670445e-05,
+      "loss": 1.494,
+      "step": 6970
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4928460692724883e-05,
+      "loss": 1.061,
+      "step": 6980
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4927737414237823e-05,
+      "loss": 0.6719,
+      "step": 6990
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4927010515561777e-05,
+      "loss": 1.0067,
+      "step": 7000
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4926279997051033e-05,
+      "loss": 0.9501,
+      "step": 7010
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4925545859061631e-05,
+      "loss": 0.8364,
+      "step": 7020
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4924808101951386e-05,
+      "loss": 0.792,
+      "step": 7030
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4924066726079868e-05,
+      "loss": 0.7958,
+      "step": 7040
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4923321731808417e-05,
+      "loss": 0.9504,
+      "step": 7050
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4922573119500133e-05,
+      "loss": 0.8696,
+      "step": 7060
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.492182088951988e-05,
+      "loss": 1.0283,
+      "step": 7070
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.492106504223429e-05,
+      "loss": 1.0415,
+      "step": 7080
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.492030557801175e-05,
+      "loss": 0.7804,
+      "step": 7090
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4919542497222414e-05,
+      "loss": 1.0154,
+      "step": 7100
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4918775800238197e-05,
+      "loss": 1.1701,
+      "step": 7110
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4918005487432779e-05,
+      "loss": 0.9983,
+      "step": 7120
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4917231559181602e-05,
+      "loss": 0.8283,
+      "step": 7130
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4916454015861869e-05,
+      "loss": 0.8256,
+      "step": 7140
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4915672857852545e-05,
+      "loss": 0.9813,
+      "step": 7150
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4914888085534355e-05,
+      "loss": 0.9492,
+      "step": 7160
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.491409969928979e-05,
+      "loss": 0.8943,
+      "step": 7170
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.49133076995031e-05,
+      "loss": 0.9886,
+      "step": 7180
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4912512086560295e-05,
+      "loss": 1.0587,
+      "step": 7190
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4911712860849147e-05,
+      "loss": 0.9218,
+      "step": 7200
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4910910022759188e-05,
+      "loss": 0.9411,
+      "step": 7210
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4910103572681715e-05,
+      "loss": 0.633,
+      "step": 7220
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4909293511009777e-05,
+      "loss": 0.7656,
+      "step": 7230
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4908479838138192e-05,
+      "loss": 1.0236,
+      "step": 7240
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4907662554463534e-05,
+      "loss": 0.7882,
+      "step": 7250
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4906841660384135e-05,
+      "loss": 0.703,
+      "step": 7260
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4906017156300092e-05,
+      "loss": 0.9532,
+      "step": 7270
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4905189042613253e-05,
+      "loss": 0.8584,
+      "step": 7280
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4904357319727236e-05,
+      "loss": 0.8987,
+      "step": 7290
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.490352198804741e-05,
+      "loss": 0.7435,
+      "step": 7300
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4902683047980905e-05,
+      "loss": 0.7606,
+      "step": 7310
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4901840499936614e-05,
+      "loss": 1.0078,
+      "step": 7320
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.490099434432518e-05,
+      "loss": 1.0276,
+      "step": 7330
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.490014458155901e-05,
+      "loss": 0.9141,
+      "step": 7340
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4899291212052271e-05,
+      "loss": 0.9084,
+      "step": 7350
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4898434236220883e-05,
+      "loss": 1.0924,
+      "step": 7360
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4897573654482523e-05,
+      "loss": 1.0189,
+      "step": 7370
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4896709467256632e-05,
+      "loss": 0.9006,
+      "step": 7380
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4895841674964404e-05,
+      "loss": 0.9912,
+      "step": 7390
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4894970278028787e-05,
+      "loss": 1.0447,
+      "step": 7400
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4894095276874493e-05,
+      "loss": 0.9935,
+      "step": 7410
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4893216671927985e-05,
+      "loss": 0.9729,
+      "step": 7420
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4892334463617486e-05,
+      "loss": 0.926,
+      "step": 7430
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4891448652372972e-05,
+      "loss": 0.9932,
+      "step": 7440
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4890559238626177e-05,
+      "loss": 0.8552,
+      "step": 7450
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.488966622281059e-05,
+      "loss": 0.8495,
+      "step": 7460
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4888769605361456e-05,
+      "loss": 0.7872,
+      "step": 7470
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4887869386715777e-05,
+      "loss": 1.157,
+      "step": 7480
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4886965567312308e-05,
+      "loss": 1.1723,
+      "step": 7490
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.488605814759156e-05,
+      "loss": 0.9442,
+      "step": 7500
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4885147127995799e-05,
+      "loss": 0.8961,
+      "step": 7510
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4884232508969042e-05,
+      "loss": 1.2466,
+      "step": 7520
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4883314290957063e-05,
+      "loss": 0.9751,
+      "step": 7530
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4882392474407394e-05,
+      "loss": 0.6522,
+      "step": 7540
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4881467059769314e-05,
+      "loss": 1.0721,
+      "step": 7550
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4880538047493862e-05,
+      "loss": 0.8575,
+      "step": 7560
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4879605438033825e-05,
+      "loss": 1.1339,
+      "step": 7570
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4878669231843745e-05,
+      "loss": 0.7888,
+      "step": 7580
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4877729429379918e-05,
+      "loss": 0.7422,
+      "step": 7590
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4876786031100395e-05,
+      "loss": 1.1973,
+      "step": 7600
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4875839037464973e-05,
+      "loss": 1.2434,
+      "step": 7610
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4874888448935207e-05,
+      "loss": 0.8379,
+      "step": 7620
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4873934265974401e-05,
+      "loss": 0.7247,
+      "step": 7630
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4872976489047615e-05,
+      "loss": 0.8807,
+      "step": 7640
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4872015118621657e-05,
+      "loss": 0.879,
+      "step": 7650
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4871050155165086e-05,
+      "loss": 1.0314,
+      "step": 7660
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4870081599148213e-05,
+      "loss": 0.8832,
+      "step": 7670
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.48691094510431e-05,
+      "loss": 1.1437,
+      "step": 7680
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4868133711323566e-05,
+      "loss": 1.0185,
+      "step": 7690
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4867154380465167e-05,
+      "loss": 0.963,
+      "step": 7700
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4866171458945218e-05,
+      "loss": 0.7426,
+      "step": 7710
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4865184947242789e-05,
+      "loss": 0.733,
+      "step": 7720
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.486419484583869e-05,
+      "loss": 1.0429,
+      "step": 7730
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.486320115521548e-05,
+      "loss": 1.0294,
+      "step": 7740
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.486220387585748e-05,
+      "loss": 0.8187,
+      "step": 7750
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.4861203008250747e-05,
+      "loss": 1.0844,
+      "step": 7760
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.4860198552883092e-05,
+      "loss": 0.6466,
+      "step": 7770
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.4859190510244076e-05,
+      "loss": 1.0297,
+      "step": 7780
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.4858178880825006e-05,
+      "loss": 1.0666,
+      "step": 7790
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.4857163665118937e-05,
+      "loss": 0.8843,
+      "step": 7800
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.4856144863620673e-05,
+      "loss": 0.9252,
+      "step": 7810
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.4855122476826767e-05,
+      "loss": 0.6275,
+      "step": 7820
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.4854096505235517e-05,
+      "loss": 0.797,
+      "step": 7830
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.4853066949346967e-05,
+      "loss": 0.8827,
+      "step": 7840
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.4852033809662914e-05,
+      "loss": 1.0379,
+      "step": 7850
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.4850997086686896e-05,
+      "loss": 1.122,
+      "step": 7860
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.4849956780924197e-05,
+      "loss": 0.8421,
+      "step": 7870
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.4848912892881852e-05,
+      "loss": 0.7973,
+      "step": 7880
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.4847865423068639e-05,
+      "loss": 1.1859,
+      "step": 7890
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.484681437199508e-05,
+      "loss": 0.9623,
+      "step": 7900
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.4845759740173448e-05,
+      "loss": 1.1052,
+      "step": 7910
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.4844701528117756e-05,
+      "loss": 0.8889,
+      "step": 7920
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.4843639736343764e-05,
+      "loss": 0.6552,
+      "step": 7930
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.4842574365368974e-05,
+      "loss": 0.9871,
+      "step": 7940
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.4841505415712638e-05,
+      "loss": 0.7968,
+      "step": 7950
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.4840432887895748e-05,
+      "loss": 1.1069,
+      "step": 7960
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.483935678244104e-05,
+      "loss": 1.0628,
+      "step": 7970
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.4838277099873e-05,
+      "loss": 0.9716,
+      "step": 7980
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.4837193840717842e-05,
+      "loss": 1.1399,
+      "step": 7990
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.4836107005503543e-05,
+      "loss": 0.9048,
+      "step": 8000
+    },
+    {
+      "epoch": 3.04,
+      "eval_accuracy": 0.7863157894736842,
+      "eval_f1": 0.7863157894736842,
+      "eval_loss": 0.8679137825965881,
+      "eval_runtime": 746.5646,
+      "eval_samples_per_second": 6.362,
+      "eval_steps_per_second": 1.591,
+      "step": 8000
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.4835016594759808e-05,
+      "loss": 0.9414,
+      "step": 8010
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.4833922609018092e-05,
+      "loss": 0.7321,
+      "step": 8020
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.483282504881159e-05,
+      "loss": 1.1217,
+      "step": 8030
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.483172391467524e-05,
+      "loss": 0.9879,
+      "step": 8040
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.483061920714572e-05,
+      "loss": 0.9518,
+      "step": 8050
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.4829510926761451e-05,
+      "loss": 0.7889,
+      "step": 8060
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.4828399074062598e-05,
+      "loss": 0.757,
+      "step": 8070
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.4827283649591061e-05,
+      "loss": 0.9488,
+      "step": 8080
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.4826164653890486e-05,
+      "loss": 0.7085,
+      "step": 8090
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.482504208750626e-05,
+      "loss": 0.8293,
+      "step": 8100
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.4823915950985504e-05,
+      "loss": 0.9005,
+      "step": 8110
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.4822786244877088e-05,
+      "loss": 1.0684,
+      "step": 8120
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.4821652969731612e-05,
+      "loss": 0.8249,
+      "step": 8130
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.4820516126101424e-05,
+      "loss": 0.8295,
+      "step": 8140
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.4819375714540607e-05,
+      "loss": 1.0305,
+      "step": 8150
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.4818231735604982e-05,
+      "loss": 1.0169,
+      "step": 8160
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.4817084189852114e-05,
+      "loss": 0.9314,
+      "step": 8170
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.48159330778413e-05,
+      "loss": 1.0152,
+      "step": 8180
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.4814778400133578e-05,
+      "loss": 0.8649,
+      "step": 8190
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.4813620157291724e-05,
+      "loss": 0.8508,
+      "step": 8200
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.4812458349880255e-05,
+      "loss": 0.8145,
+      "step": 8210
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.4811292978465416e-05,
+      "loss": 1.0982,
+      "step": 8220
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.48101240436152e-05,
+      "loss": 0.6951,
+      "step": 8230
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.480895154589933e-05,
+      "loss": 0.9137,
+      "step": 8240
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.4807775485889265e-05,
+      "loss": 1.2906,
+      "step": 8250
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.4806595864158203e-05,
+      "loss": 0.8748,
+      "step": 8260
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.4805412681281081e-05,
+      "loss": 1.1232,
+      "step": 8270
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.4804225937834564e-05,
+      "loss": 0.8602,
+      "step": 8280
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.4803035634397058e-05,
+      "loss": 0.8962,
+      "step": 8290
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.4801841771548703e-05,
+      "loss": 0.61,
+      "step": 8300
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.4800644349871372e-05,
+      "loss": 0.8235,
+      "step": 8310
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.479944336994867e-05,
+      "loss": 0.9972,
+      "step": 8320
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.479823883236595e-05,
+      "loss": 0.7722,
+      "step": 8330
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.479703073771028e-05,
+      "loss": 0.9794,
+      "step": 8340
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.4795819086570476e-05,
+      "loss": 1.2318,
+      "step": 8350
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.4794603879537076e-05,
+      "loss": 0.6804,
+      "step": 8360
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.4793385117202365e-05,
+      "loss": 0.7993,
+      "step": 8370
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.4792162800160346e-05,
+      "loss": 0.9275,
+      "step": 8380
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.4790936929006766e-05,
+      "loss": 0.9467,
+      "step": 8390
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.4789707504339098e-05,
+      "loss": 0.8108,
+      "step": 8400
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.4788474526756547e-05,
+      "loss": 0.9307,
+      "step": 8410
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.4787237996860056e-05,
+      "loss": 0.7687,
+      "step": 8420
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.478599791525229e-05,
+      "loss": 0.9544,
+      "step": 8430
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.478475428253765e-05,
+      "loss": 0.968,
+      "step": 8440
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.4783507099322271e-05,
+      "loss": 1.0598,
+      "step": 8450
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.4782256366214012e-05,
+      "loss": 0.7067,
+      "step": 8460
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.4781002083822465e-05,
+      "loss": 0.8091,
+      "step": 8470
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.4779744252758955e-05,
+      "loss": 0.8498,
+      "step": 8480
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.4778482873636531e-05,
+      "loss": 1.1503,
+      "step": 8490
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.4777217947069972e-05,
+      "loss": 0.9158,
+      "step": 8500
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.4775949473675794e-05,
+      "loss": 0.7677,
+      "step": 8510
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.477467745407223e-05,
+      "loss": 0.5724,
+      "step": 8520
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.4773401888879247e-05,
+      "loss": 0.919,
+      "step": 8530
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.4772122778718545e-05,
+      "loss": 0.9545,
+      "step": 8540
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.4770840124213543e-05,
+      "loss": 0.9738,
+      "step": 8550
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.476955392598939e-05,
+      "loss": 0.9716,
+      "step": 8560
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.4768264184672965e-05,
+      "loss": 1.2282,
+      "step": 8570
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.4766970900892873e-05,
+      "loss": 0.8605,
+      "step": 8580
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.4765674075279444e-05,
+      "loss": 0.9213,
+      "step": 8590
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.4764373708464733e-05,
+      "loss": 0.7695,
+      "step": 8600
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.4763069801082527e-05,
+      "loss": 0.6323,
+      "step": 8610
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.4761762353768333e-05,
+      "loss": 0.9472,
+      "step": 8620
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.4760451367159385e-05,
+      "loss": 0.7674,
+      "step": 8630
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.475913684189464e-05,
+      "loss": 0.6949,
+      "step": 8640
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.4757818778614786e-05,
+      "loss": 1.1363,
+      "step": 8650
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.4756497177962224e-05,
+      "loss": 0.9669,
+      "step": 8660
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.4755172040581093e-05,
+      "loss": 0.7092,
+      "step": 8670
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.4753843367117248e-05,
+      "loss": 0.937,
+      "step": 8680
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.4752511158218263e-05,
+      "loss": 1.281,
+      "step": 8690
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.4751175414533447e-05,
+      "loss": 0.8582,
+      "step": 8700
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.4749836136713819e-05,
+      "loss": 1.0818,
+      "step": 8710
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.4748493325412132e-05,
+      "loss": 1.1468,
+      "step": 8720
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.4747146981282854e-05,
+      "loss": 0.9654,
+      "step": 8730
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.4745797104982177e-05,
+      "loss": 0.9186,
+      "step": 8740
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.4744443697168013e-05,
+      "loss": 0.9825,
+      "step": 8750
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.4743086758499996e-05,
+      "loss": 0.7791,
+      "step": 8760
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.4741726289639485e-05,
+      "loss": 0.8682,
+      "step": 8770
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.4740362291249555e-05,
+      "loss": 1.0464,
+      "step": 8780
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.4738994763995e-05,
+      "loss": 0.7383,
+      "step": 8790
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.4737623708542336e-05,
+      "loss": 0.9431,
+      "step": 8800
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.47362491255598e-05,
+      "loss": 0.8399,
+      "step": 8810
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.473487101571735e-05,
+      "loss": 0.7528,
+      "step": 8820
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.4733489379686654e-05,
+      "loss": 1.0939,
+      "step": 8830
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.473210421814111e-05,
+      "loss": 1.1854,
+      "step": 8840
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.4730715531755826e-05,
+      "loss": 0.8147,
+      "step": 8850
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.472932332120763e-05,
+      "loss": 0.9854,
+      "step": 8860
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.4727927587175074e-05,
+      "loss": 0.9375,
+      "step": 8870
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.4726528330338416e-05,
+      "loss": 1.1204,
+      "step": 8880
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.4725125551379637e-05,
+      "loss": 0.7848,
+      "step": 8890
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.4723719250982437e-05,
+      "loss": 0.937,
+      "step": 8900
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.4722309429832228e-05,
+      "loss": 0.6796,
+      "step": 8910
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.4720896088616142e-05,
+      "loss": 0.6637,
+      "step": 8920
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.4719479228023022e-05,
+      "loss": 0.7359,
+      "step": 8930
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.471805884874343e-05,
+      "loss": 0.8415,
+      "step": 8940
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.471663495146964e-05,
+      "loss": 0.8642,
+      "step": 8950
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.4715207536895644e-05,
+      "loss": 1.1598,
+      "step": 8960
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.4713776605717146e-05,
+      "loss": 0.9631,
+      "step": 8970
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.4712342158631564e-05,
+      "loss": 1.028,
+      "step": 8980
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.4710904196338032e-05,
+      "loss": 0.9079,
+      "step": 8990
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.4709462719537392e-05,
+      "loss": 0.7848,
+      "step": 9000
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.4708017728932204e-05,
+      "loss": 0.7513,
+      "step": 9010
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.4706569225226741e-05,
+      "loss": 1.0347,
+      "step": 9020
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.470511720912698e-05,
+      "loss": 0.9487,
+      "step": 9030
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.4703661681340624e-05,
+      "loss": 0.649,
+      "step": 9040
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.4702202642577073e-05,
+      "loss": 0.8312,
+      "step": 9050
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.470074009354745e-05,
+      "loss": 0.883,
+      "step": 9060
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.4699274034964577e-05,
+      "loss": 0.635,
+      "step": 9070
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.4697804467543001e-05,
+      "loss": 1.1506,
+      "step": 9080
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.4696331391998966e-05,
+      "loss": 1.3274,
+      "step": 9090
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.4694854809050431e-05,
+      "loss": 0.4753,
+      "step": 9100
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.4693374719417069e-05,
+      "loss": 0.9106,
+      "step": 9110
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.4691891123820253e-05,
+      "loss": 1.1933,
+      "step": 9120
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.469040402298307e-05,
+      "loss": 1.4015,
+      "step": 9130
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.468891341763032e-05,
+      "loss": 0.9833,
+      "step": 9140
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.46874193084885e-05,
+      "loss": 1.1419,
+      "step": 9150
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.4685921696285823e-05,
+      "loss": 0.7536,
+      "step": 9160
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.4684420581752207e-05,
+      "loss": 0.9597,
+      "step": 9170
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.4682915965619275e-05,
+      "loss": 1.0005,
+      "step": 9180
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.4681407848620362e-05,
+      "loss": 1.1734,
+      "step": 9190
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.4679896231490503e-05,
+      "loss": 0.9856,
+      "step": 9200
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.4678381114966447e-05,
+      "loss": 1.2223,
+      "step": 9210
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.4676862499786637e-05,
+      "loss": 0.8492,
+      "step": 9220
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.467534038669123e-05,
+      "loss": 0.8826,
+      "step": 9230
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.4673814776422084e-05,
+      "loss": 0.5519,
+      "step": 9240
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.4672285669722767e-05,
+      "loss": 1.0553,
+      "step": 9250
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.4670753067338543e-05,
+      "loss": 1.1525,
+      "step": 9260
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.4669216970016385e-05,
+      "loss": 1.0257,
+      "step": 9270
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.466767737850497e-05,
+      "loss": 0.8463,
+      "step": 9280
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.4666134293554673e-05,
+      "loss": 0.6491,
+      "step": 9290
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.4664587715917576e-05,
+      "loss": 0.7512,
+      "step": 9300
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.4663037646347467e-05,
+      "loss": 0.7006,
+      "step": 9310
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.4661484085599823e-05,
+      "loss": 0.9646,
+      "step": 9320
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.4659927034431834e-05,
+      "loss": 0.8665,
+      "step": 9330
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.465836649360239e-05,
+      "loss": 0.5898,
+      "step": 9340
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.4656802463872076e-05,
+      "loss": 0.6902,
+      "step": 9350
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.4655234946003185e-05,
+      "loss": 0.8633,
+      "step": 9360
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.4653663940759703e-05,
+      "loss": 0.8003,
+      "step": 9370
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.465208944890732e-05,
+      "loss": 0.6989,
+      "step": 9380
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.4650511471213424e-05,
+      "loss": 1.0922,
+      "step": 9390
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.4648930008447102e-05,
+      "loss": 0.9699,
+      "step": 9400
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.4647345061379142e-05,
+      "loss": 0.894,
+      "step": 9410
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.4645756630782025e-05,
+      "loss": 0.8662,
+      "step": 9420
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.4644164717429931e-05,
+      "loss": 0.7388,
+      "step": 9430
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.4642569322098747e-05,
+      "loss": 0.5905,
+      "step": 9440
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.464097044556604e-05,
+      "loss": 0.8467,
+      "step": 9450
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.463936808861109e-05,
+      "loss": 1.0525,
+      "step": 9460
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.4637762252014863e-05,
+      "loss": 0.8764,
+      "step": 9470
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.4636152936560023e-05,
+      "loss": 0.7288,
+      "step": 9480
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.4634540143030935e-05,
+      "loss": 0.6504,
+      "step": 9490
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.4632923872213653e-05,
+      "loss": 0.8992,
+      "step": 9500
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.4631304124895924e-05,
+      "loss": 0.652,
+      "step": 9510
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.46296809018672e-05,
+      "loss": 0.9174,
+      "step": 9520
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.4628054203918615e-05,
+      "loss": 1.2839,
+      "step": 9530
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.4626424031843006e-05,
+      "loss": 0.9751,
+      "step": 9540
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.4624790386434893e-05,
+      "loss": 0.8526,
+      "step": 9550
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.4623153268490502e-05,
+      "loss": 0.9823,
+      "step": 9560
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.4621512678807738e-05,
+      "loss": 1.1091,
+      "step": 9570
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.461986861818621e-05,
+      "loss": 0.8197,
+      "step": 9580
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.461822108742721e-05,
+      "loss": 0.6894,
+      "step": 9590
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.4616570087333725e-05,
+      "loss": 1.047,
+      "step": 9600
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.4614915618710431e-05,
+      "loss": 0.8924,
+      "step": 9610
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.46132576823637e-05,
+      "loss": 0.8171,
+      "step": 9620
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.4611596279101584e-05,
+      "loss": 0.9181,
+      "step": 9630
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.4609931409733837e-05,
+      "loss": 0.8818,
+      "step": 9640
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.4608263075071894e-05,
+      "loss": 0.8118,
+      "step": 9650
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.4606591275928879e-05,
+      "loss": 0.9907,
+      "step": 9660
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.4604916013119607e-05,
+      "loss": 0.9498,
+      "step": 9670
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.4603237287460582e-05,
+      "loss": 0.7819,
+      "step": 9680
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.4601555099769994e-05,
+      "loss": 0.6889,
+      "step": 9690
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.4599869450867724e-05,
+      "loss": 0.7433,
+      "step": 9700
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.4598180341575332e-05,
+      "loss": 0.8066,
+      "step": 9710
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.459648777271607e-05,
+      "loss": 0.7443,
+      "step": 9720
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.4594791745114878e-05,
+      "loss": 0.6299,
+      "step": 9730
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.4593092259598375e-05,
+      "loss": 1.0822,
+      "step": 9740
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.4591389316994878e-05,
+      "loss": 0.8876,
+      "step": 9750
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.4589682918134371e-05,
+      "loss": 1.0569,
+      "step": 9760
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.4587973063848537e-05,
+      "loss": 0.8454,
+      "step": 9770
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.4586259754970738e-05,
+      "loss": 0.687,
+      "step": 9780
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.4584542992336017e-05,
+      "loss": 1.2345,
+      "step": 9790
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.4582822776781108e-05,
+      "loss": 0.6698,
+      "step": 9800
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.4581099109144421e-05,
+      "loss": 0.6466,
+      "step": 9810
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.457937199026605e-05,
+      "loss": 1.2157,
+      "step": 9820
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.457764142098777e-05,
+      "loss": 0.9521,
+      "step": 9830
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.4575907402153044e-05,
+      "loss": 1.0293,
+      "step": 9840
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.4574169934607006e-05,
+      "loss": 0.9327,
+      "step": 9850
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.4572429019196484e-05,
+      "loss": 0.7715,
+      "step": 9860
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.4570684656769973e-05,
+      "loss": 1.1282,
+      "step": 9870
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.4568936848177657e-05,
+      "loss": 0.9545,
+      "step": 9880
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.4567185594271393e-05,
+      "loss": 0.7993,
+      "step": 9890
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.4565430895904725e-05,
+      "loss": 0.8394,
+      "step": 9900
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.456367275393287e-05,
+      "loss": 0.6563,
+      "step": 9910
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.4561911169212726e-05,
+      "loss": 0.7107,
+      "step": 9920
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.4560146142602868e-05,
+      "loss": 1.2381,
+      "step": 9930
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.4558377674963545e-05,
+      "loss": 0.6681,
+      "step": 9940
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.4556605767156693e-05,
+      "loss": 1.149,
+      "step": 9950
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.455483042004591e-05,
+      "loss": 0.9799,
+      "step": 9960
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.4553051634496486e-05,
+      "loss": 0.9604,
+      "step": 9970
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.4551269411375377e-05,
+      "loss": 0.681,
+      "step": 9980
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.4549483751551216e-05,
+      "loss": 0.8368,
+      "step": 9990
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.4547694655894313e-05,
+      "loss": 0.7977,
+      "step": 10000
+    },
+    {
+      "epoch": 4.04,
+      "eval_accuracy": 0.7810526315789473,
+      "eval_f1": 0.7810526315789474,
+      "eval_loss": 0.8845712542533875,
+      "eval_runtime": 758.7878,
+      "eval_samples_per_second": 6.26,
+      "eval_steps_per_second": 1.566,
+      "step": 10000
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.4545902125276652e-05,
+      "loss": 1.0388,
+      "step": 10010
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.4544106160571887e-05,
+      "loss": 0.9292,
+      "step": 10020
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.4542306762655355e-05,
+      "loss": 0.897,
+      "step": 10030
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.4540503932404057e-05,
+      "loss": 1.0358,
+      "step": 10040
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.453869767069667e-05,
+      "loss": 1.0391,
+      "step": 10050
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.4536887978413547e-05,
+      "loss": 0.7696,
+      "step": 10060
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.4535074856436707e-05,
+      "loss": 0.6376,
+      "step": 10070
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.4533258305649845e-05,
+      "loss": 0.8613,
+      "step": 10080
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.4531438326938328e-05,
+      "loss": 0.8559,
+      "step": 10090
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.4529614921189187e-05,
+      "loss": 0.6378,
+      "step": 10100
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.452778808929113e-05,
+      "loss": 0.6774,
+      "step": 10110
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.4525957832134532e-05,
+      "loss": 0.6826,
+      "step": 10120
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.4524124150611443e-05,
+      "loss": 0.7114,
+      "step": 10130
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.452228704561557e-05,
+      "loss": 0.8111,
+      "step": 10140
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.4520446518042301e-05,
+      "loss": 0.8248,
+      "step": 10150
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.4518602568788683e-05,
+      "loss": 0.8909,
+      "step": 10160
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.4516755198753435e-05,
+      "loss": 1.0263,
+      "step": 10170
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.4514904408836944e-05,
+      "loss": 0.7914,
+      "step": 10180
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.4513050199941261e-05,
+      "loss": 0.9401,
+      "step": 10190
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.4511192572970108e-05,
+      "loss": 0.7783,
+      "step": 10200
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.4509331528828868e-05,
+      "loss": 0.6487,
+      "step": 10210
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.4507467068424591e-05,
+      "loss": 0.7832,
+      "step": 10220
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.4505599192665993e-05,
+      "loss": 0.7375,
+      "step": 10230
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.4503727902463451e-05,
+      "loss": 0.8463,
+      "step": 10240
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.4501853198729012e-05,
+      "loss": 0.7524,
+      "step": 10250
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.4499975082376387e-05,
+      "loss": 0.5843,
+      "step": 10260
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.449809355432094e-05,
+      "loss": 1.342,
+      "step": 10270
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.449620861547971e-05,
+      "loss": 0.8964,
+      "step": 10280
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.4494320266771391e-05,
+      "loss": 0.8378,
+      "step": 10290
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.4492428509116341e-05,
+      "loss": 0.8489,
+      "step": 10300
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.4490533343436581e-05,
+      "loss": 0.5495,
+      "step": 10310
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.4488634770655793e-05,
+      "loss": 0.9422,
+      "step": 10320
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.4486732791699318e-05,
+      "loss": 1.0297,
+      "step": 10330
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.4484827407494154e-05,
+      "loss": 0.8256,
+      "step": 10340
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.4482918618968963e-05,
+      "loss": 0.5982,
+      "step": 10350
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.4481006427054067e-05,
+      "loss": 0.7247,
+      "step": 10360
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.4479090832681445e-05,
+      "loss": 1.0779,
+      "step": 10370
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.4477171836784736e-05,
+      "loss": 0.7428,
+      "step": 10380
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.4475249440299231e-05,
+      "loss": 0.6429,
+      "step": 10390
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.4473323644161886e-05,
+      "loss": 0.557,
+      "step": 10400
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.4471394449311309e-05,
+      "loss": 0.7514,
+      "step": 10410
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.4469461856687769e-05,
+      "loss": 1.0773,
+      "step": 10420
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.4467525867233184e-05,
+      "loss": 1.1142,
+      "step": 10430
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.4465586481891134e-05,
+      "loss": 0.8702,
+      "step": 10440
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.4463643701606852e-05,
+      "loss": 1.0078,
+      "step": 10450
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.4461697527327225e-05,
+      "loss": 0.6047,
+      "step": 10460
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.4459747960000795e-05,
+      "loss": 0.9054,
+      "step": 10470
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.4457795000577756e-05,
+      "loss": 0.8874,
+      "step": 10480
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.4455838650009962e-05,
+      "loss": 1.1414,
+      "step": 10490
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.4453878909250906e-05,
+      "loss": 0.8461,
+      "step": 10500
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.4451915779255748e-05,
+      "loss": 1.2347,
+      "step": 10510
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.4449949260981291e-05,
+      "loss": 0.9733,
+      "step": 10520
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.4447979355385994e-05,
+      "loss": 0.7622,
+      "step": 10530
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.4446006063429966e-05,
+      "loss": 1.0488,
+      "step": 10540
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.4444029386074961e-05,
+      "loss": 0.6356,
+      "step": 10550
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.4442049324284393e-05,
+      "loss": 0.9436,
+      "step": 10560
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.4440065879023313e-05,
+      "loss": 0.9486,
+      "step": 10570
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.4438079051258435e-05,
+      "loss": 1.1029,
+      "step": 10580
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.4436088841958113e-05,
+      "loss": 1.1475,
+      "step": 10590
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.443409525209235e-05,
+      "loss": 0.7903,
+      "step": 10600
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.4432098282632795e-05,
+      "loss": 0.6607,
+      "step": 10610
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.4430097934552751e-05,
+      "loss": 0.6229,
+      "step": 10620
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.4428094208827161e-05,
+      "loss": 1.006,
+      "step": 10630
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.4426087106432617e-05,
+      "loss": 0.7741,
+      "step": 10640
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.4424076628347357e-05,
+      "loss": 0.5454,
+      "step": 10650
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.4422062775551262e-05,
+      "loss": 0.7863,
+      "step": 10660
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.4420045549025862e-05,
+      "loss": 0.977,
+      "step": 10670
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.4418024949754326e-05,
+      "loss": 0.956,
+      "step": 10680
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.441600097872147e-05,
+      "loss": 0.9236,
+      "step": 10690
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.4413973636913754e-05,
+      "loss": 0.9959,
+      "step": 10700
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.441194292531928e-05,
+      "loss": 1.2392,
+      "step": 10710
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.4409908844927792e-05,
+      "loss": 0.6514,
+      "step": 10720
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.4407871396730672e-05,
+      "loss": 0.9126,
+      "step": 10730
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.4405830581720953e-05,
+      "loss": 0.9703,
+      "step": 10740
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.4403786400893304e-05,
+      "loss": 0.6085,
+      "step": 10750
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.4401738855244029e-05,
+      "loss": 0.7244,
+      "step": 10760
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.439968794577108e-05,
+      "loss": 0.8882,
+      "step": 10770
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.4397633673474042e-05,
+      "loss": 0.584,
+      "step": 10780
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.4395576039354148e-05,
+      "loss": 1.2507,
+      "step": 10790
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.4393515044414259e-05,
+      "loss": 0.8442,
+      "step": 10800
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.439145068965888e-05,
+      "loss": 1.0842,
+      "step": 10810
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.4389382976094155e-05,
+      "loss": 0.9413,
+      "step": 10820
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.438731190472786e-05,
+      "loss": 0.655,
+      "step": 10830
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.438523747656941e-05,
+      "loss": 0.8029,
+      "step": 10840
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.4383159692629858e-05,
+      "loss": 0.5417,
+      "step": 10850
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.4381078553921888e-05,
+      "loss": 1.0177,
+      "step": 10860
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.4378994061459826e-05,
+      "loss": 0.9567,
+      "step": 10870
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.4376906216259623e-05,
+      "loss": 0.7705,
+      "step": 10880
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.4374815019338873e-05,
+      "loss": 0.6263,
+      "step": 10890
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.4372720471716797e-05,
+      "loss": 0.5253,
+      "step": 10900
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.4370622574414254e-05,
+      "loss": 0.8735,
+      "step": 10910
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.4368521328453736e-05,
+      "loss": 0.916,
+      "step": 10920
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.4366416734859362e-05,
+      "loss": 0.8634,
+      "step": 10930
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.4364308794656881e-05,
+      "loss": 0.7812,
+      "step": 10940
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.4362197508873688e-05,
+      "loss": 0.7478,
+      "step": 10950
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.4360082878538787e-05,
+      "loss": 0.7418,
+      "step": 10960
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.4357964904682832e-05,
+      "loss": 0.6585,
+      "step": 10970
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.4355843588338092e-05,
+      "loss": 0.8206,
+      "step": 10980
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.4353718930538473e-05,
+      "loss": 0.6967,
+      "step": 10990
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.4351590932319506e-05,
+      "loss": 0.8266,
+      "step": 11000
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.4349459594718354e-05,
+      "loss": 0.8005,
+      "step": 11010
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.4347324918773805e-05,
+      "loss": 0.9922,
+      "step": 11020
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.4345186905526272e-05,
+      "loss": 0.9101,
+      "step": 11030
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.4343045556017798e-05,
+      "loss": 0.8269,
+      "step": 11040
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.4340900871292047e-05,
+      "loss": 1.0536,
+      "step": 11050
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.433875285239432e-05,
+      "loss": 0.8667,
+      "step": 11060
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.4336601500371527e-05,
+      "loss": 0.8843,
+      "step": 11070
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.4334446816272218e-05,
+      "loss": 0.7773,
+      "step": 11080
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.4332288801146554e-05,
+      "loss": 1.0272,
+      "step": 11090
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.4330127456046328e-05,
+      "loss": 0.739,
+      "step": 11100
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.4327962782024956e-05,
+      "loss": 0.7335,
+      "step": 11110
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.4325794780137468e-05,
+      "loss": 0.7412,
+      "step": 11120
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.4323623451440525e-05,
+      "loss": 1.011,
+      "step": 11130
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.4321448796992409e-05,
+      "loss": 0.9723,
+      "step": 11140
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.4319270817853014e-05,
+      "loss": 1.0028,
+      "step": 11150
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.4317089515083866e-05,
+      "loss": 0.7306,
+      "step": 11160
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.4314904889748102e-05,
+      "loss": 0.9448,
+      "step": 11170
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.4312716942910483e-05,
+      "loss": 0.617,
+      "step": 11180
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.4310525675637389e-05,
+      "loss": 0.7878,
+      "step": 11190
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.4308331088996816e-05,
+      "loss": 1.0331,
+      "step": 11200
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.4306133184058378e-05,
+      "loss": 0.6705,
+      "step": 11210
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.4303931961893309e-05,
+      "loss": 0.7288,
+      "step": 11220
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.4301727423574453e-05,
+      "loss": 1.2383,
+      "step": 11230
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.4299519570176284e-05,
+      "loss": 0.7154,
+      "step": 11240
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.4297308402774876e-05,
+      "loss": 0.6404,
+      "step": 11250
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.4295093922447927e-05,
+      "loss": 0.9904,
+      "step": 11260
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.4292876130274747e-05,
+      "loss": 1.0343,
+      "step": 11270
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.4290655027336264e-05,
+      "loss": 0.7563,
+      "step": 11280
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.428843061471501e-05,
+      "loss": 1.0234,
+      "step": 11290
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.4286202893495147e-05,
+      "loss": 0.6342,
+      "step": 11300
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.428397186476243e-05,
+      "loss": 0.7474,
+      "step": 11310
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.428173752960424e-05,
+      "loss": 0.7082,
+      "step": 11320
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.4279499889109563e-05,
+      "loss": 0.9103,
+      "step": 11330
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.4277258944369001e-05,
+      "loss": 0.6227,
+      "step": 11340
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.4275014696474758e-05,
+      "loss": 0.5647,
+      "step": 11350
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.4272767146520655e-05,
+      "loss": 0.9027,
+      "step": 11360
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.4270516295602122e-05,
+      "loss": 0.7919,
+      "step": 11370
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.4268262144816196e-05,
+      "loss": 0.831,
+      "step": 11380
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.426600469526152e-05,
+      "loss": 0.7439,
+      "step": 11390
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.4263743948038355e-05,
+      "loss": 0.9817,
+      "step": 11400
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.4261479904248552e-05,
+      "loss": 0.6809,
+      "step": 11410
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.4259212564995586e-05,
+      "loss": 0.852,
+      "step": 11420
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.4256941931384526e-05,
+      "loss": 0.6747,
+      "step": 11430
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.4254668004522053e-05,
+      "loss": 1.0233,
+      "step": 11440
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.4252390785516453e-05,
+      "loss": 0.9007,
+      "step": 11450
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.4250110275477612e-05,
+      "loss": 0.9747,
+      "step": 11460
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.4247826475517023e-05,
+      "loss": 1.2008,
+      "step": 11470
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.4245539386747784e-05,
+      "loss": 0.782,
+      "step": 11480
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.4243249010284593e-05,
+      "loss": 0.6924,
+      "step": 11490
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.4240955347243754e-05,
+      "loss": 0.8578,
+      "step": 11500
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.4238658398743167e-05,
+      "loss": 0.6872,
+      "step": 11510
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.4236358165902338e-05,
+      "loss": 0.5779,
+      "step": 11520
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.4234054649842377e-05,
+      "loss": 0.8016,
+      "step": 11530
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.4231747851685982e-05,
+      "loss": 0.7191,
+      "step": 11540
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.4229437772557463e-05,
+      "loss": 0.8966,
+      "step": 11550
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.4227124413582726e-05,
+      "loss": 0.8387,
+      "step": 11560
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.422480777588927e-05,
+      "loss": 0.6938,
+      "step": 11570
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.4222487860606197e-05,
+      "loss": 0.964,
+      "step": 11580
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.4220164668864207e-05,
+      "loss": 0.831,
+      "step": 11590
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.4217838201795596e-05,
+      "loss": 1.0617,
+      "step": 11600
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.4215508460534254e-05,
+      "loss": 0.7678,
+      "step": 11610
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.4213175446215669e-05,
+      "loss": 1.1968,
+      "step": 11620
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.4210839159976927e-05,
+      "loss": 0.8451,
+      "step": 11630
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.4208499602956699e-05,
+      "loss": 0.803,
+      "step": 11640
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.420615677629526e-05,
+      "loss": 0.5882,
+      "step": 11650
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.4203810681134479e-05,
+      "loss": 1.0378,
+      "step": 11660
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.4201461318617807e-05,
+      "loss": 0.8273,
+      "step": 11670
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.4199108689890303e-05,
+      "loss": 0.8917,
+      "step": 11680
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.4196752796098601e-05,
+      "loss": 0.8185,
+      "step": 11690
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.4194393638390943e-05,
+      "loss": 0.7411,
+      "step": 11700
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.4192031217917148e-05,
+      "loss": 0.649,
+      "step": 11710
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.4189665535828631e-05,
+      "loss": 1.0503,
+      "step": 11720
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.41872965932784e-05,
+      "loss": 0.9669,
+      "step": 11730
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.418492439142105e-05,
+      "loss": 0.7626,
+      "step": 11740
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.4182548931412758e-05,
+      "loss": 0.7923,
+      "step": 11750
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.41801702144113e-05,
+      "loss": 0.9244,
+      "step": 11760
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.417778824157603e-05,
+      "loss": 1.2815,
+      "step": 11770
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.4175403014067892e-05,
+      "loss": 1.0134,
+      "step": 11780
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.4173014533049422e-05,
+      "loss": 0.9197,
+      "step": 11790
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.4170622799684732e-05,
+      "loss": 0.9558,
+      "step": 11800
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.4168227815139526e-05,
+      "loss": 0.9409,
+      "step": 11810
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.4165829580581085e-05,
+      "loss": 0.861,
+      "step": 11820
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.416342809717829e-05,
+      "loss": 0.7087,
+      "step": 11830
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.4161023366101585e-05,
+      "loss": 1.0238,
+      "step": 11840
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.415861538852301e-05,
+      "loss": 0.782,
+      "step": 11850
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.4156204165616188e-05,
+      "loss": 0.6829,
+      "step": 11860
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.4153789698556311e-05,
+      "loss": 0.7065,
+      "step": 11870
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.4151371988520169e-05,
+      "loss": 0.7532,
+      "step": 11880
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.414895103668612e-05,
+      "loss": 0.4863,
+      "step": 11890
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.414652684423411e-05,
+      "loss": 1.1852,
+      "step": 11900
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.414409941234566e-05,
+      "loss": 0.7895,
+      "step": 11910
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.4141668742203868e-05,
+      "loss": 0.6872,
+      "step": 11920
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.4139234834993416e-05,
+      "loss": 1.1803,
+      "step": 11930
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.4136797691900557e-05,
+      "loss": 0.5786,
+      "step": 11940
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.413435731411313e-05,
+      "loss": 0.8615,
+      "step": 11950
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.4131913702820543e-05,
+      "loss": 0.7935,
+      "step": 11960
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.4129466859213782e-05,
+      "loss": 0.5791,
+      "step": 11970
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.4127016784485411e-05,
+      "loss": 1.2098,
+      "step": 11980
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.4124563479829562e-05,
+      "loss": 1.1726,
+      "step": 11990
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.4122106946441953e-05,
+      "loss": 0.9259,
+      "step": 12000
+    },
+    {
+      "epoch": 5.04,
+      "eval_accuracy": 0.8263157894736842,
+      "eval_f1": 0.8263157894736841,
+      "eval_loss": 0.8018165230751038,
+      "eval_runtime": 741.1315,
+      "eval_samples_per_second": 6.409,
+      "eval_steps_per_second": 1.603,
+      "step": 12000
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.4119647185519863e-05,
+      "loss": 0.9292,
+      "step": 12010
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.4117184198262151e-05,
+      "loss": 0.9956,
+      "step": 12020
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.4114717985869247e-05,
+      "loss": 0.9384,
+      "step": 12030
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.4112248549543151e-05,
+      "loss": 0.9028,
+      "step": 12040
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.410977589048744e-05,
+      "loss": 0.8509,
+      "step": 12050
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.410730000990726e-05,
+      "loss": 0.6837,
+      "step": 12060
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.4104820909009319e-05,
+      "loss": 0.6524,
+      "step": 12070
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.4102338589001901e-05,
+      "loss": 0.6301,
+      "step": 12080
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.4099853051094864e-05,
+      "loss": 0.8283,
+      "step": 12090
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.4097364296499624e-05,
+      "loss": 1.0106,
+      "step": 12100
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.409487232642917e-05,
+      "loss": 0.8437,
+      "step": 12110
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.4092377142098065e-05,
+      "loss": 0.7118,
+      "step": 12120
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.4089878744722421e-05,
+      "loss": 1.087,
+      "step": 12130
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.4087377135519934e-05,
+      "loss": 0.54,
+      "step": 12140
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.4084872315709853e-05,
+      "loss": 0.785,
+      "step": 12150
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.4082364286513003e-05,
+      "loss": 0.509,
+      "step": 12160
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.4079853049151762e-05,
+      "loss": 0.8785,
+      "step": 12170
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.4077338604850075e-05,
+      "loss": 0.9986,
+      "step": 12180
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.4074820954833457e-05,
+      "loss": 0.8,
+      "step": 12190
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.4072300100328976e-05,
+      "loss": 0.8324,
+      "step": 12200
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.4069776042565269e-05,
+      "loss": 0.3989,
+      "step": 12210
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.406724878277253e-05,
+      "loss": 0.9609,
+      "step": 12220
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.4064718322182512e-05,
+      "loss": 0.9633,
+      "step": 12230
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.4062184662028534e-05,
+      "loss": 0.8126,
+      "step": 12240
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.4059647803545468e-05,
+      "loss": 0.5011,
+      "step": 12250
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.4057107747969753e-05,
+      "loss": 0.5289,
+      "step": 12260
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.4054564496539378e-05,
+      "loss": 1.0525,
+      "step": 12270
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.4052018050493892e-05,
+      "loss": 0.7972,
+      "step": 12280
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.4049468411074402e-05,
+      "loss": 0.7144,
+      "step": 12290
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.4046915579523573e-05,
+      "loss": 0.6966,
+      "step": 12300
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.4044359557085624e-05,
+      "loss": 0.9538,
+      "step": 12310
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.4041800345006328e-05,
+      "loss": 0.7337,
+      "step": 12320
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.4039237944533015e-05,
+      "loss": 0.8238,
+      "step": 12330
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.4036672356914567e-05,
+      "loss": 0.7472,
+      "step": 12340
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.4034103583401422e-05,
+      "loss": 0.7119,
+      "step": 12350
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.4031531625245567e-05,
+      "loss": 0.7508,
+      "step": 12360
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.4028956483700542e-05,
+      "loss": 0.8372,
+      "step": 12370
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.4026378160021442e-05,
+      "loss": 0.8902,
+      "step": 12380
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.4023796655464912e-05,
+      "loss": 0.666,
+      "step": 12390
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.4021211971289142e-05,
+      "loss": 0.9766,
+      "step": 12400
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.401862410875388e-05,
+      "loss": 0.9295,
+      "step": 12410
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.4016033069120414e-05,
+      "loss": 0.8617,
+      "step": 12420
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.4013438853651591e-05,
+      "loss": 0.6898,
+      "step": 12430
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.4010841463611795e-05,
+      "loss": 0.6646,
+      "step": 12440
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.4008240900266964e-05,
+      "loss": 0.4307,
+      "step": 12450
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.400563716488458e-05,
+      "loss": 0.7693,
+      "step": 12460
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.4003030258733676e-05,
+      "loss": 0.735,
+      "step": 12470
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.4000420183084823e-05,
+      "loss": 0.818,
+      "step": 12480
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.3997806939210139e-05,
+      "loss": 1.0814,
+      "step": 12490
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.3995190528383292e-05,
+      "loss": 1.1466,
+      "step": 12500
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.3992570951879483e-05,
+      "loss": 0.8816,
+      "step": 12510
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.3989948210975466e-05,
+      "loss": 0.8825,
+      "step": 12520
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.3987322306949532e-05,
+      "loss": 0.9807,
+      "step": 12530
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.3984693241081512e-05,
+      "loss": 0.8425,
+      "step": 12540
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.3982061014652787e-05,
+      "loss": 0.8364,
+      "step": 12550
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.3979425628946263e-05,
+      "loss": 0.8375,
+      "step": 12560
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.3976787085246405e-05,
+      "loss": 0.7715,
+      "step": 12570
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.3974145384839203e-05,
+      "loss": 0.7157,
+      "step": 12580
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.3971500529012188e-05,
+      "loss": 0.8325,
+      "step": 12590
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.396885251905443e-05,
+      "loss": 0.5132,
+      "step": 12600
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.3966201356256543e-05,
+      "loss": 1.0001,
+      "step": 12610
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.3963547041910663e-05,
+      "loss": 0.8268,
+      "step": 12620
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.3960889577310476e-05,
+      "loss": 0.7412,
+      "step": 12630
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.3958228963751197e-05,
+      "loss": 0.5222,
+      "step": 12640
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.3955565202529577e-05,
+      "loss": 0.6424,
+      "step": 12650
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.39528982949439e-05,
+      "loss": 0.9083,
+      "step": 12660
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.3950228242293985e-05,
+      "loss": 0.6788,
+      "step": 12670
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.3947555045881183e-05,
+      "loss": 0.9128,
+      "step": 12680
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.3944878707008378e-05,
+      "loss": 1.0018,
+      "step": 12690
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.3942199226979984e-05,
+      "loss": 0.9426,
+      "step": 12700
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.3939516607101947e-05,
+      "loss": 0.7346,
+      "step": 12710
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.3936830848681743e-05,
+      "loss": 0.7292,
+      "step": 12720
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.393414195302838e-05,
+      "loss": 0.9039,
+      "step": 12730
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.3931449921452392e-05,
+      "loss": 0.4155,
+      "step": 12740
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.3928754755265844e-05,
+      "loss": 0.5747,
+      "step": 12750
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.3926056455782322e-05,
+      "loss": 0.7444,
+      "step": 12760
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.392335502431695e-05,
+      "loss": 1.033,
+      "step": 12770
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.392065046218637e-05,
+      "loss": 0.5894,
+      "step": 12780
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.3917942770708757e-05,
+      "loss": 0.7544,
+      "step": 12790
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.39152319512038e-05,
+      "loss": 0.6563,
+      "step": 12800
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.3912518004992724e-05,
+      "loss": 0.7536,
+      "step": 12810
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.3909800933398273e-05,
+      "loss": 0.577,
+      "step": 12820
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.3907080737744714e-05,
+      "loss": 0.6801,
+      "step": 12830
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.3904357419357838e-05,
+      "loss": 0.8534,
+      "step": 12840
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.3901630979564955e-05,
+      "loss": 0.9727,
+      "step": 12850
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.3898901419694903e-05,
+      "loss": 0.8376,
+      "step": 12860
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.3896168741078033e-05,
+      "loss": 0.9828,
+      "step": 12870
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.3893432945046219e-05,
+      "loss": 0.81,
+      "step": 12880
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.3890694032932857e-05,
+      "loss": 0.9423,
+      "step": 12890
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.3887952006072857e-05,
+      "loss": 0.9811,
+      "step": 12900
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.388520686580265e-05,
+      "loss": 1.0064,
+      "step": 12910
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.388245861346019e-05,
+      "loss": 1.0164,
+      "step": 12920
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.3879707250384934e-05,
+      "loss": 0.6345,
+      "step": 12930
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.3876952777917864e-05,
+      "loss": 0.8114,
+      "step": 12940
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.387419519740148e-05,
+      "loss": 0.8133,
+      "step": 12950
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.3871434510179791e-05,
+      "loss": 0.7562,
+      "step": 12960
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.3868670717598323e-05,
+      "loss": 0.9805,
+      "step": 12970
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.3865903821004115e-05,
+      "loss": 0.7079,
+      "step": 12980
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.3863133821745717e-05,
+      "loss": 0.9933,
+      "step": 12990
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.3860360721173195e-05,
+      "loss": 0.5244,
+      "step": 13000
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.3857584520638124e-05,
+      "loss": 0.6868,
+      "step": 13010
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.3854805221493592e-05,
+      "loss": 0.7727,
+      "step": 13020
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.3852022825094192e-05,
+      "loss": 1.0346,
+      "step": 13030
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.3849237332796034e-05,
+      "loss": 0.9681,
+      "step": 13040
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.384644874595673e-05,
+      "loss": 0.824,
+      "step": 13050
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.3843657065935406e-05,
+      "loss": 0.9007,
+      "step": 13060
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.3840862294092691e-05,
+      "loss": 1.0339,
+      "step": 13070
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.3838064431790724e-05,
+      "loss": 0.8334,
+      "step": 13080
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.3835263480393149e-05,
+      "loss": 0.9044,
+      "step": 13090
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.3832459441265114e-05,
+      "loss": 0.8958,
+      "step": 13100
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.3829652315773276e-05,
+      "loss": 0.8666,
+      "step": 13110
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.3826842105285792e-05,
+      "loss": 0.9741,
+      "step": 13120
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.3824028811172325e-05,
+      "loss": 0.8652,
+      "step": 13130
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.3821212434804042e-05,
+      "loss": 0.9392,
+      "step": 13140
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.3818392977553607e-05,
+      "loss": 0.658,
+      "step": 13150
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.3815570440795194e-05,
+      "loss": 0.636,
+      "step": 13160
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.3812744825904467e-05,
+      "loss": 0.8125,
+      "step": 13170
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.3809916134258603e-05,
+      "loss": 0.5889,
+      "step": 13180
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.3807084367236269e-05,
+      "loss": 0.6374,
+      "step": 13190
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.3804249526217633e-05,
+      "loss": 0.9499,
+      "step": 13200
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.3801411612584363e-05,
+      "loss": 0.9158,
+      "step": 13210
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.3798570627719622e-05,
+      "loss": 0.4447,
+      "step": 13220
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.3795726573008075e-05,
+      "loss": 0.976,
+      "step": 13230
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.3792879449835875e-05,
+      "loss": 0.7793,
+      "step": 13240
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.3790029259590681e-05,
+      "loss": 0.9342,
+      "step": 13250
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.3787176003661635e-05,
+      "loss": 0.6184,
+      "step": 13260
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.3784319683439385e-05,
+      "loss": 0.7955,
+      "step": 13270
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.3781460300316064e-05,
+      "loss": 0.906,
+      "step": 13280
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.37785978556853e-05,
+      "loss": 0.9102,
+      "step": 13290
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.3775732350942213e-05,
+      "loss": 0.9564,
+      "step": 13300
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.3772863787483418e-05,
+      "loss": 0.734,
+      "step": 13310
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.3769992166707014e-05,
+      "loss": 0.8327,
+      "step": 13320
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.3767117490012603e-05,
+      "loss": 0.6923,
+      "step": 13330
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.3764239758801257e-05,
+      "loss": 0.692,
+      "step": 13340
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.376135897447555e-05,
+      "loss": 0.8945,
+      "step": 13350
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.3758475138439543e-05,
+      "loss": 0.6951,
+      "step": 13360
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.3755588252098785e-05,
+      "loss": 0.6342,
+      "step": 13370
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.3752698316860305e-05,
+      "loss": 0.8719,
+      "step": 13380
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.3749805334132624e-05,
+      "loss": 0.7696,
+      "step": 13390
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.3746909305325747e-05,
+      "loss": 0.721,
+      "step": 13400
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.3744010231851161e-05,
+      "loss": 0.5691,
+      "step": 13410
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.3741108115121844e-05,
+      "loss": 1.035,
+      "step": 13420
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.373820295655225e-05,
+      "loss": 0.7772,
+      "step": 13430
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.3735294757558315e-05,
+      "loss": 0.9524,
+      "step": 13440
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.3732383519557461e-05,
+      "loss": 1.0839,
+      "step": 13450
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.3729469243968596e-05,
+      "loss": 0.4841,
+      "step": 13460
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.3726551932212094e-05,
+      "loss": 0.7338,
+      "step": 13470
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.3723631585709822e-05,
+      "loss": 1.0517,
+      "step": 13480
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.3720708205885125e-05,
+      "loss": 0.9323,
+      "step": 13490
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.3717781794162813e-05,
+      "loss": 0.6849,
+      "step": 13500
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.371485235196919e-05,
+      "loss": 0.3872,
+      "step": 13510
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.3711919880732033e-05,
+      "loss": 0.9672,
+      "step": 13520
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.3708984381880584e-05,
+      "loss": 0.6098,
+      "step": 13530
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.3706045856845579e-05,
+      "loss": 0.7626,
+      "step": 13540
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.3703104307059213e-05,
+      "loss": 0.8963,
+      "step": 13550
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.3700159733955166e-05,
+      "loss": 0.7506,
+      "step": 13560
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.3697212138968584e-05,
+      "loss": 0.9264,
+      "step": 13570
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.369426152353609e-05,
+      "loss": 0.787,
+      "step": 13580
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.3691307889095778e-05,
+      "loss": 0.7793,
+      "step": 13590
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.3688351237087214e-05,
+      "loss": 0.8481,
+      "step": 13600
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.3685391568951434e-05,
+      "loss": 0.8792,
+      "step": 13610
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.3682428886130944e-05,
+      "loss": 0.888,
+      "step": 13620
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.367946319006972e-05,
+      "loss": 0.7437,
+      "step": 13630
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.3676494482213206e-05,
+      "loss": 1.0551,
+      "step": 13640
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.3673522764008315e-05,
+      "loss": 0.9655,
+      "step": 13650
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.3670548036903425e-05,
+      "loss": 0.8299,
+      "step": 13660
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.3667570302348384e-05,
+      "loss": 0.9885,
+      "step": 13670
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.3664589561794498e-05,
+      "loss": 0.6566,
+      "step": 13680
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.3661605816694551e-05,
+      "loss": 0.6017,
+      "step": 13690
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.365861906850278e-05,
+      "loss": 0.6448,
+      "step": 13700
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.3655629318674892e-05,
+      "loss": 0.8787,
+      "step": 13710
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.3652636568668053e-05,
+      "loss": 0.8334,
+      "step": 13720
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.3649640819940891e-05,
+      "loss": 1.1023,
+      "step": 13730
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.3646642073953505e-05,
+      "loss": 0.5216,
+      "step": 13740
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.364364033216744e-05,
+      "loss": 1.0398,
+      "step": 13750
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.3640635596045708e-05,
+      "loss": 0.9455,
+      "step": 13760
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.3637627867052786e-05,
+      "loss": 0.6643,
+      "step": 13770
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.3634617146654605e-05,
+      "loss": 0.8005,
+      "step": 13780
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.3631603436318548e-05,
+      "loss": 1.2044,
+      "step": 13790
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.3628586737513463e-05,
+      "loss": 1.0157,
+      "step": 13800
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.3625567051709656e-05,
+      "loss": 0.4879,
+      "step": 13810
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.362254438037888e-05,
+      "loss": 0.6704,
+      "step": 13820
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.3619518724994351e-05,
+      "loss": 0.8101,
+      "step": 13830
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.3616490087030738e-05,
+      "loss": 0.6362,
+      "step": 13840
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.3613458467964156e-05,
+      "loss": 0.9716,
+      "step": 13850
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.3610423869272188e-05,
+      "loss": 0.6861,
+      "step": 13860
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.3607386292433854e-05,
+      "loss": 1.2453,
+      "step": 13870
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.3604345738929636e-05,
+      "loss": 0.7002,
+      "step": 13880
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.360130221024146e-05,
+      "loss": 0.7556,
+      "step": 13890
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.3598255707852707e-05,
+      "loss": 0.7298,
+      "step": 13900
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.3595206233248204e-05,
+      "loss": 0.7313,
+      "step": 13910
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.3592153787914228e-05,
+      "loss": 0.4009,
+      "step": 13920
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.3589098373338507e-05,
+      "loss": 0.829,
+      "step": 13930
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.3586039991010208e-05,
+      "loss": 0.6719,
+      "step": 13940
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.358297864241995e-05,
+      "loss": 0.5607,
+      "step": 13950
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.35799143290598e-05,
+      "loss": 0.6292,
+      "step": 13960
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.3576847052423266e-05,
+      "loss": 0.9576,
+      "step": 13970
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.35737768140053e-05,
+      "loss": 1.0345,
+      "step": 13980
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.35707036153023e-05,
+      "loss": 0.4929,
+      "step": 13990
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.3567627457812107e-05,
+      "loss": 0.6077,
+      "step": 14000
+    },
+    {
+      "epoch": 6.04,
+      "eval_accuracy": 0.8189473684210526,
+      "eval_f1": 0.8189473684210526,
+      "eval_loss": 0.8212071061134338,
+      "eval_runtime": 747.465,
+      "eval_samples_per_second": 6.355,
+      "eval_steps_per_second": 1.589,
+      "step": 14000
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.3564548343034e-05,
+      "loss": 0.8925,
+      "step": 14010
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.3561466272468704e-05,
+      "loss": 1.021,
+      "step": 14020
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.3558381247618381e-05,
+      "loss": 0.8947,
+      "step": 14030
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.3555293269986639e-05,
+      "loss": 0.7799,
+      "step": 14040
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.3552202341078515e-05,
+      "loss": 0.7529,
+      "step": 14050
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.3549108462400494e-05,
+      "loss": 0.8222,
+      "step": 14060
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.3546011635460494e-05,
+      "loss": 1.1847,
+      "step": 14070
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.354291186176787e-05,
+      "loss": 0.7235,
+      "step": 14080
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.3539809142833414e-05,
+      "loss": 0.8463,
+      "step": 14090
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.3536703480169356e-05,
+      "loss": 0.512,
+      "step": 14100
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.3533594875289357e-05,
+      "loss": 0.8988,
+      "step": 14110
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.353048332970851e-05,
+      "loss": 0.8714,
+      "step": 14120
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.3527368844943349e-05,
+      "loss": 0.8919,
+      "step": 14130
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.3524251422511834e-05,
+      "loss": 0.6822,
+      "step": 14140
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.3521131063933359e-05,
+      "loss": 0.8376,
+      "step": 14150
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.351800777072875e-05,
+      "loss": 0.6984,
+      "step": 14160
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.3514881544420259e-05,
+      "loss": 0.7484,
+      "step": 14170
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.3511752386531575e-05,
+      "loss": 1.064,
+      "step": 14180
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.3508620298587809e-05,
+      "loss": 0.5924,
+      "step": 14190
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.3505485282115501e-05,
+      "loss": 0.812,
+      "step": 14200
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.3502347338642623e-05,
+      "loss": 0.7384,
+      "step": 14210
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.3499206469698571e-05,
+      "loss": 0.8245,
+      "step": 14220
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.3496062676814165e-05,
+      "loss": 0.8636,
+      "step": 14230
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.349291596152165e-05,
+      "loss": 0.6606,
+      "step": 14240
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.3489766325354697e-05,
+      "loss": 0.8448,
+      "step": 14250
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.3486613769848403e-05,
+      "loss": 0.9852,
+      "step": 14260
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.3483458296539283e-05,
+      "loss": 1.052,
+      "step": 14270
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.3480299906965276e-05,
+      "loss": 0.6644,
+      "step": 14280
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.3477138602665747e-05,
+      "loss": 0.8872,
+      "step": 14290
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.3473974385181472e-05,
+      "loss": 1.069,
+      "step": 14300
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.3470807256054654e-05,
+      "loss": 0.7077,
+      "step": 14310
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.3467637216828916e-05,
+      "loss": 0.8647,
+      "step": 14320
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.3464464269049293e-05,
+      "loss": 0.9642,
+      "step": 14330
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.3461288414262242e-05,
+      "loss": 0.7707,
+      "step": 14340
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.3458109654015637e-05,
+      "loss": 0.7427,
+      "step": 14350
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.3454927989858766e-05,
+      "loss": 0.561,
+      "step": 14360
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.3451743423342333e-05,
+      "loss": 1.0471,
+      "step": 14370
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.344855595601846e-05,
+      "loss": 0.9729,
+      "step": 14380
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.3445365589440676e-05,
+      "loss": 0.7116,
+      "step": 14390
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.344217232516393e-05,
+      "loss": 0.8295,
+      "step": 14400
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.343897616474458e-05,
+      "loss": 0.6606,
+      "step": 14410
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.3435777109740394e-05,
+      "loss": 0.8007,
+      "step": 14420
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.3432575161710552e-05,
+      "loss": 1.0616,
+      "step": 14430
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.3429370322215648e-05,
+      "loss": 0.7153,
+      "step": 14440
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.3426162592817678e-05,
+      "loss": 0.7569,
+      "step": 14450
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.3422951975080054e-05,
+      "loss": 0.9862,
+      "step": 14460
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.3419738470567587e-05,
+      "loss": 0.5145,
+      "step": 14470
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.3416522080846506e-05,
+      "loss": 1.1163,
+      "step": 14480
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.3413302807484436e-05,
+      "loss": 0.8015,
+      "step": 14490
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.3410080652050414e-05,
+      "loss": 0.7657,
+      "step": 14500
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.340685561611488e-05,
+      "loss": 0.534,
+      "step": 14510
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.3403627701249675e-05,
+      "loss": 0.8209,
+      "step": 14520
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.3400396909028046e-05,
+      "loss": 0.5046,
+      "step": 14530
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.3397163241024641e-05,
+      "loss": 0.8202,
+      "step": 14540
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.3393926698815516e-05,
+      "loss": 0.7993,
+      "step": 14550
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.3390687283978114e-05,
+      "loss": 0.6522,
+      "step": 14560
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.3387444998091294e-05,
+      "loss": 0.7665,
+      "step": 14570
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.33841998427353e-05,
+      "loss": 0.5005,
+      "step": 14580
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.3380951819491785e-05,
+      "loss": 0.8614,
+      "step": 14590
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.3377700929943799e-05,
+      "loss": 0.4467,
+      "step": 14600
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.337444717567578e-05,
+      "loss": 1.0509,
+      "step": 14610
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.3371190558273574e-05,
+      "loss": 1.0283,
+      "step": 14620
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.336793107932441e-05,
+      "loss": 0.9415,
+      "step": 14630
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.3364668740416927e-05,
+      "loss": 0.6773,
+      "step": 14640
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.3361403543141141e-05,
+      "loss": 0.7324,
+      "step": 14650
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.3358135489088473e-05,
+      "loss": 0.9419,
+      "step": 14660
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.3354864579851737e-05,
+      "loss": 0.4525,
+      "step": 14670
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.3351590817025127e-05,
+      "loss": 0.5833,
+      "step": 14680
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.3348314202204241e-05,
+      "loss": 0.8469,
+      "step": 14690
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.3345034736986057e-05,
+      "loss": 0.8856,
+      "step": 14700
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.3341752422968948e-05,
+      "loss": 0.5665,
+      "step": 14710
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.3338467261752677e-05,
+      "loss": 0.9724,
+      "step": 14720
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.3335179254938387e-05,
+      "loss": 0.6042,
+      "step": 14730
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.3331888404128615e-05,
+      "loss": 0.9173,
+      "step": 14740
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.3328594710927282e-05,
+      "loss": 1.3782,
+      "step": 14750
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.3325298176939694e-05,
+      "loss": 0.876,
+      "step": 14760
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.332199880377254e-05,
+      "loss": 0.876,
+      "step": 14770
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.3318696593033896e-05,
+      "loss": 1.0023,
+      "step": 14780
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.3315391546333219e-05,
+      "loss": 0.6538,
+      "step": 14790
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.3312083665281348e-05,
+      "loss": 0.7284,
+      "step": 14800
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.3308772951490503e-05,
+      "loss": 0.7376,
+      "step": 14810
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.330545940657429e-05,
+      "loss": 0.5036,
+      "step": 14820
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.3302143032147687e-05,
+      "loss": 0.7377,
+      "step": 14830
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.3298823829827055e-05,
+      "loss": 1.1506,
+      "step": 14840
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.3295501801230133e-05,
+      "loss": 0.6578,
+      "step": 14850
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.3292176947976038e-05,
+      "loss": 0.7625,
+      "step": 14860
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.3288849271685263e-05,
+      "loss": 0.64,
+      "step": 14870
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.3285518773979677e-05,
+      "loss": 0.6243,
+      "step": 14880
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.3282185456482522e-05,
+      "loss": 0.9075,
+      "step": 14890
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.327884932081842e-05,
+      "loss": 0.9783,
+      "step": 14900
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.327551036861336e-05,
+      "loss": 0.6165,
+      "step": 14910
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.327216860149471e-05,
+      "loss": 0.9649,
+      "step": 14920
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.3268824021091203e-05,
+      "loss": 0.7211,
+      "step": 14930
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.326547662903295e-05,
+      "loss": 0.9264,
+      "step": 14940
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.3262126426951427e-05,
+      "loss": 0.7827,
+      "step": 14950
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.3258773416479483e-05,
+      "loss": 0.947,
+      "step": 14960
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.3255417599251331e-05,
+      "loss": 0.5499,
+      "step": 14970
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.3252058976902563e-05,
+      "loss": 0.6036,
+      "step": 14980
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.3248697551070124e-05,
+      "loss": 0.8893,
+      "step": 14990
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.3245333323392335e-05,
+      "loss": 0.6627,
+      "step": 15000
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.3241966295508879e-05,
+      "loss": 0.7549,
+      "step": 15010
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.3238596469060808e-05,
+      "loss": 0.7898,
+      "step": 15020
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.3235223845690528e-05,
+      "loss": 0.7913,
+      "step": 15030
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.3231848427041817e-05,
+      "loss": 1.0171,
+      "step": 15040
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.3228470214759818e-05,
+      "loss": 0.9284,
+      "step": 15050
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.3225089210491024e-05,
+      "loss": 1.0654,
+      "step": 15060
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.3221705415883297e-05,
+      "loss": 0.762,
+      "step": 15070
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.321831883258586e-05,
+      "loss": 0.656,
+      "step": 15080
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.321492946224929e-05,
+      "loss": 0.5709,
+      "step": 15090
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.3211537306525526e-05,
+      "loss": 0.8218,
+      "step": 15100
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.3208142367067865e-05,
+      "loss": 0.5841,
+      "step": 15110
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.3204744645530956e-05,
+      "loss": 0.6966,
+      "step": 15120
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.3201344143570806e-05,
+      "loss": 0.9424,
+      "step": 15130
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.3197940862844786e-05,
+      "loss": 0.6475,
+      "step": 15140
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.3194534805011606e-05,
+      "loss": 0.91,
+      "step": 15150
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.3191125971731342e-05,
+      "loss": 0.8327,
+      "step": 15160
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.3187714364665415e-05,
+      "loss": 0.7005,
+      "step": 15170
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.31842999854766e-05,
+      "loss": 0.9117,
+      "step": 15180
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.318088283582903e-05,
+      "loss": 0.6846,
+      "step": 15190
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.3177462917388173e-05,
+      "loss": 0.8622,
+      "step": 15200
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.3174040231820863e-05,
+      "loss": 0.9937,
+      "step": 15210
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.3170614780795273e-05,
+      "loss": 0.5033,
+      "step": 15220
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.3167186565980927e-05,
+      "loss": 1.3306,
+      "step": 15230
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.3163755589048693e-05,
+      "loss": 0.9008,
+      "step": 15240
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.316032185167079e-05,
+      "loss": 0.8008,
+      "step": 15250
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.3156885355520778e-05,
+      "loss": 0.6892,
+      "step": 15260
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.3153446102273566e-05,
+      "loss": 0.7426,
+      "step": 15270
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.3150004093605403e-05,
+      "loss": 0.7207,
+      "step": 15280
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.3146559331193879e-05,
+      "loss": 0.9089,
+      "step": 15290
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.3143111816717933e-05,
+      "loss": 0.8718,
+      "step": 15300
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.3139661551857842e-05,
+      "loss": 0.8146,
+      "step": 15310
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.3136208538295221e-05,
+      "loss": 0.7586,
+      "step": 15320
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.3132752777713027e-05,
+      "loss": 0.6503,
+      "step": 15330
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.312929427179556e-05,
+      "loss": 0.6057,
+      "step": 15340
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.3125833022228448e-05,
+      "loss": 0.7167,
+      "step": 15350
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.3122369030698663e-05,
+      "loss": 0.8433,
+      "step": 15360
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.3118902298894515e-05,
+      "loss": 0.7265,
+      "step": 15370
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.3115432828505646e-05,
+      "loss": 0.6624,
+      "step": 15380
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.3111960621223035e-05,
+      "loss": 0.5974,
+      "step": 15390
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.310848567873899e-05,
+      "loss": 0.9127,
+      "step": 15400
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.3105008002747158e-05,
+      "loss": 0.8466,
+      "step": 15410
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.3101527594942517e-05,
+      "loss": 0.5286,
+      "step": 15420
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.3098044457021373e-05,
+      "loss": 0.6996,
+      "step": 15430
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.3094558590681367e-05,
+      "loss": 0.6488,
+      "step": 15440
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.3091069997621466e-05,
+      "loss": 0.5853,
+      "step": 15450
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.3087578679541972e-05,
+      "loss": 0.7664,
+      "step": 15460
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.3084084638144506e-05,
+      "loss": 0.8834,
+      "step": 15470
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.3080587875132024e-05,
+      "loss": 1.1025,
+      "step": 15480
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.3077088392208807e-05,
+      "loss": 0.82,
+      "step": 15490
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.3073586191080456e-05,
+      "loss": 0.9271,
+      "step": 15500
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.3070081273453906e-05,
+      "loss": 0.8683,
+      "step": 15510
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.3066573641037412e-05,
+      "loss": 0.7035,
+      "step": 15520
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.3063063295540545e-05,
+      "loss": 0.6797,
+      "step": 15530
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.3059550238674209e-05,
+      "loss": 0.8762,
+      "step": 15540
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.3056034472150625e-05,
+      "loss": 0.6074,
+      "step": 15550
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.3052515997683336e-05,
+      "loss": 0.6325,
+      "step": 15560
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.3048994816987201e-05,
+      "loss": 0.7776,
+      "step": 15570
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.3045470931778403e-05,
+      "loss": 0.8261,
+      "step": 15580
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.304194434377444e-05,
+      "loss": 0.6536,
+      "step": 15590
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.303841505469413e-05,
+      "loss": 1.1411,
+      "step": 15600
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.3034883066257602e-05,
+      "loss": 0.6888,
+      "step": 15610
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.3031348380186305e-05,
+      "loss": 0.7335,
+      "step": 15620
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.3027810998203005e-05,
+      "loss": 0.8149,
+      "step": 15630
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.3024270922031775e-05,
+      "loss": 1.115,
+      "step": 15640
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.302072815339801e-05,
+      "loss": 0.6974,
+      "step": 15650
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.3017182694028406e-05,
+      "loss": 1.0305,
+      "step": 15660
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.3013634545650983e-05,
+      "loss": 0.5599,
+      "step": 15670
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.3010083709995062e-05,
+      "loss": 0.7644,
+      "step": 15680
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.3006530188791278e-05,
+      "loss": 0.688,
+      "step": 15690
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.3002973983771572e-05,
+      "loss": 0.8266,
+      "step": 15700
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.2999415096669201e-05,
+      "loss": 0.9208,
+      "step": 15710
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.2995853529218718e-05,
+      "loss": 0.6909,
+      "step": 15720
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.2992289283155988e-05,
+      "loss": 0.9514,
+      "step": 15730
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.2988722360218184e-05,
+      "loss": 0.5879,
+      "step": 15740
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 1.2985152762143779e-05,
+      "loss": 0.7636,
+      "step": 15750
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 1.2981580490672554e-05,
+      "loss": 0.6555,
+      "step": 15760
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 1.2978005547545586e-05,
+      "loss": 0.4848,
+      "step": 15770
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 1.2974427934505262e-05,
+      "loss": 0.6107,
+      "step": 15780
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 1.2970847653295267e-05,
+      "loss": 1.2346,
+      "step": 15790
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 1.2967264705660587e-05,
+      "loss": 0.6772,
+      "step": 15800
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 1.2963679093347507e-05,
+      "loss": 0.7676,
+      "step": 15810
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 1.296009081810361e-05,
+      "loss": 0.8187,
+      "step": 15820
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 1.2956499881677777e-05,
+      "loss": 0.5784,
+      "step": 15830
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 1.295290628582019e-05,
+      "loss": 0.5904,
+      "step": 15840
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 1.294931003228232e-05,
+      "loss": 0.7243,
+      "step": 15850
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 1.2945711122816939e-05,
+      "loss": 0.8472,
+      "step": 15860
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 1.2942109559178113e-05,
+      "loss": 0.9001,
+      "step": 15870
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 1.2938505343121199e-05,
+      "loss": 0.6855,
+      "step": 15880
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 1.2934898476402851e-05,
+      "loss": 0.7127,
+      "step": 15890
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 1.2931288960781008e-05,
+      "loss": 0.4177,
+      "step": 15900
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 1.2927676798014904e-05,
+      "loss": 0.8813,
+      "step": 15910
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 1.2924061989865066e-05,
+      "loss": 0.6631,
+      "step": 15920
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 1.2920444538093305e-05,
+      "loss": 1.1157,
+      "step": 15930
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 1.2916824444462726e-05,
+      "loss": 0.9802,
+      "step": 15940
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 1.2913201710737716e-05,
+      "loss": 0.5958,
+      "step": 15950
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 1.2909576338683956e-05,
+      "loss": 1.1829,
+      "step": 15960
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 1.2905948330068401e-05,
+      "loss": 0.7153,
+      "step": 15970
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 1.2902317686659302e-05,
+      "loss": 1.0184,
+      "step": 15980
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 1.2898684410226192e-05,
+      "loss": 0.6815,
+      "step": 15990
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 1.2895048502539883e-05,
+      "loss": 0.7102,
+      "step": 16000
+    },
+    {
+      "epoch": 7.04,
+      "eval_accuracy": 0.8242105263157895,
+      "eval_f1": 0.8242105263157895,
+      "eval_loss": 0.7876191735267639,
+      "eval_runtime": 761.4492,
+      "eval_samples_per_second": 6.238,
+      "eval_steps_per_second": 1.56,
+      "step": 16000
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 1.2891409965372476e-05,
+      "loss": 0.903,
+      "step": 16010
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 1.2887768800497346e-05,
+      "loss": 0.7573,
+      "step": 16020
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 1.2884125009689153e-05,
+      "loss": 0.9222,
+      "step": 16030
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 1.2880478594723841e-05,
+      "loss": 0.9862,
+      "step": 16040
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 1.287682955737862e-05,
+      "loss": 0.7781,
+      "step": 16050
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 1.2873177899431994e-05,
+      "loss": 0.7142,
+      "step": 16060
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 1.2869523622663734e-05,
+      "loss": 0.9108,
+      "step": 16070
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 1.2865866728854889e-05,
+      "loss": 0.572,
+      "step": 16080
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 1.2862207219787787e-05,
+      "loss": 0.5146,
+      "step": 16090
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 1.2858545097246025e-05,
+      "loss": 0.7091,
+      "step": 16100
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 1.2854880363014482e-05,
+      "loss": 0.759,
+      "step": 16110
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 1.2851213018879298e-05,
+      "loss": 0.5985,
+      "step": 16120
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 1.2847543066627897e-05,
+      "loss": 0.8886,
+      "step": 16130
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 1.284387050804897e-05,
+      "loss": 1.0709,
+      "step": 16140
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 1.2840195344932474e-05,
+      "loss": 0.735,
+      "step": 16150
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 1.2836517579069642e-05,
+      "loss": 0.8514,
+      "step": 16160
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 1.2832837212252966e-05,
+      "loss": 0.6403,
+      "step": 16170
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 1.2829154246276222e-05,
+      "loss": 0.6856,
+      "step": 16180
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 1.2825468682934436e-05,
+      "loss": 0.7865,
+      "step": 16190
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 1.2821780524023907e-05,
+      "loss": 0.6365,
+      "step": 16200
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 1.28180897713422e-05,
+      "loss": 0.6259,
+      "step": 16210
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 1.2814396426688147e-05,
+      "loss": 0.7151,
+      "step": 16220
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 1.2810700491861833e-05,
+      "loss": 0.7079,
+      "step": 16230
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 1.2807001968664616e-05,
+      "loss": 0.7594,
+      "step": 16240
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.2803300858899106e-05,
+      "loss": 0.7652,
+      "step": 16250
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.2799597164369187e-05,
+      "loss": 0.8973,
+      "step": 16260
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.2795890886879988e-05,
+      "loss": 0.9693,
+      "step": 16270
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.2792182028237907e-05,
+      "loss": 1.0535,
+      "step": 16280
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.2788470590250594e-05,
+      "loss": 0.7048,
+      "step": 16290
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.278475657472696e-05,
+      "loss": 0.6197,
+      "step": 16300
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.2781039983477172e-05,
+      "loss": 0.6348,
+      "step": 16310
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.2777320818312652e-05,
+      "loss": 0.8078,
+      "step": 16320
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.2773599081046076e-05,
+      "loss": 1.0742,
+      "step": 16330
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.276987477349137e-05,
+      "loss": 0.6463,
+      "step": 16340
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.2766147897463718e-05,
+      "loss": 0.4302,
+      "step": 16350
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.2762418454779556e-05,
+      "loss": 0.751,
+      "step": 16360
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.2758686447256568e-05,
+      "loss": 0.9189,
+      "step": 16370
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.2754951876713688e-05,
+      "loss": 0.8189,
+      "step": 16380
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.27512147449711e-05,
+      "loss": 0.5518,
+      "step": 16390
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.2747475053850241e-05,
+      "loss": 1.0893,
+      "step": 16400
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.2743732805173786e-05,
+      "loss": 0.9392,
+      "step": 16410
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.2739988000765664e-05,
+      "loss": 1.016,
+      "step": 16420
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.2736240642451046e-05,
+      "loss": 0.3563,
+      "step": 16430
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.2732490732056348e-05,
+      "loss": 0.9955,
+      "step": 16440
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.2728738271409236e-05,
+      "loss": 0.5461,
+      "step": 16450
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.2724983262338605e-05,
+      "loss": 0.5001,
+      "step": 16460
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.2721225706674609e-05,
+      "loss": 0.6405,
+      "step": 16470
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.2717465606248632e-05,
+      "loss": 0.7436,
+      "step": 16480
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.2713702962893299e-05,
+      "loss": 0.8613,
+      "step": 16490
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.270993777844248e-05,
+      "loss": 0.929,
+      "step": 16500
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.270617005473128e-05,
+      "loss": 0.504,
+      "step": 16510
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.2702399793596037e-05,
+      "loss": 0.4618,
+      "step": 16520
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.2698626996874338e-05,
+      "loss": 0.6505,
+      "step": 16530
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.2694851666404991e-05,
+      "loss": 0.8792,
+      "step": 16540
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.2691073804028052e-05,
+      "loss": 0.7674,
+      "step": 16550
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.26872934115848e-05,
+      "loss": 1.042,
+      "step": 16560
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.2683510490917759e-05,
+      "loss": 0.9164,
+      "step": 16570
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.2679725043870672e-05,
+      "loss": 0.6523,
+      "step": 16580
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.2675937072288526e-05,
+      "loss": 0.5904,
+      "step": 16590
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.2672146578017524e-05,
+      "loss": 0.8063,
+      "step": 16600
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.2668353562905114e-05,
+      "loss": 0.4042,
+      "step": 16610
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.2664558028799963e-05,
+      "loss": 0.6684,
+      "step": 16620
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.2660759977551972e-05,
+      "loss": 0.6434,
+      "step": 16630
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.2656959411012255e-05,
+      "loss": 0.9733,
+      "step": 16640
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.2653156331033174e-05,
+      "loss": 0.5205,
+      "step": 16650
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.2649350739468295e-05,
+      "loss": 0.4417,
+      "step": 16660
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.2645542638172423e-05,
+      "loss": 0.8051,
+      "step": 16670
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.2641732029001577e-05,
+      "loss": 0.7189,
+      "step": 16680
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.2637918913813002e-05,
+      "loss": 0.7512,
+      "step": 16690
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.2634103294465166e-05,
+      "loss": 1.0825,
+      "step": 16700
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.2630285172817751e-05,
+      "loss": 1.1748,
+      "step": 16710
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.2626464550731671e-05,
+      "loss": 1.0208,
+      "step": 16720
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.2622641430069041e-05,
+      "loss": 0.5634,
+      "step": 16730
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.2618815812693213e-05,
+      "loss": 0.4922,
+      "step": 16740
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.2614987700468739e-05,
+      "loss": 0.8235,
+      "step": 16750
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.2611157095261399e-05,
+      "loss": 0.6315,
+      "step": 16760
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.2607323998938181e-05,
+      "loss": 0.8166,
+      "step": 16770
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.260348841336729e-05,
+      "loss": 0.7631,
+      "step": 16780
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.2599650340418144e-05,
+      "loss": 0.7756,
+      "step": 16790
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.2595809781961374e-05,
+      "loss": 0.7076,
+      "step": 16800
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.2591966739868822e-05,
+      "loss": 0.7914,
+      "step": 16810
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.2588121216013537e-05,
+      "loss": 0.6832,
+      "step": 16820
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.2584273212269782e-05,
+      "loss": 0.547,
+      "step": 16830
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.258042273051303e-05,
+      "loss": 0.9669,
+      "step": 16840
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.2576569772619955e-05,
+      "loss": 0.6243,
+      "step": 16850
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.2572714340468445e-05,
+      "loss": 0.6971,
+      "step": 16860
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.256885643593759e-05,
+      "loss": 0.5598,
+      "step": 16870
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.2564996060907683e-05,
+      "loss": 0.604,
+      "step": 16880
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.2561133217260227e-05,
+      "loss": 0.7139,
+      "step": 16890
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.2557267906877925e-05,
+      "loss": 0.4249,
+      "step": 16900
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.255340013164468e-05,
+      "loss": 0.4992,
+      "step": 16910
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.2549529893445603e-05,
+      "loss": 1.0948,
+      "step": 16920
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.2545657194166997e-05,
+      "loss": 0.6932,
+      "step": 16930
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.2541782035696372e-05,
+      "loss": 0.5878,
+      "step": 16940
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.2537904419922428e-05,
+      "loss": 0.6641,
+      "step": 16950
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.253402434873507e-05,
+      "loss": 0.8987,
+      "step": 16960
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.2530141824025399e-05,
+      "loss": 0.5498,
+      "step": 16970
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.2526256847685713e-05,
+      "loss": 0.7692,
+      "step": 16980
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.2522369421609497e-05,
+      "loss": 0.7997,
+      "step": 16990
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.2518479547691437e-05,
+      "loss": 0.5458,
+      "step": 17000
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.2514587227827412e-05,
+      "loss": 0.8942,
+      "step": 17010
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.2510692463914487e-05,
+      "loss": 0.461,
+      "step": 17020
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.250679525785093e-05,
+      "loss": 0.9453,
+      "step": 17030
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.2502895611536185e-05,
+      "loss": 1.1806,
+      "step": 17040
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.2498993526870893e-05,
+      "loss": 0.8235,
+      "step": 17050
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.2495089005756888e-05,
+      "loss": 0.3497,
+      "step": 17060
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.249118205009718e-05,
+      "loss": 0.9592,
+      "step": 17070
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.2487272661795976e-05,
+      "loss": 0.5981,
+      "step": 17080
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.2483360842758662e-05,
+      "loss": 0.7031,
+      "step": 17090
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.2479446594891812e-05,
+      "loss": 0.8887,
+      "step": 17100
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.2475529920103183e-05,
+      "loss": 0.5698,
+      "step": 17110
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.2471610820301714e-05,
+      "loss": 0.9545,
+      "step": 17120
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.2467689297397526e-05,
+      "loss": 0.5898,
+      "step": 17130
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.2463765353301925e-05,
+      "loss": 0.4993,
+      "step": 17140
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.245983898992739e-05,
+      "loss": 0.4658,
+      "step": 17150
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.2455910209187584e-05,
+      "loss": 0.7944,
+      "step": 17160
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.2451979012997347e-05,
+      "loss": 0.8215,
+      "step": 17170
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.2448045403272696e-05,
+      "loss": 0.5273,
+      "step": 17180
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.2444109381930825e-05,
+      "loss": 0.6735,
+      "step": 17190
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.2440170950890103e-05,
+      "loss": 0.8846,
+      "step": 17200
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.2436230112070075e-05,
+      "loss": 0.5238,
+      "step": 17210
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.2432286867391455e-05,
+      "loss": 0.9863,
+      "step": 17220
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.2428341218776135e-05,
+      "loss": 0.8749,
+      "step": 17230
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.2424393168147173e-05,
+      "loss": 0.5667,
+      "step": 17240
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.2420442717428804e-05,
+      "loss": 0.877,
+      "step": 17250
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.241648986854643e-05,
+      "loss": 0.7394,
+      "step": 17260
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.241253462342662e-05,
+      "loss": 0.6451,
+      "step": 17270
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.2408576983997112e-05,
+      "loss": 0.8236,
+      "step": 17280
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.2404616952186811e-05,
+      "loss": 0.8152,
+      "step": 17290
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.2400654529925793e-05,
+      "loss": 0.8507,
+      "step": 17300
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.2396689719145288e-05,
+      "loss": 0.7373,
+      "step": 17310
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.2392722521777699e-05,
+      "loss": 0.4609,
+      "step": 17320
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.2388752939756591e-05,
+      "loss": 0.7621,
+      "step": 17330
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.238478097501669e-05,
+      "loss": 0.7369,
+      "step": 17340
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.2380806629493882e-05,
+      "loss": 0.6674,
+      "step": 17350
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.2376829905125215e-05,
+      "loss": 0.7949,
+      "step": 17360
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.23728508038489e-05,
+      "loss": 0.9449,
+      "step": 17370
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.2368869327604297e-05,
+      "loss": 0.7967,
+      "step": 17380
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.2364885478331933e-05,
+      "loss": 0.9675,
+      "step": 17390
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.2360899257973482e-05,
+      "loss": 1.2767,
+      "step": 17400
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.2356910668471788e-05,
+      "loss": 0.816,
+      "step": 17410
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.2352919711770834e-05,
+      "loss": 0.8011,
+      "step": 17420
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.2348926389815766e-05,
+      "loss": 0.5407,
+      "step": 17430
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.2344930704552883e-05,
+      "loss": 0.5253,
+      "step": 17440
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.234093265792963e-05,
+      "loss": 0.6741,
+      "step": 17450
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.2336932251894608e-05,
+      "loss": 0.8949,
+      "step": 17460
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.2332929488397567e-05,
+      "loss": 1.0454,
+      "step": 17470
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.2328924369389404e-05,
+      "loss": 0.4574,
+      "step": 17480
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.2324916896822166e-05,
+      "loss": 0.9879,
+      "step": 17490
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.2320907072649045e-05,
+      "loss": 0.9544,
+      "step": 17500
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.2316894898824383e-05,
+      "loss": 1.1716,
+      "step": 17510
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.2312880377303664e-05,
+      "loss": 0.9311,
+      "step": 17520
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.2308863510043517e-05,
+      "loss": 0.5337,
+      "step": 17530
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.2304844299001712e-05,
+      "loss": 0.8078,
+      "step": 17540
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.2300822746137167e-05,
+      "loss": 0.8005,
+      "step": 17550
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.2296798853409937e-05,
+      "loss": 0.8335,
+      "step": 17560
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.2292772622781215e-05,
+      "loss": 0.6197,
+      "step": 17570
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.2288744056213343e-05,
+      "loss": 0.7248,
+      "step": 17580
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.228471315566979e-05,
+      "loss": 0.5651,
+      "step": 17590
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.2280679923115173e-05,
+      "loss": 0.7526,
+      "step": 17600
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.2276644360515238e-05,
+      "loss": 0.3586,
+      "step": 17610
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.2272606469836868e-05,
+      "loss": 0.8768,
+      "step": 17620
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.2268566253048081e-05,
+      "loss": 0.8052,
+      "step": 17630
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.2264523712118033e-05,
+      "loss": 0.748,
+      "step": 17640
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.2260478849017007e-05,
+      "loss": 0.7358,
+      "step": 17650
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.225643166571642e-05,
+      "loss": 0.3771,
+      "step": 17660
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.2252382164188825e-05,
+      "loss": 0.6613,
+      "step": 17670
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.2248330346407893e-05,
+      "loss": 1.1527,
+      "step": 17680
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.2244276214348435e-05,
+      "loss": 0.5922,
+      "step": 17690
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.2240219769986382e-05,
+      "loss": 0.6686,
+      "step": 17700
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.2236161015298799e-05,
+      "loss": 0.6159,
+      "step": 17710
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.2232099952263872e-05,
+      "loss": 0.9876,
+      "step": 17720
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.2228036582860917e-05,
+      "loss": 0.6581,
+      "step": 17730
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.2223970909070367e-05,
+      "loss": 0.613,
+      "step": 17740
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 1.2219902932873782e-05,
+      "loss": 0.5538,
+      "step": 17750
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 1.2215832656253846e-05,
+      "loss": 0.7896,
+      "step": 17760
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 1.2211760081194362e-05,
+      "loss": 0.9046,
+      "step": 17770
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 1.2207685209680254e-05,
+      "loss": 0.8718,
+      "step": 17780
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 1.2203608043697564e-05,
+      "loss": 0.6827,
+      "step": 17790
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 1.2199528585233455e-05,
+      "loss": 0.5505,
+      "step": 17800
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 1.2195446836276202e-05,
+      "loss": 0.5683,
+      "step": 17810
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 1.2191362798815203e-05,
+      "loss": 0.946,
+      "step": 17820
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 1.2187276474840968e-05,
+      "loss": 0.581,
+      "step": 17830
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 1.2183187866345117e-05,
+      "loss": 0.9227,
+      "step": 17840
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 1.2179096975320398e-05,
+      "loss": 0.5363,
+      "step": 17850
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 1.217500380376065e-05,
+      "loss": 0.6947,
+      "step": 17860
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 1.2170908353660844e-05,
+      "loss": 0.8875,
+      "step": 17870
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 1.216681062701705e-05,
+      "loss": 0.4628,
+      "step": 17880
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 1.2162710625826446e-05,
+      "loss": 0.8057,
+      "step": 17890
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 1.2158608352087328e-05,
+      "loss": 0.9372,
+      "step": 17900
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 1.2154503807799093e-05,
+      "loss": 0.649,
+      "step": 17910
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 1.2150396994962244e-05,
+      "loss": 0.835,
+      "step": 17920
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 1.2146287915578394e-05,
+      "loss": 0.4606,
+      "step": 17930
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 1.2142176571650259e-05,
+      "loss": 0.7991,
+      "step": 17940
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 1.2138062965181656e-05,
+      "loss": 0.7018,
+      "step": 17950
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 1.2133947098177508e-05,
+      "loss": 1.0809,
+      "step": 17960
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 1.2129828972643837e-05,
+      "loss": 0.4973,
+      "step": 17970
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 1.212570859058777e-05,
+      "loss": 0.6203,
+      "step": 17980
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 1.2121585954017528e-05,
+      "loss": 0.5908,
+      "step": 17990
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 1.2117461064942437e-05,
+      "loss": 0.5726,
+      "step": 18000
+    },
+    {
+      "epoch": 8.04,
+      "eval_accuracy": 0.8231578947368421,
+      "eval_f1": 0.8231578947368422,
+      "eval_loss": 0.8805130124092102,
+      "eval_runtime": 767.3617,
+      "eval_samples_per_second": 6.19,
+      "eval_steps_per_second": 1.548,
+      "step": 18000
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 1.2113333925372919e-05,
+      "loss": 1.2099,
+      "step": 18010
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 1.2109204537320483e-05,
+      "loss": 0.937,
+      "step": 18020
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 1.2105072902797753e-05,
+      "loss": 0.7613,
+      "step": 18030
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 1.2100939023818432e-05,
+      "loss": 0.8872,
+      "step": 18040
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 1.2096802902397324e-05,
+      "loss": 0.7832,
+      "step": 18050
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 1.2092664540550323e-05,
+      "loss": 0.8112,
+      "step": 18060
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 1.2088523940294418e-05,
+      "loss": 0.6844,
+      "step": 18070
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 1.2084381103647688e-05,
+      "loss": 0.8863,
+      "step": 18080
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 1.2080236032629298e-05,
+      "loss": 0.6259,
+      "step": 18090
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 1.207608872925951e-05,
+      "loss": 0.5141,
+      "step": 18100
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 1.207193919555966e-05,
+      "loss": 0.7785,
+      "step": 18110
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 1.2067787433552192e-05,
+      "loss": 0.8503,
+      "step": 18120
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 1.2063633445260615e-05,
+      "loss": 0.7436,
+      "step": 18130
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 1.2059477232709542e-05,
+      "loss": 0.8031,
+      "step": 18140
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 1.205531879792465e-05,
+      "loss": 1.2157,
+      "step": 18150
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 1.2051158142932718e-05,
+      "loss": 0.8124,
+      "step": 18160
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 1.204699526976159e-05,
+      "loss": 0.7364,
+      "step": 18170
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 1.2042830180440211e-05,
+      "loss": 0.7025,
+      "step": 18180
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 1.2038662876998586e-05,
+      "loss": 0.398,
+      "step": 18190
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 1.2034493361467813e-05,
+      "loss": 0.9601,
+      "step": 18200
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 1.203032163588006e-05,
+      "loss": 0.4517,
+      "step": 18210
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 1.2026147702268574e-05,
+      "loss": 0.6078,
+      "step": 18220
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 1.2021971562667687e-05,
+      "loss": 0.6127,
+      "step": 18230
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 1.201779321911279e-05,
+      "loss": 0.9812,
+      "step": 18240
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.2013612673640364e-05,
+      "loss": 0.7009,
+      "step": 18250
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.200942992828795e-05,
+      "loss": 0.5552,
+      "step": 18260
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.2005244985094171e-05,
+      "loss": 0.7145,
+      "step": 18270
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.2001057846098717e-05,
+      "loss": 0.4111,
+      "step": 18280
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.1996868513342349e-05,
+      "loss": 0.5976,
+      "step": 18290
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.1992676988866894e-05,
+      "loss": 0.7484,
+      "step": 18300
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.1988483274715256e-05,
+      "loss": 0.9863,
+      "step": 18310
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.1984287372931392e-05,
+      "loss": 0.3628,
+      "step": 18320
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.1980089285560342e-05,
+      "loss": 0.4903,
+      "step": 18330
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.1975889014648195e-05,
+      "loss": 0.8426,
+      "step": 18340
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.197168656224212e-05,
+      "loss": 0.7967,
+      "step": 18350
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.1967481930390335e-05,
+      "loss": 0.4083,
+      "step": 18360
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.196327512114213e-05,
+      "loss": 0.6116,
+      "step": 18370
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.1959066136547851e-05,
+      "loss": 0.9358,
+      "step": 18380
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.1954854978658903e-05,
+      "loss": 0.6193,
+      "step": 18390
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.1950641649527762e-05,
+      "loss": 0.677,
+      "step": 18400
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.1946426151207945e-05,
+      "loss": 0.6106,
+      "step": 18410
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.1942208485754038e-05,
+      "loss": 0.6699,
+      "step": 18420
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.193798865522168e-05,
+      "loss": 0.5351,
+      "step": 18430
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.1933766661667565e-05,
+      "loss": 0.836,
+      "step": 18440
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.192954250714944e-05,
+      "loss": 0.7118,
+      "step": 18450
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.192531619372611e-05,
+      "loss": 1.0574,
+      "step": 18460
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.1921087723457425e-05,
+      "loss": 0.7786,
+      "step": 18470
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.191685709840429e-05,
+      "loss": 0.9358,
+      "step": 18480
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.1912624320628666e-05,
+      "loss": 0.8795,
+      "step": 18490
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.1908389392193549e-05,
+      "loss": 0.7275,
+      "step": 18500
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.1904152315162996e-05,
+      "loss": 0.487,
+      "step": 18510
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.189991309160211e-05,
+      "loss": 0.6892,
+      "step": 18520
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.1895671723577032e-05,
+      "loss": 0.8501,
+      "step": 18530
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.1891428213154956e-05,
+      "loss": 1.0048,
+      "step": 18540
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.1887182562404118e-05,
+      "loss": 0.579,
+      "step": 18550
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.188293477339379e-05,
+      "loss": 1.0681,
+      "step": 18560
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.1878684848194302e-05,
+      "loss": 0.9328,
+      "step": 18570
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.187443278887701e-05,
+      "loss": 0.8938,
+      "step": 18580
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.187017859751432e-05,
+      "loss": 0.7139,
+      "step": 18590
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.1865922276179671e-05,
+      "loss": 0.7831,
+      "step": 18600
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.186166382694754e-05,
+      "loss": 0.6213,
+      "step": 18610
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.185740325189345e-05,
+      "loss": 0.901,
+      "step": 18620
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.1853140553093945e-05,
+      "loss": 0.5026,
+      "step": 18630
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.1848875732626619e-05,
+      "loss": 0.5141,
+      "step": 18640
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.1844608792570091e-05,
+      "loss": 0.7118,
+      "step": 18650
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.1840339735004018e-05,
+      "loss": 1.1035,
+      "step": 18660
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.1836068562009084e-05,
+      "loss": 0.4183,
+      "step": 18670
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.1831795275667007e-05,
+      "loss": 0.6543,
+      "step": 18680
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.1827519878060537e-05,
+      "loss": 0.5805,
+      "step": 18690
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.182324237127345e-05,
+      "loss": 0.3918,
+      "step": 18700
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.1818962757390552e-05,
+      "loss": 1.0159,
+      "step": 18710
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.1814681038497671e-05,
+      "loss": 0.5038,
+      "step": 18720
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.1810397216681665e-05,
+      "loss": 0.5818,
+      "step": 18730
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.1806111294030424e-05,
+      "loss": 0.8919,
+      "step": 18740
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.1801823272632845e-05,
+      "loss": 0.8671,
+      "step": 18750
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.1797533154578866e-05,
+      "loss": 0.858,
+      "step": 18760
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.1793240941959434e-05,
+      "loss": 0.5649,
+      "step": 18770
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.1788946636866518e-05,
+      "loss": 0.7596,
+      "step": 18780
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.1784650241393117e-05,
+      "loss": 0.8354,
+      "step": 18790
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.178035175763324e-05,
+      "loss": 0.5007,
+      "step": 18800
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.1776051187681911e-05,
+      "loss": 0.9104,
+      "step": 18810
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.177174853363518e-05,
+      "loss": 0.639,
+      "step": 18820
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.176744379759011e-05,
+      "loss": 0.6799,
+      "step": 18830
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.1763136981644773e-05,
+      "loss": 0.9681,
+      "step": 18840
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.175882808789826e-05,
+      "loss": 1.0896,
+      "step": 18850
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.1754517118450675e-05,
+      "loss": 0.5733,
+      "step": 18860
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.1750204075403128e-05,
+      "loss": 0.6242,
+      "step": 18870
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.1745888960857749e-05,
+      "loss": 0.5115,
+      "step": 18880
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.1741571776917673e-05,
+      "loss": 0.5443,
+      "step": 18890
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.1737252525687035e-05,
+      "loss": 0.6259,
+      "step": 18900
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.1732931209270995e-05,
+      "loss": 0.7084,
+      "step": 18910
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.17286078297757e-05,
+      "loss": 0.8891,
+      "step": 18920
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.1724282389308324e-05,
+      "loss": 0.8908,
+      "step": 18930
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.1719954889977027e-05,
+      "loss": 0.8218,
+      "step": 18940
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.1715625333890979e-05,
+      "loss": 0.7136,
+      "step": 18950
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.1711293723160359e-05,
+      "loss": 0.587,
+      "step": 18960
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.1706960059896336e-05,
+      "loss": 0.6956,
+      "step": 18970
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.1702624346211084e-05,
+      "loss": 0.5694,
+      "step": 18980
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.1698286584217785e-05,
+      "loss": 0.5655,
+      "step": 18990
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.1693946776030601e-05,
+      "loss": 0.5769,
+      "step": 19000
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.168960492376471e-05,
+      "loss": 0.7568,
+      "step": 19010
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.1685261029536276e-05,
+      "loss": 0.6725,
+      "step": 19020
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.1680915095462456e-05,
+      "loss": 0.6308,
+      "step": 19030
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.167656712366141e-05,
+      "loss": 0.5472,
+      "step": 19040
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.1672217116252287e-05,
+      "loss": 0.6917,
+      "step": 19050
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.1667865075355224e-05,
+      "loss": 0.8717,
+      "step": 19060
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.1663511003091356e-05,
+      "loss": 0.8956,
+      "step": 19070
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.1659154901582805e-05,
+      "loss": 0.3889,
+      "step": 19080
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.165479677295268e-05,
+      "loss": 0.494,
+      "step": 19090
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.1650436619325081e-05,
+      "loss": 0.3606,
+      "step": 19100
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.1646074442825094e-05,
+      "loss": 0.6802,
+      "step": 19110
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.164171024557879e-05,
+      "loss": 0.6772,
+      "step": 19120
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.1637344029713228e-05,
+      "loss": 0.9367,
+      "step": 19130
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.1632975797356445e-05,
+      "loss": 0.5726,
+      "step": 19140
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.1628605550637467e-05,
+      "loss": 0.5994,
+      "step": 19150
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.16242332916863e-05,
+      "loss": 0.6001,
+      "step": 19160
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.1619859022633925e-05,
+      "loss": 0.7313,
+      "step": 19170
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.1615482745612315e-05,
+      "loss": 0.5976,
+      "step": 19180
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.1611104462754406e-05,
+      "loss": 0.6038,
+      "step": 19190
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.1606724176194128e-05,
+      "loss": 0.7814,
+      "step": 19200
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.1602341888066372e-05,
+      "loss": 0.6553,
+      "step": 19210
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.1597957600507019e-05,
+      "loss": 0.7844,
+      "step": 19220
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.1593571315652912e-05,
+      "loss": 0.4785,
+      "step": 19230
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.1589183035641877e-05,
+      "loss": 0.7918,
+      "step": 19240
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.1584792762612705e-05,
+      "loss": 1.0776,
+      "step": 19250
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.1580400498705161e-05,
+      "loss": 1.1064,
+      "step": 19260
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.1576006246059987e-05,
+      "loss": 0.9584,
+      "step": 19270
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.1571610006818883e-05,
+      "loss": 0.983,
+      "step": 19280
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.1567211783124523e-05,
+      "loss": 0.5213,
+      "step": 19290
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.156281157712055e-05,
+      "loss": 0.8329,
+      "step": 19300
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.155840939095157e-05,
+      "loss": 0.9438,
+      "step": 19310
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.1554005226763153e-05,
+      "loss": 0.5701,
+      "step": 19320
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.1549599086701841e-05,
+      "loss": 0.698,
+      "step": 19330
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.1545190972915127e-05,
+      "loss": 0.5665,
+      "step": 19340
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.1540780887551473e-05,
+      "loss": 0.7165,
+      "step": 19350
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.1536368832760304e-05,
+      "loss": 0.694,
+      "step": 19360
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.1531954810692e-05,
+      "loss": 0.6035,
+      "step": 19370
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.1527538823497903e-05,
+      "loss": 0.7644,
+      "step": 19380
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.1523120873330308e-05,
+      "loss": 0.8886,
+      "step": 19390
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.1518700962342475e-05,
+      "loss": 0.7071,
+      "step": 19400
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.1514279092688612e-05,
+      "loss": 0.8012,
+      "step": 19410
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.1509855266523884e-05,
+      "loss": 0.9929,
+      "step": 19420
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.1505429486004414e-05,
+      "loss": 0.7935,
+      "step": 19430
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.150100175328727e-05,
+      "loss": 0.6962,
+      "step": 19440
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.1496572070530475e-05,
+      "loss": 0.8044,
+      "step": 19450
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.1492140439893006e-05,
+      "loss": 0.7601,
+      "step": 19460
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.148770686353478e-05,
+      "loss": 0.9783,
+      "step": 19470
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.1483271343616675e-05,
+      "loss": 0.5717,
+      "step": 19480
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.1478833882300505e-05,
+      "loss": 0.4403,
+      "step": 19490
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.1474394481749037e-05,
+      "loss": 0.6769,
+      "step": 19500
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.1469953144125981e-05,
+      "loss": 0.628,
+      "step": 19510
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.1465509871595986e-05,
+      "loss": 0.2903,
+      "step": 19520
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.1461064666324659e-05,
+      "loss": 1.0457,
+      "step": 19530
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.1456617530478528e-05,
+      "loss": 0.9678,
+      "step": 19540
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.1452168466225084e-05,
+      "loss": 0.7661,
+      "step": 19550
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.1447717475732735e-05,
+      "loss": 0.7126,
+      "step": 19560
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.144326456117085e-05,
+      "loss": 0.6659,
+      "step": 19570
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.1438809724709719e-05,
+      "loss": 0.5846,
+      "step": 19580
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.1434352968520574e-05,
+      "loss": 1.1135,
+      "step": 19590
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.1429894294775594e-05,
+      "loss": 0.4112,
+      "step": 19600
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.1425433705647872e-05,
+      "loss": 0.6958,
+      "step": 19610
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.142097120331145e-05,
+      "loss": 0.5993,
+      "step": 19620
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.1416506789941295e-05,
+      "loss": 0.9286,
+      "step": 19630
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.1412040467713309e-05,
+      "loss": 1.0423,
+      "step": 19640
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.1407572238804325e-05,
+      "loss": 0.6883,
+      "step": 19650
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.1403102105392098e-05,
+      "loss": 0.5804,
+      "step": 19660
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.139863006965532e-05,
+      "loss": 0.7858,
+      "step": 19670
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.139415613377361e-05,
+      "loss": 0.7412,
+      "step": 19680
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.1389680299927506e-05,
+      "loss": 0.8717,
+      "step": 19690
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.1385202570298477e-05,
+      "loss": 0.6939,
+      "step": 19700
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.1380722947068912e-05,
+      "loss": 0.7415,
+      "step": 19710
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.1376241432422127e-05,
+      "loss": 0.9208,
+      "step": 19720
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.1371758028542356e-05,
+      "loss": 0.9007,
+      "step": 19730
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.1367272737614758e-05,
+      "loss": 0.6171,
+      "step": 19740
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.1362785561825407e-05,
+      "loss": 0.7419,
+      "step": 19750
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.13582965033613e-05,
+      "loss": 0.8097,
+      "step": 19760
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.1353805564410347e-05,
+      "loss": 0.6628,
+      "step": 19770
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.1349312747161377e-05,
+      "loss": 0.5436,
+      "step": 19780
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.1344818053804139e-05,
+      "loss": 0.8017,
+      "step": 19790
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.1340321486529287e-05,
+      "loss": 0.6905,
+      "step": 19800
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.1335823047528395e-05,
+      "loss": 0.4657,
+      "step": 19810
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.1331322738993949e-05,
+      "loss": 0.815,
+      "step": 19820
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.1326820563119344e-05,
+      "loss": 0.6882,
+      "step": 19830
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.1322316522098883e-05,
+      "loss": 0.748,
+      "step": 19840
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.1317810618127785e-05,
+      "loss": 0.572,
+      "step": 19850
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.1313302853402172e-05,
+      "loss": 0.8386,
+      "step": 19860
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.130879323011907e-05,
+      "loss": 0.6839,
+      "step": 19870
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.1304281750476418e-05,
+      "loss": 1.0472,
+      "step": 19880
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.1299768416673056e-05,
+      "loss": 0.6152,
+      "step": 19890
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.1295253230908728e-05,
+      "loss": 0.6518,
+      "step": 19900
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.1290736195384084e-05,
+      "loss": 0.4062,
+      "step": 19910
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.1286217312300663e-05,
+      "loss": 0.5654,
+      "step": 19920
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.1281696583860923e-05,
+      "loss": 0.4624,
+      "step": 19930
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.1277174012268207e-05,
+      "loss": 0.5182,
+      "step": 19940
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.1272649599726764e-05,
+      "loss": 0.8319,
+      "step": 19950
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.1268123348441735e-05,
+      "loss": 0.7917,
+      "step": 19960
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.1263595260619163e-05,
+      "loss": 0.7896,
+      "step": 19970
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.1259065338465981e-05,
+      "loss": 0.5326,
+      "step": 19980
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.125453358419002e-05,
+      "loss": 0.7379,
+      "step": 19990
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.125e-05,
+      "loss": 0.7768,
+      "step": 20000
+    },
+    {
+      "epoch": 9.04,
+      "eval_accuracy": 0.8589473684210527,
+      "eval_f1": 0.8589473684210527,
+      "eval_loss": 0.7489694356918335,
+      "eval_runtime": 770.3856,
+      "eval_samples_per_second": 6.166,
+      "eval_steps_per_second": 1.542,
+      "step": 20000
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.1245464588105537e-05,
+      "loss": 0.4541,
+      "step": 20010
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.1240927350717133e-05,
+      "loss": 0.6959,
+      "step": 20020
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.123638829004618e-05,
+      "loss": 0.8044,
+      "step": 20030
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.123184740830497e-05,
+      "loss": 0.754,
+      "step": 20040
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.1227304707706665e-05,
+      "loss": 0.3744,
+      "step": 20050
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.1222760190465327e-05,
+      "loss": 0.8936,
+      "step": 20060
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.1218213858795896e-05,
+      "loss": 1.1047,
+      "step": 20070
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.1213665714914198e-05,
+      "loss": 0.3212,
+      "step": 20080
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.1209115761036948e-05,
+      "loss": 0.5225,
+      "step": 20090
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.1204563999381733e-05,
+      "loss": 0.5067,
+      "step": 20100
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.1200010432167028e-05,
+      "loss": 0.692,
+      "step": 20110
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.1195455061612187e-05,
+      "loss": 0.4932,
+      "step": 20120
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.1190897889937441e-05,
+      "loss": 0.528,
+      "step": 20130
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.1186338919363903e-05,
+      "loss": 0.5525,
+      "step": 20140
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.1181778152113556e-05,
+      "loss": 0.6876,
+      "step": 20150
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.1177215590409265e-05,
+      "loss": 0.7504,
+      "step": 20160
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.1172651236474768e-05,
+      "loss": 0.7971,
+      "step": 20170
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.1168085092534673e-05,
+      "loss": 0.7278,
+      "step": 20180
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.1163517160814464e-05,
+      "loss": 0.3767,
+      "step": 20190
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.1158947443540496e-05,
+      "loss": 0.924,
+      "step": 20200
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.1154375942939992e-05,
+      "loss": 0.6778,
+      "step": 20210
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.1149802661241051e-05,
+      "loss": 0.7237,
+      "step": 20220
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.1145227600672627e-05,
+      "loss": 0.7282,
+      "step": 20230
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.1140650763464555e-05,
+      "loss": 0.5504,
+      "step": 20240
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.1136072151847529e-05,
+      "loss": 0.7132,
+      "step": 20250
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.1131491768053105e-05,
+      "loss": 0.6256,
+      "step": 20260
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.1126909614313711e-05,
+      "loss": 0.3614,
+      "step": 20270
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.1122325692862631e-05,
+      "loss": 0.3448,
+      "step": 20280
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.1117740005934013e-05,
+      "loss": 0.6783,
+      "step": 20290
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.1113152555762865e-05,
+      "loss": 0.8791,
+      "step": 20300
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.1108563344585056e-05,
+      "loss": 0.7556,
+      "step": 20310
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.1103972374637305e-05,
+      "loss": 0.6908,
+      "step": 20320
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.1099379648157206e-05,
+      "loss": 0.9076,
+      "step": 20330
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.1094785167383189e-05,
+      "loss": 0.6393,
+      "step": 20340
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.1090188934554552e-05,
+      "loss": 0.4002,
+      "step": 20350
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.1085590951911442e-05,
+      "loss": 0.5696,
+      "step": 20360
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.108099122169486e-05,
+      "loss": 0.5739,
+      "step": 20370
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.1076389746146659e-05,
+      "loss": 0.8006,
+      "step": 20380
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.1071786527509544e-05,
+      "loss": 0.7159,
+      "step": 20390
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.1067181568027065e-05,
+      "loss": 0.5583,
+      "step": 20400
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.1062574869943623e-05,
+      "loss": 0.4653,
+      "step": 20410
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.1057966435504468e-05,
+      "loss": 0.8138,
+      "step": 20420
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.1053356266955699e-05,
+      "loss": 0.9701,
+      "step": 20430
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.1048744366544248e-05,
+      "loss": 1.121,
+      "step": 20440
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.1044130736517906e-05,
+      "loss": 0.6044,
+      "step": 20450
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.1039515379125297e-05,
+      "loss": 0.7948,
+      "step": 20460
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.1034898296615888e-05,
+      "loss": 0.7339,
+      "step": 20470
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.1030279491239996e-05,
+      "loss": 0.6781,
+      "step": 20480
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.1025658965248762e-05,
+      "loss": 0.5691,
+      "step": 20490
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.1021036720894182e-05,
+      "loss": 0.7411,
+      "step": 20500
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.1016412760429078e-05,
+      "loss": 0.8556,
+      "step": 20510
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.1011787086107109e-05,
+      "loss": 0.8254,
+      "step": 20520
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.100715970018278e-05,
+      "loss": 0.6314,
+      "step": 20530
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.1002530604911416e-05,
+      "loss": 0.6799,
+      "step": 20540
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.0997899802549185e-05,
+      "loss": 0.7688,
+      "step": 20550
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.0993267295353082e-05,
+      "loss": 0.755,
+      "step": 20560
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.0988633085580938e-05,
+      "loss": 0.6034,
+      "step": 20570
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.0983997175491409e-05,
+      "loss": 0.6137,
+      "step": 20580
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.0979359567343977e-05,
+      "loss": 0.5719,
+      "step": 20590
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.0974720263398964e-05,
+      "loss": 0.6048,
+      "step": 20600
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.0970079265917503e-05,
+      "loss": 0.7313,
+      "step": 20610
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.0965436577161566e-05,
+      "loss": 0.5351,
+      "step": 20620
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.0960792199393936e-05,
+      "loss": 0.71,
+      "step": 20630
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.0956146134878232e-05,
+      "loss": 0.8728,
+      "step": 20640
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.0951498385878888e-05,
+      "loss": 1.0162,
+      "step": 20650
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.0946848954661161e-05,
+      "loss": 0.7644,
+      "step": 20660
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.0942197843491125e-05,
+      "loss": 0.7666,
+      "step": 20670
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.0937545054635673e-05,
+      "loss": 0.4105,
+      "step": 20680
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.0932890590362526e-05,
+      "loss": 0.6279,
+      "step": 20690
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.0928234452940207e-05,
+      "loss": 0.6136,
+      "step": 20700
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.0923576644638063e-05,
+      "loss": 0.7239,
+      "step": 20710
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.0918917167726252e-05,
+      "loss": 0.5398,
+      "step": 20720
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.0914256024475743e-05,
+      "loss": 0.9341,
+      "step": 20730
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.090959321715833e-05,
+      "loss": 0.899,
+      "step": 20740
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.0904928748046601e-05,
+      "loss": 0.5511,
+      "step": 20750
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.0900262619413965e-05,
+      "loss": 0.8734,
+      "step": 20760
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.0895594833534635e-05,
+      "loss": 0.865,
+      "step": 20770
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.089092539268363e-05,
+      "loss": 0.5086,
+      "step": 20780
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.0886254299136787e-05,
+      "loss": 0.9701,
+      "step": 20790
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.088158155517073e-05,
+      "loss": 0.9013,
+      "step": 20800
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.0876907163062907e-05,
+      "loss": 0.8956,
+      "step": 20810
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.0872231125091554e-05,
+      "loss": 0.6707,
+      "step": 20820
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.0867553443535718e-05,
+      "loss": 0.7424,
+      "step": 20830
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.0862874120675244e-05,
+      "loss": 0.7948,
+      "step": 20840
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.0858193158790773e-05,
+      "loss": 0.895,
+      "step": 20850
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.0853510560163755e-05,
+      "loss": 0.5475,
+      "step": 20860
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.0848826327076426e-05,
+      "loss": 0.6243,
+      "step": 20870
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.0844140461811832e-05,
+      "loss": 0.2264,
+      "step": 20880
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.0839452966653798e-05,
+      "loss": 0.8413,
+      "step": 20890
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.0834763843886956e-05,
+      "loss": 0.9677,
+      "step": 20900
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.083007309579673e-05,
+      "loss": 0.4505,
+      "step": 20910
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.0825380724669328e-05,
+      "loss": 0.8381,
+      "step": 20920
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.0820686732791763e-05,
+      "loss": 0.3911,
+      "step": 20930
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.081599112245182e-05,
+      "loss": 0.4558,
+      "step": 20940
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.081129389593809e-05,
+      "loss": 0.7595,
+      "step": 20950
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.080659505553994e-05,
+      "loss": 0.6722,
+      "step": 20960
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.0801894603547529e-05,
+      "loss": 0.6681,
+      "step": 20970
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.07971925422518e-05,
+      "loss": 0.9848,
+      "step": 20980
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.0792488873944481e-05,
+      "loss": 0.8077,
+      "step": 20990
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.078778360091808e-05,
+      "loss": 0.6053,
+      "step": 21000
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.0783076725465896e-05,
+      "loss": 0.2675,
+      "step": 21010
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.0778368249881996e-05,
+      "loss": 0.7754,
+      "step": 21020
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.0773658176461242e-05,
+      "loss": 0.4894,
+      "step": 21030
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.0768946507499255e-05,
+      "loss": 0.9299,
+      "step": 21040
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.0764233245292457e-05,
+      "loss": 0.7061,
+      "step": 21050
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.0759518392138026e-05,
+      "loss": 0.5992,
+      "step": 21060
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.0754801950333931e-05,
+      "loss": 1.0293,
+      "step": 21070
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.0750083922178904e-05,
+      "loss": 0.9621,
+      "step": 21080
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.0745364309972454e-05,
+      "loss": 1.074,
+      "step": 21090
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.0740643116014868e-05,
+      "loss": 1.0342,
+      "step": 21100
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.0735920342607193e-05,
+      "loss": 0.5996,
+      "step": 21110
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.0731195992051254e-05,
+      "loss": 0.4885,
+      "step": 21120
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.0726470066649639e-05,
+      "loss": 0.8302,
+      "step": 21130
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.0721742568705713e-05,
+      "loss": 0.8496,
+      "step": 21140
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.0717013500523595e-05,
+      "loss": 0.6095,
+      "step": 21150
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.0712282864408178e-05,
+      "loss": 0.7966,
+      "step": 21160
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.0707550662665117e-05,
+      "loss": 0.7261,
+      "step": 21170
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.0702816897600825e-05,
+      "loss": 0.5543,
+      "step": 21180
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.0698081571522491e-05,
+      "loss": 0.4776,
+      "step": 21190
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.0693344686738045e-05,
+      "loss": 0.7805,
+      "step": 21200
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.0688606245556196e-05,
+      "loss": 0.4526,
+      "step": 21210
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.0683866250286394e-05,
+      "loss": 0.6296,
+      "step": 21220
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.0679124703238862e-05,
+      "loss": 0.7896,
+      "step": 21230
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.0674381606724573e-05,
+      "loss": 0.75,
+      "step": 21240
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.0669636963055247e-05,
+      "loss": 0.4908,
+      "step": 21250
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.0664890774543372e-05,
+      "loss": 0.5771,
+      "step": 21260
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.0660143043502181e-05,
+      "loss": 0.2833,
+      "step": 21270
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.0655393772245661e-05,
+      "loss": 0.495,
+      "step": 21280
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.0650642963088549e-05,
+      "loss": 0.6387,
+      "step": 21290
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.0645890618346329e-05,
+      "loss": 0.718,
+      "step": 21300
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.0641136740335238e-05,
+      "loss": 0.6408,
+      "step": 21310
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.063638133137226e-05,
+      "loss": 0.7938,
+      "step": 21320
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.0631624393775125e-05,
+      "loss": 0.4977,
+      "step": 21330
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.0626865929862303e-05,
+      "loss": 0.7993,
+      "step": 21340
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.062210594195301e-05,
+      "loss": 0.9612,
+      "step": 21350
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.0617344432367208e-05,
+      "loss": 0.6478,
+      "step": 21360
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.0612581403425603e-05,
+      "loss": 1.1047,
+      "step": 21370
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.0607816857449632e-05,
+      "loss": 0.6164,
+      "step": 21380
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.060305079676148e-05,
+      "loss": 0.5161,
+      "step": 21390
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.0598283223684064e-05,
+      "loss": 0.7603,
+      "step": 21400
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.0593514140541044e-05,
+      "loss": 0.827,
+      "step": 21410
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.0588743549656812e-05,
+      "loss": 0.9353,
+      "step": 21420
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.0583971453356499e-05,
+      "loss": 0.7165,
+      "step": 21430
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.057919785396596e-05,
+      "loss": 0.5806,
+      "step": 21440
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.0574422753811796e-05,
+      "loss": 0.8537,
+      "step": 21450
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.056964615522133e-05,
+      "loss": 0.587,
+      "step": 21460
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.0564868060522619e-05,
+      "loss": 0.6138,
+      "step": 21470
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.0560088472044448e-05,
+      "loss": 0.6889,
+      "step": 21480
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.0555307392116327e-05,
+      "loss": 1.3632,
+      "step": 21490
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.0550524823068504e-05,
+      "loss": 0.5563,
+      "step": 21500
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.0545740767231936e-05,
+      "loss": 0.6569,
+      "step": 21510
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.054095522693832e-05,
+      "loss": 0.4933,
+      "step": 21520
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.0536168204520068e-05,
+      "loss": 1.0152,
+      "step": 21530
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.0531379702310317e-05,
+      "loss": 0.5184,
+      "step": 21540
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.0526589722642927e-05,
+      "loss": 0.9484,
+      "step": 21550
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.0521798267852471e-05,
+      "loss": 0.7081,
+      "step": 21560
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.051700534027425e-05,
+      "loss": 0.6213,
+      "step": 21570
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.0512210942244275e-05,
+      "loss": 0.6249,
+      "step": 21580
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.0507415076099281e-05,
+      "loss": 0.657,
+      "step": 21590
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.0502617744176715e-05,
+      "loss": 0.6751,
+      "step": 21600
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.0497818948814732e-05,
+      "loss": 0.5777,
+      "step": 21610
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.0493018692352216e-05,
+      "loss": 0.6597,
+      "step": 21620
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.0488216977128745e-05,
+      "loss": 0.6365,
+      "step": 21630
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.0483413805484625e-05,
+      "loss": 0.5279,
+      "step": 21640
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.0478609179760854e-05,
+      "loss": 0.2789,
+      "step": 21650
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.0473803102299157e-05,
+      "loss": 0.6284,
+      "step": 21660
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.0468995575441954e-05,
+      "loss": 0.9454,
+      "step": 21670
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.0464186601532374e-05,
+      "loss": 1.0541,
+      "step": 21680
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.0459376182914256e-05,
+      "loss": 0.676,
+      "step": 21690
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.0454564321932134e-05,
+      "loss": 1.0218,
+      "step": 21700
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.0449751020931255e-05,
+      "loss": 0.6263,
+      "step": 21710
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.0444936282257564e-05,
+      "loss": 0.6173,
+      "step": 21720
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.0440120108257702e-05,
+      "loss": 1.0447,
+      "step": 21730
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.043530250127902e-05,
+      "loss": 0.5234,
+      "step": 21740
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.0430483463669552e-05,
+      "loss": 0.6918,
+      "step": 21750
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.0425662997778048e-05,
+      "loss": 0.5981,
+      "step": 21760
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.042084110595394e-05,
+      "loss": 0.6197,
+      "step": 21770
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.0416017790547357e-05,
+      "loss": 0.4061,
+      "step": 21780
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.041119305390913e-05,
+      "loss": 0.8991,
+      "step": 21790
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.0406366898390772e-05,
+      "loss": 0.3839,
+      "step": 21800
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.0401539326344498e-05,
+      "loss": 0.902,
+      "step": 21810
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.03967103401232e-05,
+      "loss": 0.9921,
+      "step": 21820
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.0391879942080475e-05,
+      "loss": 0.7106,
+      "step": 21830
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.0387048134570596e-05,
+      "loss": 0.632,
+      "step": 21840
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.0382214919948527e-05,
+      "loss": 0.6111,
+      "step": 21850
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.0377380300569923e-05,
+      "loss": 0.3693,
+      "step": 21860
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.0372544278791114e-05,
+      "loss": 0.6298,
+      "step": 21870
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.0367706856969119e-05,
+      "loss": 0.474,
+      "step": 21880
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.0362868037461638e-05,
+      "loss": 0.5683,
+      "step": 21890
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.0358027822627057e-05,
+      "loss": 0.8299,
+      "step": 21900
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.0353186214824433e-05,
+      "loss": 0.7203,
+      "step": 21910
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.034834321641351e-05,
+      "loss": 0.3176,
+      "step": 21920
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.0343498829754703e-05,
+      "loss": 0.5711,
+      "step": 21930
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.0338653057209111e-05,
+      "loss": 0.5252,
+      "step": 21940
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.0333805901138504e-05,
+      "loss": 0.5265,
+      "step": 21950
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.0328957363905325e-05,
+      "loss": 0.7713,
+      "step": 21960
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.0324107447872695e-05,
+      "loss": 0.5364,
+      "step": 21970
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.03192561554044e-05,
+      "loss": 0.6878,
+      "step": 21980
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.0314403488864907e-05,
+      "loss": 0.7172,
+      "step": 21990
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.0309549450619342e-05,
+      "loss": 0.6793,
+      "step": 22000
+    },
+    {
+      "epoch": 10.04,
+      "eval_accuracy": 0.8557894736842105,
+      "eval_f1": 0.8557894736842105,
+      "eval_loss": 0.7730118036270142,
+      "eval_runtime": 761.7959,
+      "eval_samples_per_second": 6.235,
+      "eval_steps_per_second": 1.559,
+      "step": 22000
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.0304694043033502e-05,
+      "loss": 0.9001,
+      "step": 22010
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.0299837268473863e-05,
+      "loss": 0.7062,
+      "step": 22020
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.0294979129307548e-05,
+      "loss": 0.4986,
+      "step": 22030
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.0290119627902361e-05,
+      "loss": 0.563,
+      "step": 22040
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.0285258766626762e-05,
+      "loss": 0.7271,
+      "step": 22050
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.0280396547849873e-05,
+      "loss": 0.782,
+      "step": 22060
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.0275532973941487e-05,
+      "loss": 0.686,
+      "step": 22070
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.0270668047272045e-05,
+      "loss": 0.6969,
+      "step": 22080
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.0265801770212656e-05,
+      "loss": 0.6656,
+      "step": 22090
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.0260934145135086e-05,
+      "loss": 0.5161,
+      "step": 22100
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.025606517441175e-05,
+      "loss": 0.9492,
+      "step": 22110
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.0251194860415732e-05,
+      "loss": 0.7005,
+      "step": 22120
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.0246323205520757e-05,
+      "loss": 1.0263,
+      "step": 22130
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.0241450212101217e-05,
+      "loss": 0.4128,
+      "step": 22140
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.0236575882532147e-05,
+      "loss": 0.601,
+      "step": 22150
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.0231700219189237e-05,
+      "loss": 0.391,
+      "step": 22160
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.0226823224448825e-05,
+      "loss": 0.623,
+      "step": 22170
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.0221944900687897e-05,
+      "loss": 0.5038,
+      "step": 22180
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.0217065250284094e-05,
+      "loss": 0.5027,
+      "step": 22190
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.0212184275615691e-05,
+      "loss": 0.7453,
+      "step": 22200
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.0207301979061625e-05,
+      "loss": 0.6545,
+      "step": 22210
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.0202418363001462e-05,
+      "loss": 0.8928,
+      "step": 22220
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.0197533429815416e-05,
+      "loss": 0.759,
+      "step": 22230
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.0192647181884347e-05,
+      "loss": 0.6488,
+      "step": 22240
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.018775962158975e-05,
+      "loss": 0.4284,
+      "step": 22250
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.018287075131377e-05,
+      "loss": 0.7273,
+      "step": 22260
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.0177980573439176e-05,
+      "loss": 0.8202,
+      "step": 22270
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.017308909034938e-05,
+      "loss": 0.6795,
+      "step": 22280
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.0168196304428437e-05,
+      "loss": 0.7264,
+      "step": 22290
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.0163302218061028e-05,
+      "loss": 0.5813,
+      "step": 22300
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.0158406833632473e-05,
+      "loss": 0.6163,
+      "step": 22310
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.015351015352872e-05,
+      "loss": 0.4763,
+      "step": 22320
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.0148612180136355e-05,
+      "loss": 0.5641,
+      "step": 22330
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.0143712915842589e-05,
+      "loss": 0.704,
+      "step": 22340
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.0138812363035263e-05,
+      "loss": 0.4373,
+      "step": 22350
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.013391052410285e-05,
+      "loss": 0.8222,
+      "step": 22360
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.0129007401434443e-05,
+      "loss": 0.7842,
+      "step": 22370
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.0124102997419767e-05,
+      "loss": 0.3431,
+      "step": 22380
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.0119197314449169e-05,
+      "loss": 0.7633,
+      "step": 22390
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.0114290354913615e-05,
+      "loss": 0.5489,
+      "step": 22400
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.0109382121204702e-05,
+      "loss": 0.5858,
+      "step": 22410
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.0104472615714642e-05,
+      "loss": 0.5548,
+      "step": 22420
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.0099561840836272e-05,
+      "loss": 0.6733,
+      "step": 22430
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.0094649798963037e-05,
+      "loss": 0.8389,
+      "step": 22440
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.008973649248901e-05,
+      "loss": 0.4196,
+      "step": 22450
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.0084821923808877e-05,
+      "loss": 0.7719,
+      "step": 22460
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.007990609531794e-05,
+      "loss": 0.4378,
+      "step": 22470
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.0074989009412116e-05,
+      "loss": 0.3773,
+      "step": 22480
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.0070070668487926e-05,
+      "loss": 0.7681,
+      "step": 22490
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.0065151074942516e-05,
+      "loss": 0.4419,
+      "step": 22500
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.0060230231173632e-05,
+      "loss": 0.7388,
+      "step": 22510
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.0055308139579639e-05,
+      "loss": 0.678,
+      "step": 22520
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.0050384802559497e-05,
+      "loss": 0.4424,
+      "step": 22530
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.0045460222512785e-05,
+      "loss": 0.5644,
+      "step": 22540
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.0040534401839687e-05,
+      "loss": 0.9504,
+      "step": 22550
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.003560734294098e-05,
+      "loss": 0.5278,
+      "step": 22560
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.003067904821806e-05,
+      "loss": 0.546,
+      "step": 22570
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.0025749520072912e-05,
+      "loss": 1.0441,
+      "step": 22580
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.0020818760908133e-05,
+      "loss": 0.8724,
+      "step": 22590
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.0015886773126914e-05,
+      "loss": 0.9456,
+      "step": 22600
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.0010953559133043e-05,
+      "loss": 0.662,
+      "step": 22610
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.0006019121330913e-05,
+      "loss": 0.7724,
+      "step": 22620
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.0001083462125504e-05,
+      "loss": 1.0125,
+      "step": 22630
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 9.996146583922401e-06,
+      "loss": 0.5953,
+      "step": 22640
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 9.991208489127775e-06,
+      "loss": 0.871,
+      "step": 22650
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 9.986269180148397e-06,
+      "loss": 0.6984,
+      "step": 22660
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 9.981328659391623e-06,
+      "loss": 0.6937,
+      "step": 22670
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 9.976386929265403e-06,
+      "loss": 0.6753,
+      "step": 22680
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 9.971443992178277e-06,
+      "loss": 0.6009,
+      "step": 22690
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 9.966499850539375e-06,
+      "loss": 0.9312,
+      "step": 22700
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 9.961554506758408e-06,
+      "loss": 0.5313,
+      "step": 22710
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 9.956607963245676e-06,
+      "loss": 0.5486,
+      "step": 22720
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 9.951660222412069e-06,
+      "loss": 0.6269,
+      "step": 22730
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 9.94671128666905e-06,
+      "loss": 0.7164,
+      "step": 22740
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 9.941761158428675e-06,
+      "loss": 0.4487,
+      "step": 22750
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 9.936809840103575e-06,
+      "loss": 0.851,
+      "step": 22760
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 9.931857334106958e-06,
+      "loss": 0.6416,
+      "step": 22770
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 9.926903642852621e-06,
+      "loss": 0.9216,
+      "step": 22780
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 9.921948768754931e-06,
+      "loss": 0.6046,
+      "step": 22790
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 9.916992714228835e-06,
+      "loss": 0.6689,
+      "step": 22800
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 9.91203548168985e-06,
+      "loss": 1.0292,
+      "step": 22810
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 9.90707707355407e-06,
+      "loss": 0.7029,
+      "step": 22820
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 9.90211749223817e-06,
+      "loss": 1.2995,
+      "step": 22830
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 9.897156740159384e-06,
+      "loss": 0.5994,
+      "step": 22840
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 9.892194819735525e-06,
+      "loss": 0.8809,
+      "step": 22850
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 9.887231733384972e-06,
+      "loss": 0.8827,
+      "step": 22860
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 9.882267483526669e-06,
+      "loss": 0.7032,
+      "step": 22870
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 9.877302072580139e-06,
+      "loss": 0.4283,
+      "step": 22880
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 9.872335502965455e-06,
+      "loss": 0.5838,
+      "step": 22890
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 9.867367777103269e-06,
+      "loss": 0.5258,
+      "step": 22900
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 9.862398897414786e-06,
+      "loss": 0.5202,
+      "step": 22910
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 9.857428866321781e-06,
+      "loss": 0.5962,
+      "step": 22920
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 9.852457686246583e-06,
+      "loss": 0.4088,
+      "step": 22930
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 9.847485359612086e-06,
+      "loss": 0.9281,
+      "step": 22940
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 9.842511888841744e-06,
+      "loss": 0.5964,
+      "step": 22950
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 9.837537276359564e-06,
+      "loss": 0.6418,
+      "step": 22960
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 9.832561524590115e-06,
+      "loss": 0.4186,
+      "step": 22970
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 9.82758463595851e-06,
+      "loss": 0.6558,
+      "step": 22980
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 9.822606612890431e-06,
+      "loss": 0.765,
+      "step": 22990
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 9.817627457812105e-06,
+      "loss": 0.9792,
+      "step": 23000
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 9.812647173150313e-06,
+      "loss": 0.9115,
+      "step": 23010
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 9.807665761332382e-06,
+      "loss": 0.3789,
+      "step": 23020
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 9.80268322478619e-06,
+      "loss": 0.4179,
+      "step": 23030
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 9.797699565940168e-06,
+      "loss": 0.861,
+      "step": 23040
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 9.792714787223294e-06,
+      "loss": 0.5827,
+      "step": 23050
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 9.787728891065084e-06,
+      "loss": 0.6217,
+      "step": 23060
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 9.782741879895602e-06,
+      "loss": 0.7407,
+      "step": 23070
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 9.777753756145458e-06,
+      "loss": 0.5362,
+      "step": 23080
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 9.772764522245806e-06,
+      "loss": 0.6387,
+      "step": 23090
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 9.767774180628335e-06,
+      "loss": 0.4657,
+      "step": 23100
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 9.762782733725277e-06,
+      "loss": 0.6733,
+      "step": 23110
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 9.757790183969404e-06,
+      "loss": 0.6548,
+      "step": 23120
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 9.752796533794022e-06,
+      "loss": 0.8012,
+      "step": 23130
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 9.747801785632979e-06,
+      "loss": 0.7032,
+      "step": 23140
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 9.74280594192065e-06,
+      "loss": 0.7914,
+      "step": 23150
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 9.737809005091952e-06,
+      "loss": 0.7735,
+      "step": 23160
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 9.732810977582329e-06,
+      "loss": 0.5257,
+      "step": 23170
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 9.727811861827762e-06,
+      "loss": 0.7142,
+      "step": 23180
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 9.722811660264757e-06,
+      "loss": 0.5854,
+      "step": 23190
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 9.717810375330351e-06,
+      "loss": 1.0973,
+      "step": 23200
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 9.71280800946211e-06,
+      "loss": 0.4392,
+      "step": 23210
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 9.707804565098127e-06,
+      "loss": 0.5886,
+      "step": 23220
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 9.702800044677022e-06,
+      "loss": 0.9332,
+      "step": 23230
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 9.69779445063793e-06,
+      "loss": 0.639,
+      "step": 23240
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 9.692787785420525e-06,
+      "loss": 0.4684,
+      "step": 23250
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 9.687780051464993e-06,
+      "loss": 0.7447,
+      "step": 23260
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 9.682771251212038e-06,
+      "loss": 0.5868,
+      "step": 23270
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 9.677761387102896e-06,
+      "loss": 0.7026,
+      "step": 23280
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 9.672750461579305e-06,
+      "loss": 0.4704,
+      "step": 23290
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 9.667738477083536e-06,
+      "loss": 0.7248,
+      "step": 23300
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 9.66272543605837e-06,
+      "loss": 0.7306,
+      "step": 23310
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 9.657711340947096e-06,
+      "loss": 0.578,
+      "step": 23320
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 9.652696194193527e-06,
+      "loss": 0.5586,
+      "step": 23330
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 9.647679998241984e-06,
+      "loss": 0.7175,
+      "step": 23340
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 9.642662755537301e-06,
+      "loss": 0.4754,
+      "step": 23350
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 9.63764446852482e-06,
+      "loss": 0.4033,
+      "step": 23360
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 9.632625139650395e-06,
+      "loss": 1.0351,
+      "step": 23370
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 9.627604771360382e-06,
+      "loss": 0.5524,
+      "step": 23380
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 9.622583366101652e-06,
+      "loss": 0.4591,
+      "step": 23390
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 9.617560926321575e-06,
+      "loss": 0.9319,
+      "step": 23400
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 9.612537454468024e-06,
+      "loss": 1.0524,
+      "step": 23410
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 9.607512952989385e-06,
+      "loss": 0.6202,
+      "step": 23420
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 9.602487424334532e-06,
+      "loss": 0.8657,
+      "step": 23430
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 9.597460870952849e-06,
+      "loss": 0.8354,
+      "step": 23440
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 9.59243329529422e-06,
+      "loss": 0.886,
+      "step": 23450
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 9.58740469980902e-06,
+      "loss": 0.6174,
+      "step": 23460
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 9.58237508694813e-06,
+      "loss": 0.7236,
+      "step": 23470
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 9.577344459162918e-06,
+      "loss": 0.4743,
+      "step": 23480
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 9.572312818905252e-06,
+      "loss": 0.6526,
+      "step": 23490
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 9.567280168627493e-06,
+      "loss": 1.0537,
+      "step": 23500
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 9.562246510782496e-06,
+      "loss": 0.1958,
+      "step": 23510
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 9.5572118478236e-06,
+      "loss": 0.5649,
+      "step": 23520
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 9.55217618220464e-06,
+      "loss": 0.8306,
+      "step": 23530
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 9.54713951637994e-06,
+      "loss": 1.0081,
+      "step": 23540
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 9.542101852804307e-06,
+      "loss": 0.7292,
+      "step": 23550
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 9.537063193933041e-06,
+      "loss": 0.7931,
+      "step": 23560
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 9.53202354222192e-06,
+      "loss": 0.4955,
+      "step": 23570
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 9.52698290012721e-06,
+      "loss": 0.949,
+      "step": 23580
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 9.521941270105657e-06,
+      "loss": 1.0758,
+      "step": 23590
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 9.516898654614492e-06,
+      "loss": 1.0021,
+      "step": 23600
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 9.511855056111426e-06,
+      "loss": 0.7936,
+      "step": 23610
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 9.506810477054645e-06,
+      "loss": 0.7424,
+      "step": 23620
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 9.501764919902818e-06,
+      "loss": 0.5233,
+      "step": 23630
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 9.496718387115085e-06,
+      "loss": 0.5166,
+      "step": 23640
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 9.491670881151067e-06,
+      "loss": 0.6646,
+      "step": 23650
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 9.486622404470855e-06,
+      "loss": 0.6662,
+      "step": 23660
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 9.481572959535019e-06,
+      "loss": 0.618,
+      "step": 23670
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 9.476522548804596e-06,
+      "loss": 0.6063,
+      "step": 23680
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 9.47147117474109e-06,
+      "loss": 0.3129,
+      "step": 23690
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 9.466418839806486e-06,
+      "loss": 0.9168,
+      "step": 23700
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 9.461365546463226e-06,
+      "loss": 0.7685,
+      "step": 23710
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 9.456311297174228e-06,
+      "loss": 0.3527,
+      "step": 23720
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 9.451256094402867e-06,
+      "loss": 0.5965,
+      "step": 23730
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 9.44619994061299e-06,
+      "loss": 0.9973,
+      "step": 23740
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 9.441142838268906e-06,
+      "loss": 0.8503,
+      "step": 23750
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 9.436084789835383e-06,
+      "loss": 0.8314,
+      "step": 23760
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 9.431025797777654e-06,
+      "loss": 0.4643,
+      "step": 23770
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 9.425965864561408e-06,
+      "loss": 0.9074,
+      "step": 23780
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 9.420904992652797e-06,
+      "loss": 0.4391,
+      "step": 23790
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 9.41584318451843e-06,
+      "loss": 0.9418,
+      "step": 23800
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 9.410780442625368e-06,
+      "loss": 0.575,
+      "step": 23810
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 9.405716769441129e-06,
+      "loss": 0.8775,
+      "step": 23820
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 9.400652167433687e-06,
+      "loss": 0.6625,
+      "step": 23830
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 9.39558663907147e-06,
+      "loss": 1.0413,
+      "step": 23840
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 9.390520186823354e-06,
+      "loss": 0.3134,
+      "step": 23850
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 9.385452813158662e-06,
+      "loss": 0.722,
+      "step": 23860
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 9.380384520547176e-06,
+      "loss": 0.8009,
+      "step": 23870
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 9.375315311459116e-06,
+      "loss": 0.6277,
+      "step": 23880
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 9.370245188365156e-06,
+      "loss": 0.5125,
+      "step": 23890
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 9.365174153736414e-06,
+      "loss": 0.493,
+      "step": 23900
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 9.360102210044441e-06,
+      "loss": 0.7453,
+      "step": 23910
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 9.355029359761253e-06,
+      "loss": 0.439,
+      "step": 23920
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 9.34995560535929e-06,
+      "loss": 0.691,
+      "step": 23930
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 9.344880949311437e-06,
+      "loss": 0.6713,
+      "step": 23940
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 9.33980539409102e-06,
+      "loss": 0.4107,
+      "step": 23950
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 9.334728942171805e-06,
+      "loss": 0.7445,
+      "step": 23960
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 9.329651596027992e-06,
+      "loss": 0.5816,
+      "step": 23970
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 9.324573358134219e-06,
+      "loss": 0.8575,
+      "step": 23980
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 9.319494230965556e-06,
+      "loss": 0.6872,
+      "step": 23990
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 9.314414216997507e-06,
+      "loss": 0.5765,
+      "step": 24000
+    },
+    {
+      "epoch": 11.04,
+      "eval_accuracy": 0.8368421052631579,
+      "eval_f1": 0.8368421052631579,
+      "eval_loss": 0.7751592993736267,
+      "eval_runtime": 757.0231,
+      "eval_samples_per_second": 6.275,
+      "eval_steps_per_second": 1.569,
+      "step": 24000
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 9.309333318706012e-06,
+      "loss": 0.5647,
+      "step": 24010
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 9.304251538567439e-06,
+      "loss": 0.4817,
+      "step": 24020
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 9.299168879058583e-06,
+      "loss": 1.1363,
+      "step": 24030
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 9.29408534265667e-06,
+      "loss": 0.4983,
+      "step": 24040
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 9.289000931839357e-06,
+      "loss": 0.5477,
+      "step": 24050
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 9.283915649084722e-06,
+      "loss": 0.5383,
+      "step": 24060
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 9.278829496871273e-06,
+      "loss": 0.3909,
+      "step": 24070
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 9.273742477677936e-06,
+      "loss": 0.5127,
+      "step": 24080
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 9.268654593984062e-06,
+      "loss": 0.6024,
+      "step": 24090
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 9.263565848269425e-06,
+      "loss": 0.6374,
+      "step": 24100
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 9.258476243014217e-06,
+      "loss": 0.3111,
+      "step": 24110
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 9.253385780699054e-06,
+      "loss": 0.8523,
+      "step": 24120
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 9.248294463804958e-06,
+      "loss": 0.6465,
+      "step": 24130
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 9.24320229481338e-06,
+      "loss": 0.3356,
+      "step": 24140
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 9.238109276206179e-06,
+      "loss": 0.4021,
+      "step": 24150
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 9.233015410465636e-06,
+      "loss": 0.5717,
+      "step": 24160
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 9.22792070007443e-06,
+      "loss": 0.5375,
+      "step": 24170
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 9.222825147515668e-06,
+      "loss": 0.5466,
+      "step": 24180
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 9.217728755272858e-06,
+      "loss": 0.5708,
+      "step": 24190
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 9.212631525829919e-06,
+      "loss": 0.5068,
+      "step": 24200
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 9.207533461671181e-06,
+      "loss": 0.3744,
+      "step": 24210
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 9.202434565281376e-06,
+      "loss": 0.7047,
+      "step": 24220
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 9.197334839145645e-06,
+      "loss": 0.7185,
+      "step": 24230
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 9.19223428574953e-06,
+      "loss": 0.6393,
+      "step": 24240
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 9.187132907578989e-06,
+      "loss": 0.9273,
+      "step": 24250
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 9.18203070712036e-06,
+      "loss": 0.771,
+      "step": 24260
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 9.176927686860397e-06,
+      "loss": 0.5174,
+      "step": 24270
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 9.171823849286254e-06,
+      "loss": 0.1818,
+      "step": 24280
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 9.166719196885473e-06,
+      "loss": 0.8949,
+      "step": 24290
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 9.161613732146007e-06,
+      "loss": 0.5596,
+      "step": 24300
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 9.156507457556189e-06,
+      "loss": 0.8454,
+      "step": 24310
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 9.151400375604762e-06,
+      "loss": 0.8878,
+      "step": 24320
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 9.146292488780854e-06,
+      "loss": 0.5003,
+      "step": 24330
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 9.141183799573984e-06,
+      "loss": 0.7932,
+      "step": 24340
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 9.136074310474071e-06,
+      "loss": 0.6643,
+      "step": 24350
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 9.130964023971411e-06,
+      "loss": 0.4153,
+      "step": 24360
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 9.1258529425567e-06,
+      "loss": 1.0002,
+      "step": 24370
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 9.120741068721017e-06,
+      "loss": 0.6616,
+      "step": 24380
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 9.115628404955823e-06,
+      "loss": 0.7308,
+      "step": 24390
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 9.110514953752975e-06,
+      "loss": 0.6399,
+      "step": 24400
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 9.1054007176047e-06,
+      "loss": 0.6749,
+      "step": 24410
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 9.10028569900362e-06,
+      "loss": 0.9301,
+      "step": 24420
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 9.09516990044273e-06,
+      "loss": 0.7458,
+      "step": 24430
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 9.090053324415412e-06,
+      "loss": 0.6056,
+      "step": 24440
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 9.084935973415417e-06,
+      "loss": 0.7213,
+      "step": 24450
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 9.07981784993688e-06,
+      "loss": 0.4729,
+      "step": 24460
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 9.074698956474321e-06,
+      "loss": 0.7893,
+      "step": 24470
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 9.069579295522614e-06,
+      "loss": 0.6198,
+      "step": 24480
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 9.064458869577028e-06,
+      "loss": 0.5706,
+      "step": 24490
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 9.059337681133194e-06,
+      "loss": 0.9065,
+      "step": 24500
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 9.054215732687118e-06,
+      "loss": 0.6079,
+      "step": 24510
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 9.049093026735176e-06,
+      "loss": 0.4128,
+      "step": 24520
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 9.04396956577411e-06,
+      "loss": 0.4852,
+      "step": 24530
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 9.038845352301034e-06,
+      "loss": 0.6561,
+      "step": 24540
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 9.033720388813426e-06,
+      "loss": 0.3259,
+      "step": 24550
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 9.028594677809138e-06,
+      "loss": 0.7694,
+      "step": 24560
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 9.023468221786367e-06,
+      "loss": 0.6811,
+      "step": 24570
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 9.018341023243696e-06,
+      "loss": 0.6014,
+      "step": 24580
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 9.013213084680053e-06,
+      "loss": 1.0853,
+      "step": 24590
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 9.008084408594737e-06,
+      "loss": 0.664,
+      "step": 24600
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 9.002954997487397e-06,
+      "loss": 0.643,
+      "step": 24610
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 8.997824853858052e-06,
+      "loss": 0.7941,
+      "step": 24620
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 8.992693980207069e-06,
+      "loss": 0.7122,
+      "step": 24630
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 8.987562379035175e-06,
+      "loss": 0.6897,
+      "step": 24640
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 8.982430052843447e-06,
+      "loss": 0.6138,
+      "step": 24650
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 8.97729700413332e-06,
+      "loss": 0.6242,
+      "step": 24660
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 8.97216323540658e-06,
+      "loss": 0.6159,
+      "step": 24670
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 8.967028749165362e-06,
+      "loss": 0.5243,
+      "step": 24680
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 8.961893547912155e-06,
+      "loss": 0.5596,
+      "step": 24690
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 8.95675763414979e-06,
+      "loss": 0.8293,
+      "step": 24700
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 8.951621010381454e-06,
+      "loss": 0.5012,
+      "step": 24710
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 8.946483679110668e-06,
+      "loss": 0.6794,
+      "step": 24720
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 8.941345642841312e-06,
+      "loss": 0.4822,
+      "step": 24730
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 8.936206904077598e-06,
+      "loss": 0.3251,
+      "step": 24740
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 8.931067465324087e-06,
+      "loss": 0.3446,
+      "step": 24750
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 8.925927329085677e-06,
+      "loss": 0.921,
+      "step": 24760
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 8.92078649786761e-06,
+      "loss": 0.5805,
+      "step": 24770
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 8.915644974175466e-06,
+      "loss": 0.6757,
+      "step": 24780
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 8.910502760515155e-06,
+      "loss": 0.8727,
+      "step": 24790
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 8.905359859392936e-06,
+      "loss": 0.5516,
+      "step": 24800
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 8.900216273315393e-06,
+      "loss": 0.8247,
+      "step": 24810
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 8.895072004789447e-06,
+      "loss": 1.006,
+      "step": 24820
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 8.889927056322356e-06,
+      "loss": 0.6281,
+      "step": 24830
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 8.884781430421703e-06,
+      "loss": 0.7306,
+      "step": 24840
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 8.879635129595402e-06,
+      "loss": 0.5848,
+      "step": 24850
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 8.874488156351698e-06,
+      "loss": 0.5106,
+      "step": 24860
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 8.869340513199166e-06,
+      "loss": 0.2741,
+      "step": 24870
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 8.864192202646702e-06,
+      "loss": 0.5024,
+      "step": 24880
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 8.85904322720353e-06,
+      "loss": 0.6582,
+      "step": 24890
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 8.853893589379202e-06,
+      "loss": 0.6676,
+      "step": 24900
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 8.848743291683583e-06,
+      "loss": 0.727,
+      "step": 24910
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 8.843592336626868e-06,
+      "loss": 0.827,
+      "step": 24920
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 8.83844072671957e-06,
+      "loss": 0.4957,
+      "step": 24930
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 8.833288464472524e-06,
+      "loss": 0.7769,
+      "step": 24940
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 8.828135552396875e-06,
+      "loss": 0.4891,
+      "step": 24950
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 8.822981993004093e-06,
+      "loss": 1.0911,
+      "step": 24960
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 8.81782778880596e-06,
+      "loss": 0.4612,
+      "step": 24970
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 8.812672942314572e-06,
+      "loss": 0.9634,
+      "step": 24980
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 8.807517456042335e-06,
+      "loss": 0.7368,
+      "step": 24990
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 8.80236133250198e-06,
+      "loss": 0.5336,
+      "step": 25000
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 8.797204574206529e-06,
+      "loss": 0.7426,
+      "step": 25010
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 8.792047183669327e-06,
+      "loss": 0.7129,
+      "step": 25020
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 8.786889163404021e-06,
+      "loss": 0.7036,
+      "step": 25030
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 8.781730515924576e-06,
+      "loss": 0.5815,
+      "step": 25040
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 8.776571243745244e-06,
+      "loss": 1.1234,
+      "step": 25050
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 8.771411349380598e-06,
+      "loss": 0.9104,
+      "step": 25060
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 8.766250835345503e-06,
+      "loss": 0.6302,
+      "step": 25070
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 8.761089704155136e-06,
+      "loss": 0.7441,
+      "step": 25080
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 8.755927958324966e-06,
+      "loss": 0.9698,
+      "step": 25090
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 8.750765600370768e-06,
+      "loss": 0.596,
+      "step": 25100
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 8.74560263280861e-06,
+      "loss": 0.7446,
+      "step": 25110
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 8.740439058154858e-06,
+      "loss": 0.4741,
+      "step": 25120
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 8.73527487892618e-06,
+      "loss": 0.884,
+      "step": 25130
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 8.730110097639533e-06,
+      "loss": 0.5014,
+      "step": 25140
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 8.724944716812167e-06,
+      "loss": 0.871,
+      "step": 25150
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 8.719778738961629e-06,
+      "loss": 0.7679,
+      "step": 25160
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 8.714612166605747e-06,
+      "loss": 0.6464,
+      "step": 25170
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 8.709445002262655e-06,
+      "loss": 0.4916,
+      "step": 25180
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 8.704277248450758e-06,
+      "loss": 0.9301,
+      "step": 25190
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 8.699108907688763e-06,
+      "loss": 0.6673,
+      "step": 25200
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 8.69393998249565e-06,
+      "loss": 0.7301,
+      "step": 25210
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 8.688770475390698e-06,
+      "loss": 0.73,
+      "step": 25220
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 8.683600388893454e-06,
+      "loss": 0.5313,
+      "step": 25230
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 8.67842972552376e-06,
+      "loss": 0.4326,
+      "step": 25240
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 8.673258487801733e-06,
+      "loss": 0.6077,
+      "step": 25250
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 8.668086678247771e-06,
+      "loss": 0.8506,
+      "step": 25260
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 8.662914299382555e-06,
+      "loss": 0.8871,
+      "step": 25270
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 8.657741353727034e-06,
+      "loss": 0.6094,
+      "step": 25280
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 8.652567843802442e-06,
+      "loss": 0.3646,
+      "step": 25290
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 8.647393772130287e-06,
+      "loss": 0.7661,
+      "step": 25300
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 8.642219141232343e-06,
+      "loss": 0.7078,
+      "step": 25310
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 8.637043953630668e-06,
+      "loss": 0.6364,
+      "step": 25320
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 8.63186821184758e-06,
+      "loss": 0.7387,
+      "step": 25330
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 8.626691918405678e-06,
+      "loss": 0.7593,
+      "step": 25340
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 8.621515075827822e-06,
+      "loss": 0.8417,
+      "step": 25350
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 8.616337686637142e-06,
+      "loss": 0.7509,
+      "step": 25360
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 8.611159753357035e-06,
+      "loss": 0.5077,
+      "step": 25370
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 8.605981278511162e-06,
+      "loss": 0.5972,
+      "step": 25380
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 8.60080226462345e-06,
+      "loss": 0.8684,
+      "step": 25390
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 8.595622714218088e-06,
+      "loss": 0.8425,
+      "step": 25400
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 8.590442629819523e-06,
+      "loss": 0.6553,
+      "step": 25410
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 8.585262013952469e-06,
+      "loss": 1.0543,
+      "step": 25420
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 8.580080869141891e-06,
+      "loss": 0.5518,
+      "step": 25430
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 8.574899197913021e-06,
+      "loss": 0.6666,
+      "step": 25440
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 8.569717002791338e-06,
+      "loss": 0.8162,
+      "step": 25450
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 8.564534286302583e-06,
+      "loss": 0.5814,
+      "step": 25460
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 8.559351050972751e-06,
+      "loss": 0.7725,
+      "step": 25470
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 8.554167299328082e-06,
+      "loss": 0.6521,
+      "step": 25480
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 8.548983033895081e-06,
+      "loss": 0.6795,
+      "step": 25490
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 8.543798257200491e-06,
+      "loss": 0.4602,
+      "step": 25500
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 8.538612971771311e-06,
+      "loss": 0.557,
+      "step": 25510
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 8.533427180134784e-06,
+      "loss": 0.5068,
+      "step": 25520
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 8.52824088481841e-06,
+      "loss": 0.7087,
+      "step": 25530
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 8.523054088349913e-06,
+      "loss": 0.3761,
+      "step": 25540
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 8.517866793257284e-06,
+      "loss": 0.4812,
+      "step": 25550
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 8.512679002068744e-06,
+      "loss": 0.7368,
+      "step": 25560
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 8.50749071731276e-06,
+      "loss": 0.581,
+      "step": 25570
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 8.502301941518045e-06,
+      "loss": 0.2424,
+      "step": 25580
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 8.497112677213532e-06,
+      "loss": 0.4681,
+      "step": 25590
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 8.49192292692842e-06,
+      "loss": 0.8638,
+      "step": 25600
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 8.48673269319212e-06,
+      "loss": 0.4321,
+      "step": 25610
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 8.481541978534293e-06,
+      "loss": 0.5152,
+      "step": 25620
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 8.476350785484828e-06,
+      "loss": 0.5563,
+      "step": 25630
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 8.471159116573852e-06,
+      "loss": 0.9802,
+      "step": 25640
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 8.465966974331722e-06,
+      "loss": 1.1415,
+      "step": 25650
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 8.460774361289021e-06,
+      "loss": 1.0542,
+      "step": 25660
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 8.45558127997657e-06,
+      "loss": 0.7572,
+      "step": 25670
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 8.450387732925411e-06,
+      "loss": 0.6368,
+      "step": 25680
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 8.445193722666814e-06,
+      "loss": 0.8224,
+      "step": 25690
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 8.439999251732282e-06,
+      "loss": 0.4071,
+      "step": 25700
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 8.434804322653534e-06,
+      "loss": 0.8619,
+      "step": 25710
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 8.429608937962513e-06,
+      "loss": 0.3016,
+      "step": 25720
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 8.424413100191391e-06,
+      "loss": 1.1291,
+      "step": 25730
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 8.419216811872551e-06,
+      "loss": 0.5899,
+      "step": 25740
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 8.414020075538606e-06,
+      "loss": 0.3795,
+      "step": 25750
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 8.408822893722379e-06,
+      "loss": 0.5415,
+      "step": 25760
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 8.403625268956915e-06,
+      "loss": 0.3862,
+      "step": 25770
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 8.398427203775473e-06,
+      "loss": 0.7194,
+      "step": 25780
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 8.393228700711524e-06,
+      "loss": 0.7457,
+      "step": 25790
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 8.388029762298759e-06,
+      "loss": 0.6828,
+      "step": 25800
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 8.382830391071072e-06,
+      "loss": 0.4018,
+      "step": 25810
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 8.377630589562581e-06,
+      "loss": 0.6345,
+      "step": 25820
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 8.3724303603076e-06,
+      "loss": 0.6341,
+      "step": 25830
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 8.367229705840664e-06,
+      "loss": 0.4764,
+      "step": 25840
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 8.3620286286965e-06,
+      "loss": 0.8783,
+      "step": 25850
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 8.356827131410057e-06,
+      "loss": 0.5031,
+      "step": 25860
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 8.351625216516476e-06,
+      "loss": 0.8044,
+      "step": 25870
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 8.346422886551112e-06,
+      "loss": 0.3773,
+      "step": 25880
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 8.341220144049517e-06,
+      "loss": 0.4815,
+      "step": 25890
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 8.33601699154744e-06,
+      "loss": 0.6278,
+      "step": 25900
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 8.33081343158084e-06,
+      "loss": 0.5951,
+      "step": 25910
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 8.325609466685867e-06,
+      "loss": 0.7307,
+      "step": 25920
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 8.320405099398867e-06,
+      "loss": 0.4442,
+      "step": 25930
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 8.31520033225639e-06,
+      "loss": 0.5929,
+      "step": 25940
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 8.309995167795172e-06,
+      "loss": 0.7052,
+      "step": 25950
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 8.30478960855215e-06,
+      "loss": 0.5476,
+      "step": 25960
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 8.29958365706445e-06,
+      "loss": 0.4296,
+      "step": 25970
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 8.29437731586939e-06,
+      "loss": 0.5321,
+      "step": 25980
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 8.289170587504476e-06,
+      "loss": 0.7862,
+      "step": 25990
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 8.283963474507402e-06,
+      "loss": 0.4789,
+      "step": 26000
+    },
+    {
+      "epoch": 12.04,
+      "eval_accuracy": 0.848421052631579,
+      "eval_f1": 0.848421052631579,
+      "eval_loss": 0.7902358770370483,
+      "eval_runtime": 759.5604,
+      "eval_samples_per_second": 6.254,
+      "eval_steps_per_second": 1.564,
+      "step": 26000
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 8.278755979416055e-06,
+      "loss": 0.8803,
+      "step": 26010
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 8.273548104768505e-06,
+      "loss": 0.2926,
+      "step": 26020
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 8.268339853103007e-06,
+      "loss": 0.7337,
+      "step": 26030
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 8.263131226957998e-06,
+      "loss": 0.714,
+      "step": 26040
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 8.257922228872097e-06,
+      "loss": 0.8997,
+      "step": 26050
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 8.25271286138411e-06,
+      "loss": 0.8157,
+      "step": 26060
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 8.247503127033023e-06,
+      "loss": 0.6588,
+      "step": 26070
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 8.242293028357989e-06,
+      "loss": 0.7557,
+      "step": 26080
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 8.237082567898349e-06,
+      "loss": 0.5747,
+      "step": 26090
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 8.231871748193622e-06,
+      "loss": 0.4681,
+      "step": 26100
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 8.226660571783495e-06,
+      "loss": 0.662,
+      "step": 26110
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 8.221449041207832e-06,
+      "loss": 0.5678,
+      "step": 26120
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 8.216237159006672e-06,
+      "loss": 0.4976,
+      "step": 26130
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 8.21102492772022e-06,
+      "loss": 0.754,
+      "step": 26140
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 8.20581234988886e-06,
+      "loss": 0.7292,
+      "step": 26150
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 8.200599428053131e-06,
+      "loss": 0.7095,
+      "step": 26160
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 8.19538616475375e-06,
+      "loss": 0.5002,
+      "step": 26170
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 8.190172562531606e-06,
+      "loss": 0.4976,
+      "step": 26180
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 8.184958623927732e-06,
+      "loss": 0.833,
+      "step": 26190
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 8.179744351483353e-06,
+      "loss": 0.3571,
+      "step": 26200
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 8.174529747739829e-06,
+      "loss": 0.5573,
+      "step": 26210
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 8.169314815238705e-06,
+      "loss": 0.4676,
+      "step": 26220
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 8.164099556521667e-06,
+      "loss": 0.3272,
+      "step": 26230
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 8.158883974130576e-06,
+      "loss": 0.4443,
+      "step": 26240
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 8.153668070607439e-06,
+      "loss": 0.5747,
+      "step": 26250
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 8.148451848494422e-06,
+      "loss": 0.7184,
+      "step": 26260
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 8.143235310333854e-06,
+      "loss": 0.5533,
+      "step": 26270
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 8.13801845866821e-06,
+      "loss": 0.4728,
+      "step": 26280
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 8.132801296040121e-06,
+      "loss": 0.3192,
+      "step": 26290
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 8.127583824992369e-06,
+      "loss": 0.7083,
+      "step": 26300
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 8.12236604806788e-06,
+      "loss": 0.8348,
+      "step": 26310
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 8.117147967809741e-06,
+      "loss": 0.9413,
+      "step": 26320
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 8.111929586761183e-06,
+      "loss": 0.7851,
+      "step": 26330
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 8.106710907465576e-06,
+      "loss": 0.7636,
+      "step": 26340
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 8.101491932466441e-06,
+      "loss": 0.5119,
+      "step": 26350
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 8.096272664307448e-06,
+      "loss": 0.7746,
+      "step": 26360
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 8.091053105532402e-06,
+      "loss": 0.3223,
+      "step": 26370
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 8.085833258685251e-06,
+      "loss": 0.8672,
+      "step": 26380
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 8.08061312631009e-06,
+      "loss": 0.5817,
+      "step": 26390
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 8.07539271095114e-06,
+      "loss": 0.6081,
+      "step": 26400
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 8.070172015152773e-06,
+      "loss": 0.5154,
+      "step": 26410
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 8.064951041459496e-06,
+      "loss": 0.664,
+      "step": 26420
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 8.059729792415942e-06,
+      "loss": 0.8406,
+      "step": 26430
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 8.054508270566888e-06,
+      "loss": 0.5142,
+      "step": 26440
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 8.049286478457237e-06,
+      "loss": 0.7025,
+      "step": 26450
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 8.044064418632032e-06,
+      "loss": 0.4909,
+      "step": 26460
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 8.038842093636438e-06,
+      "loss": 0.8021,
+      "step": 26470
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 8.033619506015754e-06,
+      "loss": 0.8264,
+      "step": 26480
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 8.028396658315402e-06,
+      "loss": 0.7216,
+      "step": 26490
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 8.02317355308094e-06,
+      "loss": 0.8995,
+      "step": 26500
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 8.017950192858045e-06,
+      "loss": 0.6565,
+      "step": 26510
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 8.012726580192515e-06,
+      "loss": 0.8193,
+      "step": 26520
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 8.007502717630282e-06,
+      "loss": 0.5443,
+      "step": 26530
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 8.002278607717387e-06,
+      "loss": 0.6395,
+      "step": 26540
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 7.997054253000003e-06,
+      "loss": 0.56,
+      "step": 26550
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 7.991829656024412e-06,
+      "loss": 0.5525,
+      "step": 26560
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 7.986604819337021e-06,
+      "loss": 0.4242,
+      "step": 26570
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 7.981379745484353e-06,
+      "loss": 0.533,
+      "step": 26580
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 7.976154437013045e-06,
+      "loss": 0.597,
+      "step": 26590
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 7.970928896469851e-06,
+      "loss": 0.7054,
+      "step": 26600
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 7.965703126401631e-06,
+      "loss": 0.4532,
+      "step": 26610
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 7.960477129355367e-06,
+      "loss": 0.7798,
+      "step": 26620
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 7.955250907878143e-06,
+      "loss": 0.9396,
+      "step": 26630
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 7.950024464517157e-06,
+      "loss": 0.6024,
+      "step": 26640
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 7.944797801819713e-06,
+      "loss": 0.4973,
+      "step": 26650
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 7.939570922333223e-06,
+      "loss": 0.7844,
+      "step": 26660
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 7.934343828605207e-06,
+      "loss": 0.852,
+      "step": 26670
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 7.929116523183287e-06,
+      "loss": 0.9642,
+      "step": 26680
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 7.923889008615186e-06,
+      "loss": 0.7534,
+      "step": 26690
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 7.91866128744873e-06,
+      "loss": 0.6524,
+      "step": 26700
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 7.913433362231847e-06,
+      "loss": 0.6843,
+      "step": 26710
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 7.908205235512568e-06,
+      "loss": 0.8911,
+      "step": 26720
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 7.902976909839015e-06,
+      "loss": 0.711,
+      "step": 26730
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 7.897748387759413e-06,
+      "loss": 0.6965,
+      "step": 26740
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 7.89251967182208e-06,
+      "loss": 0.906,
+      "step": 26750
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 7.887290764575424e-06,
+      "loss": 0.7533,
+      "step": 26760
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 7.882061668567957e-06,
+      "loss": 0.4359,
+      "step": 26770
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 7.876832386348273e-06,
+      "loss": 0.3871,
+      "step": 26780
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 7.87160292046506e-06,
+      "loss": 0.4603,
+      "step": 26790
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 7.8663732734671e-06,
+      "loss": 0.5004,
+      "step": 26800
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 7.861143447903256e-06,
+      "loss": 0.7104,
+      "step": 26810
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 7.855913446322486e-06,
+      "loss": 0.4944,
+      "step": 26820
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 7.850683271273822e-06,
+      "loss": 0.61,
+      "step": 26830
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 7.845452925306393e-06,
+      "loss": 0.607,
+      "step": 26840
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 7.840222410969402e-06,
+      "loss": 0.7068,
+      "step": 26850
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 7.834991730812147e-06,
+      "loss": 0.3703,
+      "step": 26860
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 7.829760887383987e-06,
+      "loss": 0.7156,
+      "step": 26870
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 7.824529883234379e-06,
+      "loss": 0.6012,
+      "step": 26880
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 7.819298720912848e-06,
+      "loss": 0.8801,
+      "step": 26890
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 7.814067402968998e-06,
+      "loss": 0.6596,
+      "step": 26900
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 7.808835931952513e-06,
+      "loss": 0.6712,
+      "step": 26910
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 7.803604310413144e-06,
+      "loss": 0.3717,
+      "step": 26920
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 7.798372540900723e-06,
+      "loss": 0.7806,
+      "step": 26930
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 7.793140625965152e-06,
+      "loss": 0.4922,
+      "step": 26940
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 7.7879085681564e-06,
+      "loss": 0.2948,
+      "step": 26950
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 7.78267637002451e-06,
+      "loss": 0.8103,
+      "step": 26960
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 7.77744403411959e-06,
+      "loss": 0.2707,
+      "step": 26970
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 7.772211562991819e-06,
+      "loss": 0.6444,
+      "step": 26980
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 7.766978959191438e-06,
+      "loss": 1.2529,
+      "step": 26990
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 7.76174622526876e-06,
+      "loss": 0.6869,
+      "step": 27000
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 7.756513363774147e-06,
+      "loss": 0.5928,
+      "step": 27010
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 7.751280377258036e-06,
+      "loss": 0.2015,
+      "step": 27020
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 7.746047268270923e-06,
+      "loss": 0.5429,
+      "step": 27030
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 7.740814039363363e-06,
+      "loss": 0.7598,
+      "step": 27040
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 7.735580693085962e-06,
+      "loss": 0.6728,
+      "step": 27050
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 7.730347231989397e-06,
+      "loss": 0.455,
+      "step": 27060
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 7.725113658624384e-06,
+      "loss": 0.7046,
+      "step": 27070
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 7.719879975541714e-06,
+      "loss": 0.4476,
+      "step": 27080
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 7.714646185292211e-06,
+      "loss": 0.3398,
+      "step": 27090
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 7.709412290426768e-06,
+      "loss": 0.4458,
+      "step": 27100
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 7.704178293496315e-06,
+      "loss": 1.237,
+      "step": 27110
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 7.698944197051845e-06,
+      "loss": 0.495,
+      "step": 27120
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 7.693710003644391e-06,
+      "loss": 0.918,
+      "step": 27130
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 7.688475715825032e-06,
+      "loss": 1.0978,
+      "step": 27140
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 7.6832413361449e-06,
+      "loss": 0.5083,
+      "step": 27150
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 7.678006867155165e-06,
+      "loss": 0.7703,
+      "step": 27160
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 7.672772311407047e-06,
+      "loss": 0.4654,
+      "step": 27170
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 7.667537671451803e-06,
+      "loss": 0.9089,
+      "step": 27180
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 7.662302949840734e-06,
+      "loss": 0.4724,
+      "step": 27190
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 7.657068149125177e-06,
+      "loss": 0.752,
+      "step": 27200
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 7.651833271856514e-06,
+      "loss": 0.6876,
+      "step": 27210
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 7.64659832058616e-06,
+      "loss": 0.8171,
+      "step": 27220
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 7.641363297865563e-06,
+      "loss": 0.5231,
+      "step": 27230
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 7.636128206246212e-06,
+      "loss": 0.6129,
+      "step": 27240
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 7.630893048279627e-06,
+      "loss": 0.498,
+      "step": 27250
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 7.625657826517362e-06,
+      "loss": 0.5614,
+      "step": 27260
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 7.620422543510997e-06,
+      "loss": 0.4612,
+      "step": 27270
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 7.615187201812148e-06,
+      "loss": 0.3073,
+      "step": 27280
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 7.609951803972455e-06,
+      "loss": 0.3593,
+      "step": 27290
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 7.604716352543591e-06,
+      "loss": 0.4811,
+      "step": 27300
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 7.5994808500772465e-06,
+      "loss": 0.5341,
+      "step": 27310
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 7.594245299125145e-06,
+      "loss": 0.5838,
+      "step": 27320
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 7.589009702239029e-06,
+      "loss": 0.6061,
+      "step": 27330
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 7.583774061970667e-06,
+      "loss": 1.0167,
+      "step": 27340
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 7.578538380871844e-06,
+      "loss": 0.3962,
+      "step": 27350
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 7.573302661494369e-06,
+      "loss": 0.7158,
+      "step": 27360
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 7.568066906390064e-06,
+      "loss": 0.6698,
+      "step": 27370
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 7.56283111811078e-06,
+      "loss": 0.4631,
+      "step": 27380
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 7.557595299208372e-06,
+      "loss": 0.6564,
+      "step": 27390
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 7.552359452234712e-06,
+      "loss": 1.0689,
+      "step": 27400
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 7.547123579741694e-06,
+      "loss": 0.4199,
+      "step": 27410
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 7.541887684281212e-06,
+      "loss": 0.1127,
+      "step": 27420
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 7.536651768405185e-06,
+      "loss": 0.6529,
+      "step": 27430
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 7.53141583466553e-06,
+      "loss": 0.8881,
+      "step": 27440
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 7.526179885614178e-06,
+      "loss": 0.6771,
+      "step": 27450
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 7.52094392380307e-06,
+      "loss": 0.3566,
+      "step": 27460
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 7.515707951784145e-06,
+      "loss": 0.4737,
+      "step": 27470
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 7.510471972109359e-06,
+      "loss": 0.5579,
+      "step": 27480
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 7.505235987330656e-06,
+      "loss": 0.7683,
+      "step": 27490
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 7.5e-06,
+      "loss": 0.8997,
+      "step": 27500
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 7.494764012669344e-06,
+      "loss": 0.8071,
+      "step": 27510
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 7.489528027890643e-06,
+      "loss": 0.5877,
+      "step": 27520
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 7.484292048215854e-06,
+      "loss": 0.8307,
+      "step": 27530
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 7.479056076196931e-06,
+      "loss": 0.8793,
+      "step": 27540
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 7.473820114385825e-06,
+      "loss": 0.5178,
+      "step": 27550
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 7.468584165334472e-06,
+      "loss": 0.5048,
+      "step": 27560
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 7.463348231594817e-06,
+      "loss": 0.6739,
+      "step": 27570
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 7.458112315718789e-06,
+      "loss": 0.5874,
+      "step": 27580
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 7.452876420258308e-06,
+      "loss": 0.5699,
+      "step": 27590
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 7.447640547765289e-06,
+      "loss": 0.884,
+      "step": 27600
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 7.442404700791629e-06,
+      "loss": 0.4479,
+      "step": 27610
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 7.43716888188922e-06,
+      "loss": 0.4794,
+      "step": 27620
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 7.4319330936099345e-06,
+      "loss": 0.5059,
+      "step": 27630
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 7.426697338505634e-06,
+      "loss": 0.5991,
+      "step": 27640
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 7.421461619128159e-06,
+      "loss": 0.9087,
+      "step": 27650
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 7.4162259380293345e-06,
+      "loss": 0.8848,
+      "step": 27660
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 7.4109902977609716e-06,
+      "loss": 0.7439,
+      "step": 27670
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 7.405754700874855e-06,
+      "loss": 1.0823,
+      "step": 27680
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 7.400519149922755e-06,
+      "loss": 0.3812,
+      "step": 27690
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 7.395283647456411e-06,
+      "loss": 0.474,
+      "step": 27700
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 7.390048196027545e-06,
+      "loss": 0.3511,
+      "step": 27710
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 7.3848127981878525e-06,
+      "loss": 0.4437,
+      "step": 27720
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 7.379577456489006e-06,
+      "loss": 0.7786,
+      "step": 27730
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 7.374342173482641e-06,
+      "loss": 0.7242,
+      "step": 27740
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 7.369106951720374e-06,
+      "loss": 0.4872,
+      "step": 27750
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 7.36387179375379e-06,
+      "loss": 0.7691,
+      "step": 27760
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 7.35863670213444e-06,
+      "loss": 0.7271,
+      "step": 27770
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 7.353401679413842e-06,
+      "loss": 0.2824,
+      "step": 27780
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 7.348166728143486e-06,
+      "loss": 0.323,
+      "step": 27790
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 7.342931850874824e-06,
+      "loss": 0.8196,
+      "step": 27800
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 7.337697050159266e-06,
+      "loss": 0.7817,
+      "step": 27810
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 7.332462328548198e-06,
+      "loss": 0.6831,
+      "step": 27820
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 7.327227688592955e-06,
+      "loss": 0.4716,
+      "step": 27830
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 7.321993132844837e-06,
+      "loss": 0.4233,
+      "step": 27840
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 7.316758663855102e-06,
+      "loss": 0.5285,
+      "step": 27850
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 7.31152428417497e-06,
+      "loss": 0.7737,
+      "step": 27860
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 7.30628999635561e-06,
+      "loss": 0.4997,
+      "step": 27870
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 7.301055802948154e-06,
+      "loss": 0.4997,
+      "step": 27880
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 7.295821706503684e-06,
+      "loss": 0.7782,
+      "step": 27890
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 7.290587709573235e-06,
+      "loss": 0.8473,
+      "step": 27900
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 7.285353814707791e-06,
+      "loss": 0.7146,
+      "step": 27910
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 7.280120024458289e-06,
+      "loss": 0.7538,
+      "step": 27920
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 7.274886341375616e-06,
+      "loss": 0.5318,
+      "step": 27930
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 7.269652768010605e-06,
+      "loss": 0.5479,
+      "step": 27940
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 7.264419306914038e-06,
+      "loss": 0.5562,
+      "step": 27950
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 7.259185960636639e-06,
+      "loss": 0.6011,
+      "step": 27960
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 7.253952731729076e-06,
+      "loss": 0.67,
+      "step": 27970
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 7.2487196227419636e-06,
+      "loss": 0.8015,
+      "step": 27980
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 7.243486636225856e-06,
+      "loss": 0.528,
+      "step": 27990
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 7.238253774731245e-06,
+      "loss": 0.7398,
+      "step": 28000
+    },
+    {
+      "epoch": 13.04,
+      "eval_accuracy": 0.8568421052631578,
+      "eval_f1": 0.8568421052631578,
+      "eval_loss": 0.7603176832199097,
+      "eval_runtime": 752.2156,
+      "eval_samples_per_second": 6.315,
+      "eval_steps_per_second": 1.579,
+      "step": 28000
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 7.233021040808562e-06,
+      "loss": 0.6519,
+      "step": 28010
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 7.227788437008182e-06,
+      "loss": 0.4855,
+      "step": 28020
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 7.2225559658804115e-06,
+      "loss": 0.547,
+      "step": 28030
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 7.21732362997549e-06,
+      "loss": 0.4211,
+      "step": 28040
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 7.212091431843601e-06,
+      "loss": 0.6844,
+      "step": 28050
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 7.206859374034849e-06,
+      "loss": 0.6981,
+      "step": 28060
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 7.201627459099275e-06,
+      "loss": 0.3789,
+      "step": 28070
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 7.196395689586858e-06,
+      "loss": 0.418,
+      "step": 28080
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 7.19116406804749e-06,
+      "loss": 0.8616,
+      "step": 28090
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 7.1859325970310044e-06,
+      "loss": 0.4312,
+      "step": 28100
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 7.1807012790871536e-06,
+      "loss": 0.3848,
+      "step": 28110
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 7.175470116765623e-06,
+      "loss": 0.4381,
+      "step": 28120
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 7.170239112616015e-06,
+      "loss": 0.7408,
+      "step": 28130
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 7.165008269187855e-06,
+      "loss": 0.5329,
+      "step": 28140
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 7.159777589030597e-06,
+      "loss": 0.3478,
+      "step": 28150
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 7.1545470746936075e-06,
+      "loss": 0.6114,
+      "step": 28160
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 7.149316728726182e-06,
+      "loss": 0.4138,
+      "step": 28170
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 7.144086553677518e-06,
+      "loss": 0.6431,
+      "step": 28180
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 7.138856552096746e-06,
+      "loss": 0.7836,
+      "step": 28190
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 7.1336267265329e-06,
+      "loss": 0.71,
+      "step": 28200
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 7.128397079534941e-06,
+      "loss": 0.7157,
+      "step": 28210
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 7.123167613651729e-06,
+      "loss": 0.2906,
+      "step": 28220
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 7.117938331432043e-06,
+      "loss": 0.548,
+      "step": 28230
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 7.112709235424576e-06,
+      "loss": 0.3176,
+      "step": 28240
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 7.107480328177922e-06,
+      "loss": 0.5765,
+      "step": 28250
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 7.102251612240589e-06,
+      "loss": 0.9896,
+      "step": 28260
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 7.097023090160984e-06,
+      "loss": 0.5716,
+      "step": 28270
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 7.091794764487433e-06,
+      "loss": 0.7398,
+      "step": 28280
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 7.086566637768154e-06,
+      "loss": 0.6816,
+      "step": 28290
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 7.081338712551271e-06,
+      "loss": 0.5037,
+      "step": 28300
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 7.076110991384817e-06,
+      "loss": 0.8894,
+      "step": 28310
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 7.070883476816714e-06,
+      "loss": 0.6717,
+      "step": 28320
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 7.065656171394791e-06,
+      "loss": 0.4743,
+      "step": 28330
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 7.0604290776667756e-06,
+      "loss": 0.5505,
+      "step": 28340
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 7.055202198180289e-06,
+      "loss": 0.6482,
+      "step": 28350
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 7.049975535482847e-06,
+      "loss": 0.5837,
+      "step": 28360
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 7.044749092121859e-06,
+      "loss": 0.8117,
+      "step": 28370
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 7.039522870644635e-06,
+      "loss": 0.4058,
+      "step": 28380
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 7.03429687359837e-06,
+      "loss": 0.4443,
+      "step": 28390
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 7.02907110353015e-06,
+      "loss": 0.5815,
+      "step": 28400
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 7.023845562986955e-06,
+      "loss": 0.3337,
+      "step": 28410
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 7.018620254515645e-06,
+      "loss": 0.5415,
+      "step": 28420
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 7.013395180662979e-06,
+      "loss": 0.8324,
+      "step": 28430
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 7.00817034397559e-06,
+      "loss": 0.4286,
+      "step": 28440
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 7.002945747000001e-06,
+      "loss": 0.5507,
+      "step": 28450
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 6.997721392282614e-06,
+      "loss": 0.6761,
+      "step": 28460
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 6.99249728236972e-06,
+      "loss": 0.8651,
+      "step": 28470
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 6.987273419807486e-06,
+      "loss": 0.9459,
+      "step": 28480
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 6.982049807141956e-06,
+      "loss": 0.8787,
+      "step": 28490
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 6.976826446919061e-06,
+      "loss": 0.759,
+      "step": 28500
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 6.971603341684598e-06,
+      "loss": 0.3774,
+      "step": 28510
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 6.966380493984251e-06,
+      "loss": 0.4496,
+      "step": 28520
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 6.961157906363564e-06,
+      "loss": 0.5926,
+      "step": 28530
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 6.95593558136797e-06,
+      "loss": 0.231,
+      "step": 28540
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 6.950713521542764e-06,
+      "loss": 0.4263,
+      "step": 28550
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 6.945491729433113e-06,
+      "loss": 0.6419,
+      "step": 28560
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 6.940270207584059e-06,
+      "loss": 0.5247,
+      "step": 28570
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 6.935048958540506e-06,
+      "loss": 0.5439,
+      "step": 28580
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 6.929827984847225e-06,
+      "loss": 0.4805,
+      "step": 28590
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 6.9246072890488605e-06,
+      "loss": 0.3178,
+      "step": 28600
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 6.919386873689914e-06,
+      "loss": 0.6941,
+      "step": 28610
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 6.91416674131475e-06,
+      "loss": 0.5058,
+      "step": 28620
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 6.9089468944675996e-06,
+      "loss": 0.4293,
+      "step": 28630
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 6.903727335692553e-06,
+      "loss": 0.3477,
+      "step": 28640
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 6.8985080675335594e-06,
+      "loss": 0.7547,
+      "step": 28650
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 6.893289092534425e-06,
+      "loss": 0.43,
+      "step": 28660
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 6.888070413238819e-06,
+      "loss": 0.5564,
+      "step": 28670
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 6.882852032190257e-06,
+      "loss": 0.9997,
+      "step": 28680
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 6.87763395193212e-06,
+      "loss": 0.6674,
+      "step": 28690
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 6.8724161750076355e-06,
+      "loss": 0.485,
+      "step": 28700
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 6.867198703959881e-06,
+      "loss": 0.7944,
+      "step": 28710
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 6.86198154133179e-06,
+      "loss": 0.4259,
+      "step": 28720
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 6.856764689666146e-06,
+      "loss": 0.8059,
+      "step": 28730
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 6.85154815150558e-06,
+      "loss": 0.5936,
+      "step": 28740
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 6.8463319293925634e-06,
+      "loss": 0.301,
+      "step": 28750
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 6.841116025869426e-06,
+      "loss": 0.5836,
+      "step": 28760
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 6.835900443478333e-06,
+      "loss": 0.673,
+      "step": 28770
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 6.830685184761296e-06,
+      "loss": 0.3068,
+      "step": 28780
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 6.8254702522601715e-06,
+      "loss": 0.8024,
+      "step": 28790
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 6.82025564851665e-06,
+      "loss": 0.5062,
+      "step": 28800
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 6.815041376072268e-06,
+      "loss": 0.5104,
+      "step": 28810
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 6.809827437468397e-06,
+      "loss": 0.5171,
+      "step": 28820
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 6.80461383524625e-06,
+      "loss": 0.6172,
+      "step": 28830
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 6.799400571946872e-06,
+      "loss": 0.8842,
+      "step": 28840
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 6.7941876501111426e-06,
+      "loss": 0.7714,
+      "step": 28850
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 6.78897507227978e-06,
+      "loss": 0.82,
+      "step": 28860
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 6.7837628409933274e-06,
+      "loss": 0.9166,
+      "step": 28870
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 6.778550958792171e-06,
+      "loss": 0.4862,
+      "step": 28880
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 6.773339428216507e-06,
+      "loss": 0.5772,
+      "step": 28890
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 6.76812825180638e-06,
+      "loss": 0.3489,
+      "step": 28900
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 6.762917432101653e-06,
+      "loss": 0.7883,
+      "step": 28910
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 6.7577069716420125e-06,
+      "loss": 0.6087,
+      "step": 28920
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 6.752496872966979e-06,
+      "loss": 0.588,
+      "step": 28930
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 6.747287138615887e-06,
+      "loss": 0.5674,
+      "step": 28940
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 6.742077771127902e-06,
+      "loss": 0.3757,
+      "step": 28950
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 6.7368687730420035e-06,
+      "loss": 0.4141,
+      "step": 28960
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 6.731660146896996e-06,
+      "loss": 0.7038,
+      "step": 28970
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 6.726451895231497e-06,
+      "loss": 0.6075,
+      "step": 28980
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 6.721244020583946e-06,
+      "loss": 0.6138,
+      "step": 28990
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 6.7160365254926005e-06,
+      "loss": 0.6848,
+      "step": 29000
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 6.710829412495527e-06,
+      "loss": 0.5487,
+      "step": 29010
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 6.705622684130612e-06,
+      "loss": 0.6123,
+      "step": 29020
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 6.700416342935551e-06,
+      "loss": 0.572,
+      "step": 29030
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 6.69521039144785e-06,
+      "loss": 1.101,
+      "step": 29040
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 6.690004832204828e-06,
+      "loss": 0.5468,
+      "step": 29050
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 6.684799667743613e-06,
+      "loss": 0.4592,
+      "step": 29060
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 6.679594900601136e-06,
+      "loss": 0.4517,
+      "step": 29070
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 6.674390533314135e-06,
+      "loss": 0.3175,
+      "step": 29080
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 6.66918656841916e-06,
+      "loss": 0.5221,
+      "step": 29090
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 6.66398300845256e-06,
+      "loss": 0.6589,
+      "step": 29100
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 6.658779855950483e-06,
+      "loss": 0.7721,
+      "step": 29110
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 6.653577113448887e-06,
+      "loss": 0.5447,
+      "step": 29120
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 6.648374783483521e-06,
+      "loss": 0.6826,
+      "step": 29130
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 6.643172868589947e-06,
+      "loss": 0.8476,
+      "step": 29140
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 6.6379713713035015e-06,
+      "loss": 0.5372,
+      "step": 29150
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 6.63277029415934e-06,
+      "loss": 0.426,
+      "step": 29160
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 6.627569639692401e-06,
+      "loss": 0.4579,
+      "step": 29170
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 6.62236941043742e-06,
+      "loss": 0.3956,
+      "step": 29180
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 6.617169608928927e-06,
+      "loss": 0.663,
+      "step": 29190
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 6.611970237701242e-06,
+      "loss": 0.5617,
+      "step": 29200
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 6.606771299288477e-06,
+      "loss": 0.5615,
+      "step": 29210
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 6.6015727962245286e-06,
+      "loss": 0.5542,
+      "step": 29220
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 6.596374731043087e-06,
+      "loss": 0.4503,
+      "step": 29230
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 6.591177106277623e-06,
+      "loss": 0.7225,
+      "step": 29240
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 6.5859799244613955e-06,
+      "loss": 0.9153,
+      "step": 29250
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 6.580783188127449e-06,
+      "loss": 0.6467,
+      "step": 29260
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 6.57558689980861e-06,
+      "loss": 0.4762,
+      "step": 29270
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 6.570391062037487e-06,
+      "loss": 0.5648,
+      "step": 29280
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 6.5651956773464675e-06,
+      "loss": 0.6234,
+      "step": 29290
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 6.560000748267717e-06,
+      "loss": 0.9244,
+      "step": 29300
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 6.554806277333185e-06,
+      "loss": 0.3941,
+      "step": 29310
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 6.549612267074592e-06,
+      "loss": 0.5927,
+      "step": 29320
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 6.544418720023433e-06,
+      "loss": 0.5772,
+      "step": 29330
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 6.53922563871098e-06,
+      "loss": 0.4929,
+      "step": 29340
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 6.5340330256682805e-06,
+      "loss": 0.853,
+      "step": 29350
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 6.5288408834261485e-06,
+      "loss": 0.516,
+      "step": 29360
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 6.523649214515172e-06,
+      "loss": 0.5695,
+      "step": 29370
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 6.5184580214657085e-06,
+      "loss": 0.3876,
+      "step": 29380
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 6.513267306807882e-06,
+      "loss": 0.6751,
+      "step": 29390
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 6.508077073071581e-06,
+      "loss": 0.6608,
+      "step": 29400
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 6.502887322786468e-06,
+      "loss": 0.5213,
+      "step": 29410
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 6.497698058481959e-06,
+      "loss": 0.9044,
+      "step": 29420
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 6.49250928268724e-06,
+      "loss": 0.7779,
+      "step": 29430
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 6.487320997931256e-06,
+      "loss": 0.8036,
+      "step": 29440
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 6.482133206742718e-06,
+      "loss": 0.9571,
+      "step": 29450
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 6.476945911650088e-06,
+      "loss": 0.5353,
+      "step": 29460
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 6.471759115181593e-06,
+      "loss": 0.3019,
+      "step": 29470
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 6.466572819865215e-06,
+      "loss": 0.4097,
+      "step": 29480
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 6.461387028228688e-06,
+      "loss": 0.7054,
+      "step": 29490
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 6.456201742799511e-06,
+      "loss": 0.9272,
+      "step": 29500
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 6.451016966104921e-06,
+      "loss": 0.7606,
+      "step": 29510
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 6.445832700671919e-06,
+      "loss": 0.4147,
+      "step": 29520
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 6.440648949027251e-06,
+      "loss": 0.4601,
+      "step": 29530
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 6.435465713697417e-06,
+      "loss": 0.4947,
+      "step": 29540
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 6.430282997208663e-06,
+      "loss": 0.4206,
+      "step": 29550
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 6.425100802086979e-06,
+      "loss": 0.8854,
+      "step": 29560
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 6.419919130858109e-06,
+      "loss": 0.4316,
+      "step": 29570
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 6.414737986047532e-06,
+      "loss": 0.3306,
+      "step": 29580
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 6.409557370180479e-06,
+      "loss": 1.0803,
+      "step": 29590
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 6.404377285781914e-06,
+      "loss": 0.5344,
+      "step": 29600
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 6.399197735376551e-06,
+      "loss": 0.5595,
+      "step": 29610
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 6.394018721488839e-06,
+      "loss": 0.9271,
+      "step": 29620
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 6.388840246642965e-06,
+      "loss": 0.328,
+      "step": 29630
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 6.383662313362858e-06,
+      "loss": 0.6804,
+      "step": 29640
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 6.37848492417218e-06,
+      "loss": 0.231,
+      "step": 29650
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 6.373308081594322e-06,
+      "loss": 0.6374,
+      "step": 29660
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 6.368131788152419e-06,
+      "loss": 0.6959,
+      "step": 29670
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 6.362956046369335e-06,
+      "loss": 0.346,
+      "step": 29680
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 6.3577808587676585e-06,
+      "loss": 0.5246,
+      "step": 29690
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 6.3526062278697154e-06,
+      "loss": 0.8182,
+      "step": 29700
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 6.347432156197558e-06,
+      "loss": 0.7254,
+      "step": 29710
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 6.342258646272966e-06,
+      "loss": 0.6114,
+      "step": 29720
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 6.3370857006174464e-06,
+      "loss": 0.526,
+      "step": 29730
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 6.3319133217522295e-06,
+      "loss": 0.4828,
+      "step": 29740
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 6.326741512198267e-06,
+      "loss": 0.5321,
+      "step": 29750
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 6.321570274476245e-06,
+      "loss": 0.7288,
+      "step": 29760
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 6.316399611106549e-06,
+      "loss": 0.5285,
+      "step": 29770
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 6.311229524609307e-06,
+      "loss": 0.656,
+      "step": 29780
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 6.3060600175043494e-06,
+      "loss": 0.8339,
+      "step": 29790
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 6.300891092311239e-06,
+      "loss": 0.506,
+      "step": 29800
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 6.295722751549243e-06,
+      "loss": 0.5651,
+      "step": 29810
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 6.290554997737346e-06,
+      "loss": 0.4679,
+      "step": 29820
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 6.2853878333942526e-06,
+      "loss": 0.7864,
+      "step": 29830
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 6.2802212610383735e-06,
+      "loss": 0.5258,
+      "step": 29840
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 6.2750552831878354e-06,
+      "loss": 0.7134,
+      "step": 29850
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 6.269889902360468e-06,
+      "loss": 0.7398,
+      "step": 29860
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 6.26472512107382e-06,
+      "loss": 0.439,
+      "step": 29870
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 6.259560941845143e-06,
+      "loss": 0.5584,
+      "step": 29880
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 6.254397367191391e-06,
+      "loss": 0.6233,
+      "step": 29890
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 6.249234399629234e-06,
+      "loss": 0.702,
+      "step": 29900
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 6.244072041675034e-06,
+      "loss": 0.5829,
+      "step": 29910
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 6.238910295844863e-06,
+      "loss": 0.3531,
+      "step": 29920
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 6.233749164654496e-06,
+      "loss": 0.6007,
+      "step": 29930
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 6.228588650619405e-06,
+      "loss": 0.4124,
+      "step": 29940
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 6.223428756254758e-06,
+      "loss": 0.5734,
+      "step": 29950
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 6.218269484075426e-06,
+      "loss": 0.6953,
+      "step": 29960
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 6.213110836595978e-06,
+      "loss": 0.716,
+      "step": 29970
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 6.207952816330676e-06,
+      "loss": 0.708,
+      "step": 29980
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 6.202795425793473e-06,
+      "loss": 0.6968,
+      "step": 29990
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 6.197638667498023e-06,
+      "loss": 0.6807,
+      "step": 30000
+    },
+    {
+      "epoch": 14.04,
+      "eval_accuracy": 0.871578947368421,
+      "eval_f1": 0.871578947368421,
+      "eval_loss": 0.7531183362007141,
+      "eval_runtime": 743.1398,
+      "eval_samples_per_second": 6.392,
+      "eval_steps_per_second": 1.599,
+      "step": 30000
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 6.1924825439576625e-06,
+      "loss": 0.3591,
+      "step": 30010
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 6.1873270576854295e-06,
+      "loss": 0.8104,
+      "step": 30020
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 6.182172211194042e-06,
+      "loss": 0.6743,
+      "step": 30030
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 6.177018006995909e-06,
+      "loss": 0.3849,
+      "step": 30040
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 6.171864447603126e-06,
+      "loss": 0.4138,
+      "step": 30050
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 6.1667115355274785e-06,
+      "loss": 0.252,
+      "step": 30060
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 6.161559273280431e-06,
+      "loss": 0.5183,
+      "step": 30070
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 6.156407663373133e-06,
+      "loss": 0.5016,
+      "step": 30080
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 6.1512567083164184e-06,
+      "loss": 0.7197,
+      "step": 30090
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 6.146106410620801e-06,
+      "loss": 0.5565,
+      "step": 30100
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 6.140956772796469e-06,
+      "loss": 0.4194,
+      "step": 30110
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 6.1358077973533e-06,
+      "loss": 0.4268,
+      "step": 30120
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 6.130659486800836e-06,
+      "loss": 0.4098,
+      "step": 30130
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 6.125511843648304e-06,
+      "loss": 0.3364,
+      "step": 30140
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 6.1203648704045986e-06,
+      "loss": 0.5037,
+      "step": 30150
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 6.115218569578299e-06,
+      "loss": 0.3156,
+      "step": 30160
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 6.110072943677645e-06,
+      "loss": 0.6378,
+      "step": 30170
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 6.104927995210551e-06,
+      "loss": 0.2939,
+      "step": 30180
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 6.099783726684608e-06,
+      "loss": 0.595,
+      "step": 30190
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 6.094640140607064e-06,
+      "loss": 0.4013,
+      "step": 30200
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 6.089497239484847e-06,
+      "loss": 0.751,
+      "step": 30210
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 6.084355025824538e-06,
+      "loss": 0.7006,
+      "step": 30220
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 6.079213502132392e-06,
+      "loss": 0.3794,
+      "step": 30230
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 6.074072670914325e-06,
+      "loss": 0.4284,
+      "step": 30240
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 6.068932534675914e-06,
+      "loss": 0.7241,
+      "step": 30250
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 6.063793095922403e-06,
+      "loss": 0.4987,
+      "step": 30260
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 6.058654357158688e-06,
+      "loss": 0.5056,
+      "step": 30270
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 6.053516320889331e-06,
+      "loss": 0.7833,
+      "step": 30280
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 6.048378989618548e-06,
+      "loss": 0.6071,
+      "step": 30290
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 6.043242365850212e-06,
+      "loss": 0.3674,
+      "step": 30300
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 6.038106452087847e-06,
+      "loss": 0.6286,
+      "step": 30310
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 6.032971250834639e-06,
+      "loss": 0.695,
+      "step": 30320
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 6.027836764593422e-06,
+      "loss": 0.4895,
+      "step": 30330
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 6.022702995866681e-06,
+      "loss": 0.42,
+      "step": 30340
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 6.017569947156554e-06,
+      "loss": 0.2868,
+      "step": 30350
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 6.012437620964827e-06,
+      "loss": 0.4577,
+      "step": 30360
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 6.0073060197929295e-06,
+      "loss": 0.662,
+      "step": 30370
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 6.0021751461419485e-06,
+      "loss": 0.6371,
+      "step": 30380
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 5.997045002512604e-06,
+      "loss": 0.3516,
+      "step": 30390
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 5.991915591405267e-06,
+      "loss": 0.5922,
+      "step": 30400
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 5.986786915319949e-06,
+      "loss": 0.7923,
+      "step": 30410
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 5.9816589767563065e-06,
+      "loss": 0.3149,
+      "step": 30420
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 5.9765317782136346e-06,
+      "loss": 0.4833,
+      "step": 30430
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 5.971405322190864e-06,
+      "loss": 0.7275,
+      "step": 30440
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 5.966279611186573e-06,
+      "loss": 0.6083,
+      "step": 30450
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 5.961154647698965e-06,
+      "loss": 0.5943,
+      "step": 30460
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 5.956030434225892e-06,
+      "loss": 0.7019,
+      "step": 30470
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 5.9509069732648255e-06,
+      "loss": 0.5745,
+      "step": 30480
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 5.945784267312882e-06,
+      "loss": 0.7729,
+      "step": 30490
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 5.9406623188668065e-06,
+      "loss": 0.5774,
+      "step": 30500
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 5.935541130422972e-06,
+      "loss": 0.468,
+      "step": 30510
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 5.930420704477387e-06,
+      "loss": 0.5499,
+      "step": 30520
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 5.925301043525682e-06,
+      "loss": 0.6378,
+      "step": 30530
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 5.9201821500631195e-06,
+      "loss": 1.0126,
+      "step": 30540
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 5.915064026584586e-06,
+      "loss": 1.0802,
+      "step": 30550
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 5.9099466755845925e-06,
+      "loss": 0.7932,
+      "step": 30560
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 5.904830099557271e-06,
+      "loss": 0.5136,
+      "step": 30570
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 5.899714300996381e-06,
+      "loss": 0.4871,
+      "step": 30580
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 5.8945992823953e-06,
+      "loss": 0.4485,
+      "step": 30590
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 5.889485046247026e-06,
+      "loss": 0.5692,
+      "step": 30600
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 5.8843715950441765e-06,
+      "loss": 0.6649,
+      "step": 30610
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 5.8792589312789855e-06,
+      "loss": 0.6113,
+      "step": 30620
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 5.874147057443301e-06,
+      "loss": 0.4845,
+      "step": 30630
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 5.869035976028589e-06,
+      "loss": 0.6766,
+      "step": 30640
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 5.863925689525933e-06,
+      "loss": 0.385,
+      "step": 30650
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 5.858816200426018e-06,
+      "loss": 0.5752,
+      "step": 30660
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 5.853707511219148e-06,
+      "loss": 0.7434,
+      "step": 30670
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 5.848599624395239e-06,
+      "loss": 0.8021,
+      "step": 30680
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 5.8434925424438125e-06,
+      "loss": 0.5919,
+      "step": 30690
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 5.838386267853996e-06,
+      "loss": 0.7574,
+      "step": 30700
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 5.833280803114527e-06,
+      "loss": 0.7047,
+      "step": 30710
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 5.828176150713747e-06,
+      "loss": 0.4341,
+      "step": 30720
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 5.823072313139602e-06,
+      "loss": 0.8295,
+      "step": 30730
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 5.817969292879642e-06,
+      "loss": 0.2953,
+      "step": 30740
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 5.812867092421014e-06,
+      "loss": 0.2567,
+      "step": 30750
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 5.807765714250469e-06,
+      "loss": 0.6689,
+      "step": 30760
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 5.802665160854357e-06,
+      "loss": 0.6787,
+      "step": 30770
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 5.797565434718626e-06,
+      "loss": 0.7291,
+      "step": 30780
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 5.79246653832882e-06,
+      "loss": 0.4689,
+      "step": 30790
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 5.787368474170081e-06,
+      "loss": 0.144,
+      "step": 30800
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 5.782271244727144e-06,
+      "loss": 0.598,
+      "step": 30810
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 5.777174852484333e-06,
+      "loss": 0.7588,
+      "step": 30820
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 5.772079299925573e-06,
+      "loss": 0.4024,
+      "step": 30830
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 5.766984589534368e-06,
+      "loss": 0.6319,
+      "step": 30840
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 5.761890723793821e-06,
+      "loss": 0.5051,
+      "step": 30850
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 5.75679770518662e-06,
+      "loss": 0.428,
+      "step": 30860
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 5.751705536195043e-06,
+      "loss": 0.6227,
+      "step": 30870
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 5.7466142193009485e-06,
+      "loss": 0.5371,
+      "step": 30880
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 5.7415237569857805e-06,
+      "loss": 0.5355,
+      "step": 30890
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 5.736434151730575e-06,
+      "loss": 0.5317,
+      "step": 30900
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 5.731345406015938e-06,
+      "loss": 0.7229,
+      "step": 30910
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 5.726257522322067e-06,
+      "loss": 0.5791,
+      "step": 30920
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 5.7211705031287285e-06,
+      "loss": 0.8684,
+      "step": 30930
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 5.716084350915279e-06,
+      "loss": 0.2345,
+      "step": 30940
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 5.710999068160645e-06,
+      "loss": 0.4998,
+      "step": 30950
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 5.70591465734333e-06,
+      "loss": 0.599,
+      "step": 30960
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 5.7008311209414195e-06,
+      "loss": 0.5463,
+      "step": 30970
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 5.695748461432562e-06,
+      "loss": 0.4036,
+      "step": 30980
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 5.690666681293987e-06,
+      "loss": 0.393,
+      "step": 30990
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 5.685585783002493e-06,
+      "loss": 0.4896,
+      "step": 31000
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 5.680505769034446e-06,
+      "loss": 0.4201,
+      "step": 31010
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 5.675426641865783e-06,
+      "loss": 0.8405,
+      "step": 31020
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 5.6703484039720085e-06,
+      "loss": 0.5632,
+      "step": 31030
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 5.665271057828196e-06,
+      "loss": 0.3888,
+      "step": 31040
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 5.66019460590898e-06,
+      "loss": 0.398,
+      "step": 31050
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 5.655119050688565e-06,
+      "loss": 0.6701,
+      "step": 31060
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 5.650044394640713e-06,
+      "loss": 0.5749,
+      "step": 31070
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 5.644970640238747e-06,
+      "loss": 0.5094,
+      "step": 31080
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 5.63989778995556e-06,
+      "loss": 0.5268,
+      "step": 31090
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 5.6348258462635905e-06,
+      "loss": 0.4821,
+      "step": 31100
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 5.629754811634846e-06,
+      "loss": 0.5058,
+      "step": 31110
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 5.624684688540884e-06,
+      "loss": 0.2582,
+      "step": 31120
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 5.619615479452826e-06,
+      "loss": 0.7632,
+      "step": 31130
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 5.614547186841339e-06,
+      "loss": 0.6387,
+      "step": 31140
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 5.609479813176648e-06,
+      "loss": 0.4985,
+      "step": 31150
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 5.6044133609285305e-06,
+      "loss": 0.454,
+      "step": 31160
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 5.5993478325663125e-06,
+      "loss": 0.7597,
+      "step": 31170
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 5.594283230558874e-06,
+      "loss": 0.3568,
+      "step": 31180
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 5.589219557374635e-06,
+      "loss": 0.3692,
+      "step": 31190
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 5.5841568154815725e-06,
+      "loss": 0.6227,
+      "step": 31200
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 5.579095007347204e-06,
+      "loss": 0.3188,
+      "step": 31210
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 5.574034135438591e-06,
+      "loss": 0.5097,
+      "step": 31220
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 5.568974202222348e-06,
+      "loss": 0.7134,
+      "step": 31230
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 5.5639152101646184e-06,
+      "loss": 0.6648,
+      "step": 31240
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 5.558857161731094e-06,
+      "loss": 0.9005,
+      "step": 31250
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 5.55380005938701e-06,
+      "loss": 0.6035,
+      "step": 31260
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 5.548743905597135e-06,
+      "loss": 0.3174,
+      "step": 31270
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 5.543688702825775e-06,
+      "loss": 0.7496,
+      "step": 31280
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 5.538634453536774e-06,
+      "loss": 0.6956,
+      "step": 31290
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 5.533581160193516e-06,
+      "loss": 0.6995,
+      "step": 31300
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 5.5285288252589095e-06,
+      "loss": 0.4295,
+      "step": 31310
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 5.5234774511954065e-06,
+      "loss": 0.5449,
+      "step": 31320
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 5.5184270404649815e-06,
+      "loss": 0.7376,
+      "step": 31330
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 5.513377595529143e-06,
+      "loss": 0.5837,
+      "step": 31340
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 5.508329118848934e-06,
+      "loss": 1.0156,
+      "step": 31350
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 5.503281612884917e-06,
+      "loss": 0.6391,
+      "step": 31360
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 5.498235080097185e-06,
+      "loss": 0.4268,
+      "step": 31370
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 5.493189522945356e-06,
+      "loss": 0.4658,
+      "step": 31380
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 5.488144943888575e-06,
+      "loss": 0.7193,
+      "step": 31390
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 5.483101345385508e-06,
+      "loss": 1.1998,
+      "step": 31400
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 5.478058729894342e-06,
+      "loss": 0.3759,
+      "step": 31410
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 5.4730170998727915e-06,
+      "loss": 1.0222,
+      "step": 31420
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 5.4679764577780815e-06,
+      "loss": 0.6979,
+      "step": 31430
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 5.4629368060669585e-06,
+      "loss": 0.6705,
+      "step": 31440
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 5.457898147195693e-06,
+      "loss": 0.2831,
+      "step": 31450
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 5.452860483620062e-06,
+      "loss": 0.5871,
+      "step": 31460
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 5.447823817795362e-06,
+      "loss": 0.6708,
+      "step": 31470
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 5.442788152176401e-06,
+      "loss": 0.3945,
+      "step": 31480
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 5.437753489217505e-06,
+      "loss": 1.0316,
+      "step": 31490
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 5.432719831372507e-06,
+      "loss": 0.4664,
+      "step": 31500
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 5.427687181094748e-06,
+      "loss": 0.4658,
+      "step": 31510
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 5.422655540837083e-06,
+      "loss": 0.4841,
+      "step": 31520
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 5.417624913051869e-06,
+      "loss": 0.4642,
+      "step": 31530
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 5.412595300190981e-06,
+      "loss": 0.5075,
+      "step": 31540
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 5.407566704705782e-06,
+      "loss": 0.7411,
+      "step": 31550
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 5.402539129047152e-06,
+      "loss": 1.0875,
+      "step": 31560
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 5.397512575665469e-06,
+      "loss": 0.7321,
+      "step": 31570
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 5.392487047010618e-06,
+      "loss": 0.537,
+      "step": 31580
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 5.387462545531976e-06,
+      "loss": 0.4299,
+      "step": 31590
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 5.382439073678426e-06,
+      "loss": 0.5034,
+      "step": 31600
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 5.377416633898348e-06,
+      "loss": 0.7486,
+      "step": 31610
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 5.372395228639619e-06,
+      "loss": 0.6522,
+      "step": 31620
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 5.367374860349607e-06,
+      "loss": 0.5415,
+      "step": 31630
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 5.362355531475179e-06,
+      "loss": 0.592,
+      "step": 31640
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 5.357337244462699e-06,
+      "loss": 0.2441,
+      "step": 31650
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 5.352320001758017e-06,
+      "loss": 0.6928,
+      "step": 31660
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 5.3473038058064725e-06,
+      "loss": 0.3521,
+      "step": 31670
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 5.342288659052905e-06,
+      "loss": 0.5533,
+      "step": 31680
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 5.337274563941633e-06,
+      "loss": 0.6759,
+      "step": 31690
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 5.332261522916461e-06,
+      "loss": 0.4252,
+      "step": 31700
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 5.3272495384206945e-06,
+      "loss": 0.8168,
+      "step": 31710
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 5.322238612897107e-06,
+      "loss": 0.619,
+      "step": 31720
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 5.317228748787963e-06,
+      "loss": 0.8278,
+      "step": 31730
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 5.3122199485350085e-06,
+      "loss": 0.6556,
+      "step": 31740
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 5.307212214579475e-06,
+      "loss": 0.3492,
+      "step": 31750
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 5.3022055493620705e-06,
+      "loss": 0.5529,
+      "step": 31760
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 5.29719995532298e-06,
+      "loss": 0.5318,
+      "step": 31770
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 5.292195434901873e-06,
+      "loss": 0.6974,
+      "step": 31780
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 5.2871919905378896e-06,
+      "loss": 0.4782,
+      "step": 31790
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 5.282189624669653e-06,
+      "loss": 0.6933,
+      "step": 31800
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 5.277188339735244e-06,
+      "loss": 0.5515,
+      "step": 31810
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 5.272188138172239e-06,
+      "loss": 0.4032,
+      "step": 31820
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 5.267189022417672e-06,
+      "loss": 0.7064,
+      "step": 31830
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 5.2621909949080485e-06,
+      "loss": 0.814,
+      "step": 31840
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 5.2571940580793506e-06,
+      "loss": 0.5002,
+      "step": 31850
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 5.252198214367021e-06,
+      "loss": 0.6971,
+      "step": 31860
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 5.247203466205977e-06,
+      "loss": 0.6614,
+      "step": 31870
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 5.242209816030597e-06,
+      "loss": 0.4285,
+      "step": 31880
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 5.237217266274726e-06,
+      "loss": 0.4315,
+      "step": 31890
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 5.232225819371667e-06,
+      "loss": 0.4179,
+      "step": 31900
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 5.227235477754197e-06,
+      "loss": 0.5745,
+      "step": 31910
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 5.222246243854544e-06,
+      "loss": 0.2488,
+      "step": 31920
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 5.217258120104401e-06,
+      "loss": 0.5487,
+      "step": 31930
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 5.212271108934919e-06,
+      "loss": 0.3712,
+      "step": 31940
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 5.207285212776708e-06,
+      "loss": 0.6056,
+      "step": 31950
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 5.20230043405983e-06,
+      "loss": 0.3832,
+      "step": 31960
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 5.19731677521381e-06,
+      "loss": 0.4765,
+      "step": 31970
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 5.192334238667622e-06,
+      "loss": 0.6145,
+      "step": 31980
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 5.18735282684969e-06,
+      "loss": 0.3832,
+      "step": 31990
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 5.182372542187895e-06,
+      "loss": 0.3262,
+      "step": 32000
+    },
+    {
+      "epoch": 15.04,
+      "eval_accuracy": 0.8768421052631579,
+      "eval_f1": 0.876842105263158,
+      "eval_loss": 0.7663307785987854,
+      "eval_runtime": 750.321,
+      "eval_samples_per_second": 6.331,
+      "eval_steps_per_second": 1.583,
+      "step": 32000
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 5.177393387109569e-06,
+      "loss": 0.4934,
+      "step": 32010
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 5.172415364041492e-06,
+      "loss": 0.7969,
+      "step": 32020
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 5.167438475409888e-06,
+      "loss": 0.6713,
+      "step": 32030
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 5.162462723640436e-06,
+      "loss": 0.4962,
+      "step": 32040
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 5.157488111158255e-06,
+      "loss": 0.465,
+      "step": 32050
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 5.152514640387912e-06,
+      "loss": 0.4099,
+      "step": 32060
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 5.147542313753419e-06,
+      "loss": 0.228,
+      "step": 32070
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 5.142571133678222e-06,
+      "loss": 0.85,
+      "step": 32080
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 5.137601102585216e-06,
+      "loss": 0.5891,
+      "step": 32090
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 5.132632222896733e-06,
+      "loss": 0.3639,
+      "step": 32100
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 5.127664497034546e-06,
+      "loss": 0.5626,
+      "step": 32110
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 5.122697927419861e-06,
+      "loss": 0.7871,
+      "step": 32120
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 5.1177325164733296e-06,
+      "loss": 0.3236,
+      "step": 32130
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 5.11276826661503e-06,
+      "loss": 0.6567,
+      "step": 32140
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 5.107805180264475e-06,
+      "loss": 0.5179,
+      "step": 32150
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 5.102843259840616e-06,
+      "loss": 0.6067,
+      "step": 32160
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 5.097882507761832e-06,
+      "loss": 0.3634,
+      "step": 32170
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 5.09292292644593e-06,
+      "loss": 0.6364,
+      "step": 32180
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 5.087964518310153e-06,
+      "loss": 0.9535,
+      "step": 32190
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 5.083007285771168e-06,
+      "loss": 0.712,
+      "step": 32200
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 5.07805123124507e-06,
+      "loss": 0.3794,
+      "step": 32210
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 5.073096357147379e-06,
+      "loss": 0.4944,
+      "step": 32220
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 5.068142665893042e-06,
+      "loss": 1.1231,
+      "step": 32230
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 5.063190159896428e-06,
+      "loss": 0.619,
+      "step": 32240
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 5.058238841571326e-06,
+      "loss": 0.6341,
+      "step": 32250
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 5.05328871333095e-06,
+      "loss": 0.6589,
+      "step": 32260
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 5.0483397775879325e-06,
+      "loss": 0.5169,
+      "step": 32270
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 5.043392036754324e-06,
+      "loss": 0.4261,
+      "step": 32280
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 5.038445493241593e-06,
+      "loss": 0.4723,
+      "step": 32290
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 5.033500149460626e-06,
+      "loss": 0.653,
+      "step": 32300
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 5.028556007821722e-06,
+      "loss": 0.535,
+      "step": 32310
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 5.023613070734597e-06,
+      "loss": 0.706,
+      "step": 32320
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 5.01867134060838e-06,
+      "loss": 0.9763,
+      "step": 32330
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 5.013730819851606e-06,
+      "loss": 0.6044,
+      "step": 32340
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 5.008791510872227e-06,
+      "loss": 0.6842,
+      "step": 32350
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 5.003853416077601e-06,
+      "loss": 0.4799,
+      "step": 32360
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 4.998916537874498e-06,
+      "loss": 0.7376,
+      "step": 32370
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 4.9939808786690885e-06,
+      "loss": 0.3854,
+      "step": 32380
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 4.989046440866958e-06,
+      "loss": 0.8238,
+      "step": 32390
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 4.984113226873088e-06,
+      "loss": 0.5112,
+      "step": 32400
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 4.979181239091867e-06,
+      "loss": 0.4544,
+      "step": 32410
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 4.97425047992709e-06,
+      "loss": 0.5501,
+      "step": 32420
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 4.969320951781942e-06,
+      "loss": 0.4778,
+      "step": 32430
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 4.964392657059022e-06,
+      "loss": 0.7256,
+      "step": 32440
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 4.959465598160315e-06,
+      "loss": 0.5249,
+      "step": 32450
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 4.9545397774872146e-06,
+      "loss": 0.3695,
+      "step": 32460
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 4.9496151974405045e-06,
+      "loss": 0.6224,
+      "step": 32470
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 4.944691860420363e-06,
+      "loss": 0.5238,
+      "step": 32480
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 4.939769768826367e-06,
+      "loss": 0.5127,
+      "step": 32490
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 4.934848925057485e-06,
+      "loss": 0.8212,
+      "step": 32500
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 4.929929331512076e-06,
+      "loss": 0.4451,
+      "step": 32510
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 4.925010990587886e-06,
+      "loss": 0.472,
+      "step": 32520
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 4.92009390468206e-06,
+      "loss": 0.9875,
+      "step": 32530
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 4.915178076191123e-06,
+      "loss": 0.4381,
+      "step": 32540
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 4.9102635075109905e-06,
+      "loss": 0.4436,
+      "step": 32550
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 4.905350201036965e-06,
+      "loss": 0.3102,
+      "step": 32560
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 4.9004381591637295e-06,
+      "loss": 0.7939,
+      "step": 32570
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 4.895527384285357e-06,
+      "loss": 0.8265,
+      "step": 32580
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 4.890617878795299e-06,
+      "loss": 0.4009,
+      "step": 32590
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 4.885709645086388e-06,
+      "loss": 0.4055,
+      "step": 32600
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 4.880802685550836e-06,
+      "loss": 1.0053,
+      "step": 32610
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 4.875897002580235e-06,
+      "loss": 0.2254,
+      "step": 32620
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 4.870992598565559e-06,
+      "loss": 0.3148,
+      "step": 32630
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 4.86608947589715e-06,
+      "loss": 0.5144,
+      "step": 32640
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 4.861187636964736e-06,
+      "loss": 0.5433,
+      "step": 32650
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 4.8562870841574115e-06,
+      "loss": 0.4401,
+      "step": 32660
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 4.851387819863644e-06,
+      "loss": 0.6315,
+      "step": 32670
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 4.846489846471279e-06,
+      "loss": 0.5919,
+      "step": 32680
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 4.841593166367529e-06,
+      "loss": 1.168,
+      "step": 32690
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 4.836697781938973e-06,
+      "loss": 0.5355,
+      "step": 32700
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 4.831803695571563e-06,
+      "loss": 0.3735,
+      "step": 32710
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 4.82691090965062e-06,
+      "loss": 0.4671,
+      "step": 32720
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 4.822019426560827e-06,
+      "loss": 0.7564,
+      "step": 32730
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 4.817129248686231e-06,
+      "loss": 0.3753,
+      "step": 32740
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 4.812240378410248e-06,
+      "loss": 0.5225,
+      "step": 32750
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 4.8073528181156535e-06,
+      "loss": 0.78,
+      "step": 32760
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 4.802466570184584e-06,
+      "loss": 0.6062,
+      "step": 32770
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 4.797581636998541e-06,
+      "loss": 0.6152,
+      "step": 32780
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 4.792698020938377e-06,
+      "loss": 0.2874,
+      "step": 32790
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 4.787815724384309e-06,
+      "loss": 0.5955,
+      "step": 32800
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 4.782934749715908e-06,
+      "loss": 0.654,
+      "step": 32810
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 4.778055099312103e-06,
+      "loss": 0.3857,
+      "step": 32820
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 4.7731767755511775e-06,
+      "loss": 0.5285,
+      "step": 32830
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 4.768299780810763e-06,
+      "loss": 0.7611,
+      "step": 32840
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 4.763424117467852e-06,
+      "loss": 0.759,
+      "step": 32850
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 4.758549787898785e-06,
+      "loss": 0.6901,
+      "step": 32860
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 4.753676794479244e-06,
+      "loss": 0.8388,
+      "step": 32870
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 4.74880513958427e-06,
+      "loss": 0.3117,
+      "step": 32880
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 4.743934825588252e-06,
+      "loss": 0.4733,
+      "step": 32890
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 4.739065854864916e-06,
+      "loss": 0.297,
+      "step": 32900
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 4.734198229787343e-06,
+      "loss": 0.4909,
+      "step": 32910
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 4.729331952727954e-06,
+      "loss": 0.4957,
+      "step": 32920
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 4.724467026058512e-06,
+      "loss": 0.3845,
+      "step": 32930
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 4.719603452150125e-06,
+      "loss": 0.3681,
+      "step": 32940
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 4.714741233373241e-06,
+      "loss": 0.5048,
+      "step": 32950
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 4.709880372097642e-06,
+      "loss": 0.5012,
+      "step": 32960
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 4.705020870692452e-06,
+      "loss": 0.5737,
+      "step": 32970
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 4.7001627315261395e-06,
+      "loss": 0.7233,
+      "step": 32980
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 4.695305956966498e-06,
+      "loss": 0.5549,
+      "step": 32990
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 4.6904505493806595e-06,
+      "loss": 0.442,
+      "step": 33000
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 4.685596511135094e-06,
+      "loss": 0.7247,
+      "step": 33010
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 4.6807438445956e-06,
+      "loss": 0.6389,
+      "step": 33020
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 4.675892552127305e-06,
+      "loss": 0.4903,
+      "step": 33030
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 4.671042636094677e-06,
+      "loss": 0.6385,
+      "step": 33040
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 4.666194098861498e-06,
+      "loss": 0.6107,
+      "step": 33050
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 4.661346942790891e-06,
+      "loss": 0.8643,
+      "step": 33060
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 4.656501170245297e-06,
+      "loss": 0.2314,
+      "step": 33070
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 4.651656783586493e-06,
+      "loss": 0.2656,
+      "step": 33080
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 4.646813785175569e-06,
+      "loss": 0.5808,
+      "step": 33090
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 4.641972177372944e-06,
+      "loss": 0.3246,
+      "step": 33100
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 4.637131962538362e-06,
+      "loss": 0.4211,
+      "step": 33110
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 4.632293143030881e-06,
+      "loss": 0.599,
+      "step": 33120
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 4.62745572120889e-06,
+      "loss": 0.5716,
+      "step": 33130
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 4.6226196994300784e-06,
+      "loss": 0.229,
+      "step": 33140
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 4.6177850800514735e-06,
+      "loss": 0.5619,
+      "step": 33150
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 4.612951865429405e-06,
+      "loss": 0.3864,
+      "step": 33160
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 4.6081200579195265e-06,
+      "loss": 0.6615,
+      "step": 33170
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 4.6032896598768006e-06,
+      "loss": 0.6474,
+      "step": 33180
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 4.5984606736555045e-06,
+      "loss": 0.5483,
+      "step": 33190
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 4.5936331016092285e-06,
+      "loss": 0.3562,
+      "step": 33200
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 4.58880694609087e-06,
+      "loss": 0.5079,
+      "step": 33210
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 4.583982209452645e-06,
+      "loss": 0.3317,
+      "step": 33220
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 4.579158894046062e-06,
+      "loss": 0.3543,
+      "step": 33230
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 4.574337002221953e-06,
+      "loss": 0.6051,
+      "step": 33240
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 4.569516536330448e-06,
+      "loss": 0.5412,
+      "step": 33250
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 4.564697498720982e-06,
+      "loss": 0.8271,
+      "step": 33260
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 4.5598798917422975e-06,
+      "loss": 0.5189,
+      "step": 33270
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 4.555063717742438e-06,
+      "loss": 0.4982,
+      "step": 33280
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 4.550248979068744e-06,
+      "loss": 0.3839,
+      "step": 33290
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 4.5454356780678666e-06,
+      "loss": 0.7477,
+      "step": 33300
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 4.540623817085747e-06,
+      "loss": 0.3904,
+      "step": 33310
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 4.535813398467629e-06,
+      "loss": 0.4538,
+      "step": 33320
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 4.531004424558048e-06,
+      "loss": 0.3612,
+      "step": 33330
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 4.526196897700844e-06,
+      "loss": 0.4546,
+      "step": 33340
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 4.521390820239146e-06,
+      "loss": 0.3839,
+      "step": 33350
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 4.516586194515378e-06,
+      "loss": 0.8155,
+      "step": 33360
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 4.511783022871256e-06,
+      "loss": 0.289,
+      "step": 33370
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 4.506981307647786e-06,
+      "loss": 1.0293,
+      "step": 33380
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 4.502181051185268e-06,
+      "loss": 0.4668,
+      "step": 33390
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 4.497382255823289e-06,
+      "loss": 0.4752,
+      "step": 33400
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 4.492584923900722e-06,
+      "loss": 0.7419,
+      "step": 33410
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 4.487789057755726e-06,
+      "loss": 0.5496,
+      "step": 33420
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 4.482994659725753e-06,
+      "loss": 0.3482,
+      "step": 33430
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 4.478201732147531e-06,
+      "loss": 0.4638,
+      "step": 33440
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 4.4734102773570745e-06,
+      "loss": 0.6718,
+      "step": 33450
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 4.468620297689682e-06,
+      "loss": 0.2801,
+      "step": 33460
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 4.463831795479932e-06,
+      "loss": 0.5753,
+      "step": 33470
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 4.459044773061682e-06,
+      "loss": 0.6162,
+      "step": 33480
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 4.454259232768066e-06,
+      "loss": 0.5012,
+      "step": 33490
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 4.4494751769315e-06,
+      "loss": 0.4228,
+      "step": 33500
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 4.444692607883674e-06,
+      "loss": 0.5142,
+      "step": 33510
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 4.439911527955554e-06,
+      "loss": 0.442,
+      "step": 33520
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 4.4351319394773836e-06,
+      "loss": 0.4225,
+      "step": 33530
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 4.430353844778671e-06,
+      "loss": 0.5155,
+      "step": 33540
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 4.425577246188204e-06,
+      "loss": 0.8971,
+      "step": 33550
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 4.420802146034041e-06,
+      "loss": 0.8625,
+      "step": 33560
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 4.416028546643505e-06,
+      "loss": 0.619,
+      "step": 33570
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 4.4112564503431904e-06,
+      "loss": 0.756,
+      "step": 33580
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 4.406485859458957e-06,
+      "loss": 0.3665,
+      "step": 33590
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 4.401716776315938e-06,
+      "loss": 0.4867,
+      "step": 33600
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 4.396949203238524e-06,
+      "loss": 0.8304,
+      "step": 33610
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 4.392183142550369e-06,
+      "loss": 0.6656,
+      "step": 33620
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 4.387418596574399e-06,
+      "loss": 0.5113,
+      "step": 33630
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 4.382655567632791e-06,
+      "loss": 0.4387,
+      "step": 33640
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 4.377894058046992e-06,
+      "loss": 0.4837,
+      "step": 33650
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 4.3731340701377e-06,
+      "loss": 0.6714,
+      "step": 33660
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 4.368375606224877e-06,
+      "loss": 0.7669,
+      "step": 33670
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 4.36361866862774e-06,
+      "loss": 0.2445,
+      "step": 33680
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 4.358863259664763e-06,
+      "loss": 0.4635,
+      "step": 33690
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 4.354109381653672e-06,
+      "loss": 0.4453,
+      "step": 33700
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 4.349357036911453e-06,
+      "loss": 0.5797,
+      "step": 33710
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 4.344606227754341e-06,
+      "loss": 0.4042,
+      "step": 33720
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 4.339856956497821e-06,
+      "loss": 0.4951,
+      "step": 33730
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 4.33510922545663e-06,
+      "loss": 0.7429,
+      "step": 33740
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 4.330363036944755e-06,
+      "loss": 0.6737,
+      "step": 33750
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 4.325618393275432e-06,
+      "loss": 0.5931,
+      "step": 33760
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 4.320875296761139e-06,
+      "loss": 0.6849,
+      "step": 33770
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 4.316133749713608e-06,
+      "loss": 0.3549,
+      "step": 33780
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 4.311393754443805e-06,
+      "loss": 0.6275,
+      "step": 33790
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 4.306655313261955e-06,
+      "loss": 0.4788,
+      "step": 33800
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 4.301918428477511e-06,
+      "loss": 0.5819,
+      "step": 33810
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 4.2971831023991745e-06,
+      "loss": 0.4174,
+      "step": 33820
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 4.292449337334886e-06,
+      "loss": 0.6669,
+      "step": 33830
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 4.287717135591823e-06,
+      "loss": 0.4027,
+      "step": 33840
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 4.2829864994764075e-06,
+      "loss": 0.6037,
+      "step": 33850
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 4.278257431294289e-06,
+      "loss": 0.4935,
+      "step": 33860
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 4.2735299333503615e-06,
+      "loss": 0.2093,
+      "step": 33870
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 4.268804007948749e-06,
+      "loss": 0.4776,
+      "step": 33880
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 4.264079657392807e-06,
+      "loss": 0.4668,
+      "step": 33890
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 4.259356883985131e-06,
+      "loss": 0.8342,
+      "step": 33900
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 4.254635690027544e-06,
+      "loss": 0.569,
+      "step": 33910
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 4.249916077821096e-06,
+      "loss": 0.3638,
+      "step": 33920
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 4.245198049666069e-06,
+      "loss": 0.832,
+      "step": 33930
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 4.240481607861973e-06,
+      "loss": 0.8977,
+      "step": 33940
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 4.235766754707545e-06,
+      "loss": 0.5267,
+      "step": 33950
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 4.231053492500745e-06,
+      "loss": 0.4278,
+      "step": 33960
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 4.2263418235387615e-06,
+      "loss": 0.5325,
+      "step": 33970
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 4.221631750118005e-06,
+      "loss": 0.6714,
+      "step": 33980
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 4.216923274534104e-06,
+      "loss": 1.2647,
+      "step": 33990
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 4.212216399081919e-06,
+      "loss": 0.4387,
+      "step": 34000
+    },
+    {
+      "epoch": 16.04,
+      "eval_accuracy": 0.88,
+      "eval_f1": 0.88,
+      "eval_loss": 0.7549371719360352,
+      "eval_runtime": 770.409,
+      "eval_samples_per_second": 6.166,
+      "eval_steps_per_second": 1.542,
+      "step": 34000
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 4.2075111260555195e-06,
+      "loss": 0.5721,
+      "step": 34010
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 4.202807457748205e-06,
+      "loss": 0.4209,
+      "step": 34020
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 4.1981053964524725e-06,
+      "loss": 0.2641,
+      "step": 34030
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 4.193404944460062e-06,
+      "loss": 0.4175,
+      "step": 34040
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 4.188706104061912e-06,
+      "loss": 0.3068,
+      "step": 34050
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 4.18400887754818e-06,
+      "loss": 0.5447,
+      "step": 34060
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 4.17931326720824e-06,
+      "loss": 0.4912,
+      "step": 34070
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 4.1746192753306685e-06,
+      "loss": 0.3235,
+      "step": 34080
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 4.16992690420327e-06,
+      "loss": 0.695,
+      "step": 34090
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 4.165236156113046e-06,
+      "loss": 0.7171,
+      "step": 34100
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 4.160547033346206e-06,
+      "loss": 0.2027,
+      "step": 34110
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 4.15585953818817e-06,
+      "loss": 0.4179,
+      "step": 34120
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 4.151173672923573e-06,
+      "loss": 0.545,
+      "step": 34130
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 4.146489439836246e-06,
+      "loss": 0.4815,
+      "step": 34140
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 4.141806841209228e-06,
+      "loss": 0.2793,
+      "step": 34150
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 4.137125879324759e-06,
+      "loss": 0.5699,
+      "step": 34160
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 4.1324465564642845e-06,
+      "loss": 0.7039,
+      "step": 34170
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 4.127768874908445e-06,
+      "loss": 0.4332,
+      "step": 34180
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 4.123092836937095e-06,
+      "loss": 0.3581,
+      "step": 34190
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 4.118418444829271e-06,
+      "loss": 0.6377,
+      "step": 34200
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 4.1137457008632175e-06,
+      "loss": 0.8027,
+      "step": 34210
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 4.1090746073163686e-06,
+      "loss": 0.3202,
+      "step": 34220
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 4.104405166465368e-06,
+      "loss": 0.325,
+      "step": 34230
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 4.099737380586036e-06,
+      "loss": 0.3973,
+      "step": 34240
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 4.095071251953399e-06,
+      "loss": 0.6119,
+      "step": 34250
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 4.090406782841671e-06,
+      "loss": 0.3482,
+      "step": 34260
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 4.085743975524253e-06,
+      "loss": 0.2866,
+      "step": 34270
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 4.081082832273752e-06,
+      "loss": 0.593,
+      "step": 34280
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 4.07642335536194e-06,
+      "loss": 0.3209,
+      "step": 34290
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 4.071765547059796e-06,
+      "loss": 0.4869,
+      "step": 34300
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 4.067109409637476e-06,
+      "loss": 0.5368,
+      "step": 34310
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 4.062454945364326e-06,
+      "loss": 0.5865,
+      "step": 34320
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 4.057802156508877e-06,
+      "loss": 0.9716,
+      "step": 34330
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 4.053151045338842e-06,
+      "loss": 0.5427,
+      "step": 34340
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 4.0485016141211134e-06,
+      "loss": 0.728,
+      "step": 34350
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 4.04385386512177e-06,
+      "loss": 0.4432,
+      "step": 34360
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 4.039207800606065e-06,
+      "loss": 0.6848,
+      "step": 34370
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 4.034563422838437e-06,
+      "loss": 0.2282,
+      "step": 34380
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 4.029920734082499e-06,
+      "loss": 0.2571,
+      "step": 34390
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 4.025279736601039e-06,
+      "loss": 0.4565,
+      "step": 34400
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 4.0206404326560205e-06,
+      "loss": 0.3911,
+      "step": 34410
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 4.016002824508593e-06,
+      "loss": 0.4225,
+      "step": 34420
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 4.011366914419062e-06,
+      "loss": 0.1458,
+      "step": 34430
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 4.006732704646918e-06,
+      "loss": 0.2513,
+      "step": 34440
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 4.002100197450817e-06,
+      "loss": 0.5512,
+      "step": 34450
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 3.997469395088586e-06,
+      "loss": 0.4876,
+      "step": 34460
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 3.992840299817223e-06,
+      "loss": 0.9696,
+      "step": 34470
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 3.988212913892892e-06,
+      "loss": 0.4729,
+      "step": 34480
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 3.983587239570926e-06,
+      "loss": 0.5691,
+      "step": 34490
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 3.978963279105821e-06,
+      "loss": 0.6559,
+      "step": 34500
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 3.974341034751237e-06,
+      "loss": 0.5865,
+      "step": 34510
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 3.969720508760006e-06,
+      "loss": 0.5114,
+      "step": 34520
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 3.965101703384111e-06,
+      "loss": 0.8371,
+      "step": 34530
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 3.960484620874706e-06,
+      "loss": 0.7119,
+      "step": 34540
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 3.955869263482096e-06,
+      "loss": 0.6099,
+      "step": 34550
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 3.951255633455752e-06,
+      "loss": 0.9088,
+      "step": 34560
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 3.946643733044303e-06,
+      "loss": 0.6616,
+      "step": 34570
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 3.942033564495532e-06,
+      "loss": 0.6684,
+      "step": 34580
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 3.937425130056378e-06,
+      "loss": 0.8524,
+      "step": 34590
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 3.9328184319729346e-06,
+      "loss": 0.3119,
+      "step": 34600
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 3.9282134724904555e-06,
+      "loss": 0.8073,
+      "step": 34610
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 3.92361025385334e-06,
+      "loss": 0.4739,
+      "step": 34620
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 3.919008778305139e-06,
+      "loss": 0.7186,
+      "step": 34630
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 3.914409048088559e-06,
+      "loss": 0.3444,
+      "step": 34640
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 3.90981106544545e-06,
+      "loss": 0.3746,
+      "step": 34650
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 3.905214832616813e-06,
+      "loss": 0.3366,
+      "step": 34660
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 3.900620351842797e-06,
+      "loss": 0.5922,
+      "step": 34670
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 3.896027625362695e-06,
+      "loss": 0.5424,
+      "step": 34680
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 3.891436655414948e-06,
+      "loss": 0.4661,
+      "step": 34690
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 3.886847444237134e-06,
+      "loss": 0.5704,
+      "step": 34700
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 3.882259994065986e-06,
+      "loss": 0.3393,
+      "step": 34710
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 3.877674307137371e-06,
+      "loss": 0.6378,
+      "step": 34720
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 3.873090385686292e-06,
+      "loss": 0.2401,
+      "step": 34730
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 3.868508231946894e-06,
+      "loss": 0.6446,
+      "step": 34740
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.8639278481524725e-06,
+      "loss": 0.7911,
+      "step": 34750
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.859349236535446e-06,
+      "loss": 0.4049,
+      "step": 34760
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.854772399327374e-06,
+      "loss": 0.5855,
+      "step": 34770
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.850197338758952e-06,
+      "loss": 0.5569,
+      "step": 34780
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.845624057060006e-06,
+      "loss": 0.5836,
+      "step": 34790
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.841052556459503e-06,
+      "loss": 0.5529,
+      "step": 34800
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.83648283918554e-06,
+      "loss": 0.413,
+      "step": 34810
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.831914907465331e-06,
+      "loss": 0.4782,
+      "step": 34820
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.8273487635252355e-06,
+      "loss": 0.2445,
+      "step": 34830
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.822784409590735e-06,
+      "loss": 0.7761,
+      "step": 34840
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.818221847886444e-06,
+      "loss": 0.3729,
+      "step": 34850
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.813661080636098e-06,
+      "loss": 0.6889,
+      "step": 34860
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.809102110062559e-06,
+      "loss": 0.6488,
+      "step": 34870
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.8045449383878144e-06,
+      "loss": 0.5869,
+      "step": 34880
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.799989567832971e-06,
+      "loss": 0.6171,
+      "step": 34890
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.795436000618269e-06,
+      "loss": 0.7484,
+      "step": 34900
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.7908842389630542e-06,
+      "loss": 0.7111,
+      "step": 34910
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.7863342850858034e-06,
+      "loss": 0.4369,
+      "step": 34920
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.7817861412041044e-06,
+      "loss": 0.3915,
+      "step": 34930
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.7772398095346738e-06,
+      "loss": 0.3937,
+      "step": 34940
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.7726952922933353e-06,
+      "loss": 0.597,
+      "step": 34950
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.768152591695031e-06,
+      "loss": 0.6374,
+      "step": 34960
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.763611709953819e-06,
+      "loss": 0.5747,
+      "step": 34970
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.7590726492828704e-06,
+      "loss": 0.413,
+      "step": 34980
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.7545354118944664e-06,
+      "loss": 0.7505,
+      "step": 34990
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.750000000000002e-06,
+      "loss": 0.7681,
+      "step": 35000
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.7454664158099814e-06,
+      "loss": 0.7081,
+      "step": 35010
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.7409346615340203e-06,
+      "loss": 0.3066,
+      "step": 35020
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.736404739380836e-06,
+      "loss": 0.7753,
+      "step": 35030
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.731876651558265e-06,
+      "loss": 0.5647,
+      "step": 35040
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.727350400273237e-06,
+      "loss": 0.4773,
+      "step": 35050
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.7228259877317942e-06,
+      "loss": 0.5379,
+      "step": 35060
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.7183034161390787e-06,
+      "loss": 0.2418,
+      "step": 35070
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.7137826876993383e-06,
+      "loss": 0.4506,
+      "step": 35080
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.7092638046159198e-06,
+      "loss": 0.6835,
+      "step": 35090
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.704746769091273e-06,
+      "loss": 0.2637,
+      "step": 35100
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.7002315833269455e-06,
+      "loss": 0.3303,
+      "step": 35110
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.695718249523581e-06,
+      "loss": 0.7317,
+      "step": 35120
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.6912067698809306e-06,
+      "loss": 0.4164,
+      "step": 35130
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.6866971465978297e-06,
+      "loss": 0.5373,
+      "step": 35140
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.6821893818722152e-06,
+      "loss": 0.6486,
+      "step": 35150
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.677683477901117e-06,
+      "loss": 0.8014,
+      "step": 35160
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.673179436880658e-06,
+      "loss": 0.5934,
+      "step": 35170
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.6686772610060517e-06,
+      "loss": 0.8094,
+      "step": 35180
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.6641769524716058e-06,
+      "loss": 0.3018,
+      "step": 35190
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.6596785134707146e-06,
+      "loss": 0.7329,
+      "step": 35200
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.6551819461958634e-06,
+      "loss": 0.3554,
+      "step": 35210
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.650687252838622e-06,
+      "loss": 0.4408,
+      "step": 35220
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.6461944355896537e-06,
+      "loss": 0.457,
+      "step": 35230
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.641703496638701e-06,
+      "loss": 0.4134,
+      "step": 35240
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.637214438174593e-06,
+      "loss": 0.5221,
+      "step": 35250
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.632727262385243e-06,
+      "loss": 0.4694,
+      "step": 35260
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.6282419714576444e-06,
+      "loss": 0.5371,
+      "step": 35270
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.6237585675778735e-06,
+      "loss": 0.3918,
+      "step": 35280
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.619277052931089e-06,
+      "loss": 0.6303,
+      "step": 35290
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.6147974297015245e-06,
+      "loss": 0.7591,
+      "step": 35300
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.6103197000724957e-06,
+      "loss": 0.6012,
+      "step": 35310
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.6058438662263894e-06,
+      "loss": 0.6378,
+      "step": 35320
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.6013699303446784e-06,
+      "loss": 0.5929,
+      "step": 35330
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.596897894607905e-06,
+      "loss": 0.4922,
+      "step": 35340
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.5924277611956798e-06,
+      "loss": 0.5399,
+      "step": 35350
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.587959532286692e-06,
+      "loss": 0.8259,
+      "step": 35360
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.583493210058706e-06,
+      "loss": 0.7079,
+      "step": 35370
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.579028796688551e-06,
+      "loss": 0.6888,
+      "step": 35380
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.574566294352129e-06,
+      "loss": 0.8479,
+      "step": 35390
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.5701057052244076e-06,
+      "loss": 0.6815,
+      "step": 35400
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.565647031479422e-06,
+      "loss": 0.725,
+      "step": 35410
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.5611902752902814e-06,
+      "loss": 1.1404,
+      "step": 35420
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.556735438829154e-06,
+      "loss": 0.3504,
+      "step": 35430
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.5522825242672673e-06,
+      "loss": 0.3773,
+      "step": 35440
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.5478315337749183e-06,
+      "loss": 1.0465,
+      "step": 35450
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.543382469521471e-06,
+      "loss": 0.6458,
+      "step": 35460
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.5389353336753426e-06,
+      "loss": 0.4453,
+      "step": 35470
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.5344901284040122e-06,
+      "loss": 0.2837,
+      "step": 35480
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.530046855874021e-06,
+      "loss": 0.618,
+      "step": 35490
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.525605518250964e-06,
+      "loss": 0.4564,
+      "step": 35500
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.521166117699493e-06,
+      "loss": 0.8331,
+      "step": 35510
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.5167286563833268e-06,
+      "loss": 0.4391,
+      "step": 35520
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.512293136465221e-06,
+      "loss": 0.4842,
+      "step": 35530
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.507859560106998e-06,
+      "loss": 0.7373,
+      "step": 35540
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.5034279294695245e-06,
+      "loss": 0.3786,
+      "step": 35550
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.498998246712731e-06,
+      "loss": 0.322,
+      "step": 35560
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.4945705139955865e-06,
+      "loss": 0.4963,
+      "step": 35570
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.490144733476116e-06,
+      "loss": 0.6952,
+      "step": 35580
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.4857209073113893e-06,
+      "loss": 0.5512,
+      "step": 35590
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.4812990376575235e-06,
+      "loss": 0.4881,
+      "step": 35600
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.4768791266696926e-06,
+      "loss": 0.5135,
+      "step": 35610
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.4724611765021e-06,
+      "loss": 0.7624,
+      "step": 35620
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.4680451893080028e-06,
+      "loss": 0.3704,
+      "step": 35630
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.4636311672396958e-06,
+      "loss": 0.2933,
+      "step": 35640
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.4592191124485265e-06,
+      "loss": 0.5389,
+      "step": 35650
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.454809027084874e-06,
+      "loss": 0.5478,
+      "step": 35660
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.4504009132981614e-06,
+      "loss": 0.5282,
+      "step": 35670
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.445994773236847e-06,
+      "loss": 0.6104,
+      "step": 35680
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.4415906090484316e-06,
+      "loss": 0.5048,
+      "step": 35690
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.437188422879452e-06,
+      "loss": 0.2955,
+      "step": 35700
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.4327882168754793e-06,
+      "loss": 0.404,
+      "step": 35710
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.4283899931811203e-06,
+      "loss": 0.5149,
+      "step": 35720
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.4239937539400167e-06,
+      "loss": 0.7681,
+      "step": 35730
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.4195995012948382e-06,
+      "loss": 0.4023,
+      "step": 35740
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 3.415207237387297e-06,
+      "loss": 0.5534,
+      "step": 35750
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 3.4108169643581244e-06,
+      "loss": 0.7659,
+      "step": 35760
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 3.4064286843470882e-06,
+      "loss": 0.6809,
+      "step": 35770
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 3.4020423994929824e-06,
+      "loss": 0.4474,
+      "step": 35780
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 3.397658111933628e-06,
+      "loss": 0.496,
+      "step": 35790
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 3.3932758238058738e-06,
+      "loss": 0.2837,
+      "step": 35800
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 3.3888955372455947e-06,
+      "loss": 0.1942,
+      "step": 35810
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 3.3845172543876883e-06,
+      "loss": 0.5356,
+      "step": 35820
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 3.380140977366077e-06,
+      "loss": 0.7328,
+      "step": 35830
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 3.3757667083137015e-06,
+      "loss": 0.2156,
+      "step": 35840
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 3.371394449362533e-06,
+      "loss": 0.4024,
+      "step": 35850
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 3.367024202643555e-06,
+      "loss": 0.3176,
+      "step": 35860
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 3.3626559702867738e-06,
+      "loss": 0.627,
+      "step": 35870
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 3.358289754421211e-06,
+      "loss": 0.3277,
+      "step": 35880
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 3.353925557174907e-06,
+      "loss": 0.5174,
+      "step": 35890
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 3.3495633806749203e-06,
+      "loss": 0.4933,
+      "step": 35900
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 3.345203227047322e-06,
+      "loss": 0.4456,
+      "step": 35910
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 3.340845098417197e-06,
+      "loss": 0.6054,
+      "step": 35920
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 3.3364889969086428e-06,
+      "loss": 0.7091,
+      "step": 35930
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 3.3321349246447753e-06,
+      "loss": 0.7329,
+      "step": 35940
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 3.327782883747714e-06,
+      "loss": 0.562,
+      "step": 35950
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 3.323432876338593e-06,
+      "loss": 0.6282,
+      "step": 35960
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 3.319084904537545e-06,
+      "loss": 0.574,
+      "step": 35970
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 3.3147389704637263e-06,
+      "loss": 0.3558,
+      "step": 35980
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 3.310395076235291e-06,
+      "loss": 0.834,
+      "step": 35990
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 3.3060532239694e-06,
+      "loss": 0.5013,
+      "step": 36000
+    },
+    {
+      "epoch": 17.04,
+      "eval_accuracy": 0.8736842105263158,
+      "eval_f1": 0.8736842105263158,
+      "eval_loss": 0.7713430523872375,
+      "eval_runtime": 768.879,
+      "eval_samples_per_second": 6.178,
+      "eval_steps_per_second": 1.545,
+      "step": 36000
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 3.3017134157822186e-06,
+      "loss": 0.3416,
+      "step": 36010
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 3.2973756537889164e-06,
+      "loss": 0.519,
+      "step": 36020
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 3.2930399401036647e-06,
+      "loss": 0.7151,
+      "step": 36030
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 3.288706276839642e-06,
+      "loss": 0.7257,
+      "step": 36040
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 3.2843746661090225e-06,
+      "loss": 0.3615,
+      "step": 36050
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 3.2800451100229773e-06,
+      "loss": 0.2905,
+      "step": 36060
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 3.275717610691677e-06,
+      "loss": 1.0138,
+      "step": 36070
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 3.2713921702242993e-06,
+      "loss": 0.5712,
+      "step": 36080
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 3.267068790729008e-06,
+      "loss": 0.4467,
+      "step": 36090
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 3.262747474312966e-06,
+      "loss": 0.3777,
+      "step": 36100
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 3.2584282230823308e-06,
+      "loss": 0.5909,
+      "step": 36110
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 3.2541110391422493e-06,
+      "loss": 0.3593,
+      "step": 36120
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 3.2497959245968696e-06,
+      "loss": 0.3639,
+      "step": 36130
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 3.2454828815493277e-06,
+      "loss": 0.6528,
+      "step": 36140
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 3.241171912101743e-06,
+      "loss": 0.3752,
+      "step": 36150
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 3.23686301835523e-06,
+      "loss": 0.5945,
+      "step": 36160
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 3.2325562024098906e-06,
+      "loss": 0.3346,
+      "step": 36170
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 3.2282514663648196e-06,
+      "loss": 0.3911,
+      "step": 36180
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 3.22394881231809e-06,
+      "loss": 0.7658,
+      "step": 36190
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 3.219648242366763e-06,
+      "loss": 0.352,
+      "step": 36200
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 3.215349758606885e-06,
+      "loss": 0.6865,
+      "step": 36210
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 3.2110533631334805e-06,
+      "loss": 0.5107,
+      "step": 36220
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 3.20675905804057e-06,
+      "loss": 0.4335,
+      "step": 36230
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 3.2024668454211373e-06,
+      "loss": 0.556,
+      "step": 36240
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 3.1981767273671563e-06,
+      "loss": 0.6473,
+      "step": 36250
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 3.1938887059695775e-06,
+      "loss": 0.737,
+      "step": 36260
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 3.189602783318334e-06,
+      "loss": 0.5317,
+      "step": 36270
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 3.185318961502331e-06,
+      "loss": 1.087,
+      "step": 36280
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 3.1810372426094507e-06,
+      "loss": 0.9457,
+      "step": 36290
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 3.1767576287265512e-06,
+      "loss": 0.3877,
+      "step": 36300
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 3.1724801219394645e-06,
+      "loss": 0.1997,
+      "step": 36310
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 3.1682047243329947e-06,
+      "loss": 0.4113,
+      "step": 36320
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 3.1639314379909188e-06,
+      "loss": 0.5148,
+      "step": 36330
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 3.1596602649959844e-06,
+      "loss": 0.5285,
+      "step": 36340
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 3.1553912074299106e-06,
+      "loss": 0.3002,
+      "step": 36350
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 3.151124267373381e-06,
+      "loss": 0.9515,
+      "step": 36360
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 3.1468594469060546e-06,
+      "loss": 0.3972,
+      "step": 36370
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 3.1425967481065515e-06,
+      "loss": 0.3017,
+      "step": 36380
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 3.1383361730524596e-06,
+      "loss": 0.3262,
+      "step": 36390
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 3.13407772382033e-06,
+      "loss": 0.2876,
+      "step": 36400
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 3.1298214024856807e-06,
+      "loss": 0.4643,
+      "step": 36410
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 3.1255672111229896e-06,
+      "loss": 0.2409,
+      "step": 36420
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 3.121315151805699e-06,
+      "loss": 0.5946,
+      "step": 36430
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 3.11706522660621e-06,
+      "loss": 0.5023,
+      "step": 36440
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 3.1128174375958836e-06,
+      "loss": 0.5401,
+      "step": 36450
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 3.1085717868450442e-06,
+      "loss": 0.8222,
+      "step": 36460
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 3.104328276422968e-06,
+      "loss": 0.3783,
+      "step": 36470
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 3.100086908397891e-06,
+      "loss": 0.5229,
+      "step": 36480
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 3.095847684837003e-06,
+      "loss": 0.2334,
+      "step": 36490
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 3.0916106078064522e-06,
+      "loss": 0.471,
+      "step": 36500
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 3.0873756793713374e-06,
+      "loss": 0.4497,
+      "step": 36510
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 3.083142901595711e-06,
+      "loss": 0.924,
+      "step": 36520
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 3.0789122765425775e-06,
+      "loss": 0.7265,
+      "step": 36530
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 3.0746838062738935e-06,
+      "loss": 0.676,
+      "step": 36540
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 3.0704574928505588e-06,
+      "loss": 0.45,
+      "step": 36550
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 3.066233338332435e-06,
+      "loss": 0.4074,
+      "step": 36560
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 3.06201134477832e-06,
+      "loss": 0.4613,
+      "step": 36570
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 3.057791514245965e-06,
+      "loss": 0.4177,
+      "step": 36580
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 3.0535738487920557e-06,
+      "loss": 0.3157,
+      "step": 36590
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 3.04935835047224e-06,
+      "loss": 0.402,
+      "step": 36600
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 3.045145021341097e-06,
+      "loss": 0.5566,
+      "step": 36610
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 3.0409338634521526e-06,
+      "loss": 0.8892,
+      "step": 36620
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 3.036724878857873e-06,
+      "loss": 0.3931,
+      "step": 36630
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 3.032518069609665e-06,
+      "loss": 0.4067,
+      "step": 36640
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 3.0283134377578804e-06,
+      "loss": 0.5997,
+      "step": 36650
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 3.024110985351804e-06,
+      "loss": 0.6039,
+      "step": 36660
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 3.019910714439662e-06,
+      "loss": 1.072,
+      "step": 36670
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 3.0157126270686113e-06,
+      "loss": 0.6233,
+      "step": 36680
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 3.011516725284747e-06,
+      "loss": 0.619,
+      "step": 36690
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 3.007323011133107e-06,
+      "loss": 0.6432,
+      "step": 36700
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 3.0031314866576535e-06,
+      "loss": 0.5534,
+      "step": 36710
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 2.9989421539012843e-06,
+      "loss": 0.8209,
+      "step": 36720
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 2.9947550149058308e-06,
+      "loss": 0.3964,
+      "step": 36730
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 2.99057007171205e-06,
+      "loss": 0.2976,
+      "step": 36740
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.9863873263596374e-06,
+      "loss": 0.4838,
+      "step": 36750
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.982206780887212e-06,
+      "loss": 0.6236,
+      "step": 36760
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.9780284373323167e-06,
+      "loss": 0.4662,
+      "step": 36770
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.9738522977314257e-06,
+      "loss": 0.3537,
+      "step": 36780
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.9696783641199416e-06,
+      "loss": 0.5034,
+      "step": 36790
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.9655066385321886e-06,
+      "loss": 0.2501,
+      "step": 36800
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.9613371230014146e-06,
+      "loss": 0.6887,
+      "step": 36810
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.9571698195597902e-06,
+      "loss": 0.5212,
+      "step": 36820
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.953004730238409e-06,
+      "loss": 0.4871,
+      "step": 36830
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.9488418570672825e-06,
+      "loss": 0.3243,
+      "step": 36840
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.9446812020753518e-06,
+      "loss": 0.4076,
+      "step": 36850
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.9405227672904613e-06,
+      "loss": 0.3996,
+      "step": 36860
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.9363665547393856e-06,
+      "loss": 0.7454,
+      "step": 36870
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.932212566447809e-06,
+      "loss": 0.6302,
+      "step": 36880
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.928060804440339e-06,
+      "loss": 0.3576,
+      "step": 36890
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.9239112707404932e-06,
+      "loss": 0.3181,
+      "step": 36900
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.9197639673707034e-06,
+      "loss": 0.608,
+      "step": 36910
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.915618896352314e-06,
+      "loss": 0.2541,
+      "step": 36920
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.9114760597055798e-06,
+      "loss": 0.2384,
+      "step": 36930
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.907335459449678e-06,
+      "loss": 0.529,
+      "step": 36940
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.9031970976026783e-06,
+      "loss": 0.5254,
+      "step": 36950
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.89906097618157e-06,
+      "loss": 0.9418,
+      "step": 36960
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.8949270972022473e-06,
+      "loss": 0.3854,
+      "step": 36970
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.8907954626795167e-06,
+      "loss": 0.5665,
+      "step": 36980
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.886666074627084e-06,
+      "loss": 0.5238,
+      "step": 36990
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.882538935057563e-06,
+      "loss": 0.2699,
+      "step": 37000
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.8784140459824717e-06,
+      "loss": 0.7433,
+      "step": 37010
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.8742914094122317e-06,
+      "loss": 0.6341,
+      "step": 37020
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.8701710273561635e-06,
+      "loss": 0.4695,
+      "step": 37030
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.8660529018224937e-06,
+      "loss": 0.6994,
+      "step": 37040
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.861937034818345e-06,
+      "loss": 0.2551,
+      "step": 37050
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.8578234283497423e-06,
+      "loss": 0.4353,
+      "step": 37060
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.8537120844216044e-06,
+      "loss": 0.5129,
+      "step": 37070
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.8496030050377552e-06,
+      "loss": 0.8072,
+      "step": 37080
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.845496192200908e-06,
+      "loss": 0.6905,
+      "step": 37090
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.841391647912673e-06,
+      "loss": 0.489,
+      "step": 37100
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.8372893741735546e-06,
+      "loss": 0.5638,
+      "step": 37110
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.8331893729829527e-06,
+      "loss": 0.5291,
+      "step": 37120
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.8290916463391576e-06,
+      "loss": 0.5672,
+      "step": 37130
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.8249961962393506e-06,
+      "loss": 0.5564,
+      "step": 37140
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.8209030246796054e-06,
+      "loss": 0.4435,
+      "step": 37150
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.8168121336548832e-06,
+      "loss": 0.3198,
+      "step": 37160
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.812723525159033e-06,
+      "loss": 0.4359,
+      "step": 37170
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.808637201184797e-06,
+      "loss": 0.2043,
+      "step": 37180
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.8045531637237984e-06,
+      "loss": 0.6797,
+      "step": 37190
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.8004714147665487e-06,
+      "loss": 0.6557,
+      "step": 37200
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.7963919563024366e-06,
+      "loss": 0.7913,
+      "step": 37210
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.7923147903197473e-06,
+      "loss": 0.5721,
+      "step": 37220
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.7882399188056393e-06,
+      "loss": 0.7529,
+      "step": 37230
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.7841673437461557e-06,
+      "loss": 0.2419,
+      "step": 37240
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.7800970671262205e-06,
+      "loss": 0.5299,
+      "step": 37250
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.7760290909296344e-06,
+      "loss": 0.8419,
+      "step": 37260
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.771963417139084e-06,
+      "loss": 0.6347,
+      "step": 37270
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.7679000477361276e-06,
+      "loss": 0.6209,
+      "step": 37280
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.763838984701204e-06,
+      "loss": 0.389,
+      "step": 37290
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.759780230013618e-06,
+      "loss": 0.5476,
+      "step": 37300
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.755723785651567e-06,
+      "loss": 0.8579,
+      "step": 37310
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.7516696535921084e-06,
+      "loss": 0.7398,
+      "step": 37320
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.7476178358111765e-06,
+      "loss": 0.5888,
+      "step": 37330
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.7435683342835794e-06,
+      "loss": 0.5358,
+      "step": 37340
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.739521150982994e-06,
+      "loss": 0.4386,
+      "step": 37350
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.735476287881967e-06,
+      "loss": 0.2455,
+      "step": 37360
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.731433746951919e-06,
+      "loss": 0.2797,
+      "step": 37370
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.727393530163136e-06,
+      "loss": 0.5183,
+      "step": 37380
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.7233556394847664e-06,
+      "loss": 0.3987,
+      "step": 37390
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.719320076884827e-06,
+      "loss": 0.4604,
+      "step": 37400
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.7152868443302092e-06,
+      "loss": 0.5954,
+      "step": 37410
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.7112559437866583e-06,
+      "loss": 0.5356,
+      "step": 37420
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.707227377218785e-06,
+      "loss": 0.5328,
+      "step": 37430
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.7032011465900655e-06,
+      "loss": 0.4616,
+      "step": 37440
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.6991772538628325e-06,
+      "loss": 0.4083,
+      "step": 37450
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.6951557009982875e-06,
+      "loss": 0.5883,
+      "step": 37460
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.691136489956486e-06,
+      "loss": 0.4895,
+      "step": 37470
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.687119622696339e-06,
+      "loss": 0.6126,
+      "step": 37480
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.683105101175617e-06,
+      "loss": 0.2373,
+      "step": 37490
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.6790929273509547e-06,
+      "loss": 0.445,
+      "step": 37500
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.6750831031778352e-06,
+      "loss": 0.5903,
+      "step": 37510
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.6710756306105973e-06,
+      "loss": 0.2374,
+      "step": 37520
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.6670705116024345e-06,
+      "loss": 0.5996,
+      "step": 37530
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.663067748105393e-06,
+      "loss": 0.3053,
+      "step": 37540
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.6590673420703694e-06,
+      "loss": 0.325,
+      "step": 37550
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.6550692954471202e-06,
+      "loss": 0.351,
+      "step": 37560
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.6510736101842365e-06,
+      "loss": 0.8695,
+      "step": 37570
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.64708028822917e-06,
+      "loss": 0.4065,
+      "step": 37580
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.643089331528214e-06,
+      "loss": 0.563,
+      "step": 37590
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.6391007420265183e-06,
+      "loss": 0.8252,
+      "step": 37600
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.63511452166807e-06,
+      "loss": 0.2165,
+      "step": 37610
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.631130672395705e-06,
+      "loss": 0.5479,
+      "step": 37620
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.6271491961511017e-06,
+      "loss": 0.5603,
+      "step": 37630
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.623170094874782e-06,
+      "loss": 0.4072,
+      "step": 37640
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.619193370506119e-06,
+      "loss": 0.5133,
+      "step": 37650
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.615219024983312e-06,
+      "loss": 0.6001,
+      "step": 37660
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.6112470602434107e-06,
+      "loss": 0.4305,
+      "step": 37670
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.6072774782223036e-06,
+      "loss": 0.5292,
+      "step": 37680
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.6033102808547136e-06,
+      "loss": 0.538,
+      "step": 37690
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.599345470074209e-06,
+      "loss": 0.4507,
+      "step": 37700
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.595383047813189e-06,
+      "loss": 0.2525,
+      "step": 37710
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.591423016002889e-06,
+      "loss": 0.3462,
+      "step": 37720
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.5874653765733816e-06,
+      "loss": 0.4602,
+      "step": 37730
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.5835101314535712e-06,
+      "loss": 0.5111,
+      "step": 37740
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 2.5795572825711963e-06,
+      "loss": 0.3537,
+      "step": 37750
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 2.5756068318528278e-06,
+      "loss": 0.5182,
+      "step": 37760
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 2.571658781223867e-06,
+      "loss": 0.2725,
+      "step": 37770
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 2.5677131326085447e-06,
+      "loss": 0.6178,
+      "step": 37780
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 2.5637698879299253e-06,
+      "loss": 0.4089,
+      "step": 37790
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 2.5598290491098956e-06,
+      "loss": 0.4844,
+      "step": 37800
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 2.5558906180691747e-06,
+      "loss": 0.525,
+      "step": 37810
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 2.5519545967273048e-06,
+      "loss": 0.8549,
+      "step": 37820
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 2.548020987002654e-06,
+      "loss": 0.4982,
+      "step": 37830
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 2.5440897908124176e-06,
+      "loss": 0.5018,
+      "step": 37840
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 2.5401610100726124e-06,
+      "loss": 0.3848,
+      "step": 37850
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 2.536234646698077e-06,
+      "loss": 0.3812,
+      "step": 37860
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 2.5323107026024746e-06,
+      "loss": 0.4099,
+      "step": 37870
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 2.528389179698286e-06,
+      "loss": 0.5027,
+      "step": 37880
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 2.5244700798968166e-06,
+      "loss": 0.3912,
+      "step": 37890
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 2.520553405108188e-06,
+      "loss": 0.3629,
+      "step": 37900
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 2.5166391572413403e-06,
+      "loss": 0.951,
+      "step": 37910
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 2.5127273382040244e-06,
+      "loss": 0.3341,
+      "step": 37920
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 2.5088179499028205e-06,
+      "loss": 0.8679,
+      "step": 37930
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 2.5049109942431135e-06,
+      "loss": 0.8129,
+      "step": 37940
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 2.5010064731291072e-06,
+      "loss": 0.6507,
+      "step": 37950
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 2.497104388463818e-06,
+      "loss": 1.0909,
+      "step": 37960
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 2.4932047421490708e-06,
+      "loss": 0.3817,
+      "step": 37970
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 2.4893075360855116e-06,
+      "loss": 0.8744,
+      "step": 37980
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 2.485412772172589e-06,
+      "loss": 0.4002,
+      "step": 37990
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 2.4815204523085656e-06,
+      "loss": 0.9572,
+      "step": 38000
+    },
+    {
+      "epoch": 18.04,
+      "eval_accuracy": 0.868421052631579,
+      "eval_f1": 0.868421052631579,
+      "eval_loss": 0.761253297328949,
+      "eval_runtime": 768.4537,
+      "eval_samples_per_second": 6.181,
+      "eval_steps_per_second": 1.546,
+      "step": 38000
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 2.4776305783905063e-06,
+      "loss": 1.176,
+      "step": 38010
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 2.473743152314288e-06,
+      "loss": 0.3656,
+      "step": 38020
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 2.4698581759746e-06,
+      "loss": 0.3558,
+      "step": 38030
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 2.465975651264931e-06,
+      "loss": 0.723,
+      "step": 38040
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 2.462095580077575e-06,
+      "loss": 0.747,
+      "step": 38050
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 2.4582179643036316e-06,
+      "loss": 0.5305,
+      "step": 38060
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 2.4543428058330024e-06,
+      "loss": 0.5172,
+      "step": 38070
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 2.450470106554396e-06,
+      "loss": 0.4605,
+      "step": 38080
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 2.44659986835532e-06,
+      "loss": 0.3962,
+      "step": 38090
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 2.442732093122077e-06,
+      "loss": 0.4428,
+      "step": 38100
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 2.4388667827397724e-06,
+      "loss": 0.5217,
+      "step": 38110
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 2.4350039390923164e-06,
+      "loss": 0.6184,
+      "step": 38120
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 2.431143564062411e-06,
+      "loss": 0.4994,
+      "step": 38130
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 2.427285659531556e-06,
+      "loss": 0.578,
+      "step": 38140
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 2.4234302273800455e-06,
+      "loss": 0.2993,
+      "step": 38150
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 2.4195772694869712e-06,
+      "loss": 0.3344,
+      "step": 38160
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 2.415726787730216e-06,
+      "loss": 0.4295,
+      "step": 38170
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 2.411878783986465e-06,
+      "loss": 0.697,
+      "step": 38180
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 2.4080332601311813e-06,
+      "loss": 0.3132,
+      "step": 38190
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 2.404190218038628e-06,
+      "loss": 0.5248,
+      "step": 38200
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 2.4003496595818557e-06,
+      "loss": 0.457,
+      "step": 38210
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 2.396511586632711e-06,
+      "loss": 0.3919,
+      "step": 38220
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 2.39267600106182e-06,
+      "loss": 0.2957,
+      "step": 38230
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 2.3888429047386022e-06,
+      "loss": 0.4636,
+      "step": 38240
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 2.3850122995312623e-06,
+      "loss": 0.2432,
+      "step": 38250
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.3811841873067865e-06,
+      "loss": 0.4494,
+      "step": 38260
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.3773585699309594e-06,
+      "loss": 0.6232,
+      "step": 38270
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.3735354492683315e-06,
+      "loss": 0.4032,
+      "step": 38280
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.3697148271822498e-06,
+      "loss": 0.4197,
+      "step": 38290
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.3658967055348347e-06,
+      "loss": 0.4868,
+      "step": 38300
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.3620810861869978e-06,
+      "loss": 0.2848,
+      "step": 38310
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.3582679709984236e-06,
+      "loss": 0.3721,
+      "step": 38320
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.3544573618275777e-06,
+      "loss": 0.2844,
+      "step": 38330
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.3506492605317054e-06,
+      "loss": 0.9406,
+      "step": 38340
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.3468436689668285e-06,
+      "loss": 0.6308,
+      "step": 38350
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.343040588987745e-06,
+      "loss": 0.4966,
+      "step": 38360
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.3392400224480327e-06,
+      "loss": 0.2921,
+      "step": 38370
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.3354419712000383e-06,
+      "loss": 0.7538,
+      "step": 38380
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.331646437094888e-06,
+      "loss": 0.146,
+      "step": 38390
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.3278534219824766e-06,
+      "loss": 0.588,
+      "step": 38400
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.3240629277114765e-06,
+      "loss": 0.4177,
+      "step": 38410
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.320274956129328e-06,
+      "loss": 0.1569,
+      "step": 38420
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.316489509082243e-06,
+      "loss": 0.2267,
+      "step": 38430
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.3127065884152e-06,
+      "loss": 0.4758,
+      "step": 38440
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.30892619597195e-06,
+      "loss": 0.7699,
+      "step": 38450
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.3051483335950098e-06,
+      "loss": 0.3688,
+      "step": 38460
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.3013730031256647e-06,
+      "loss": 0.842,
+      "step": 38470
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.297600206403964e-06,
+      "loss": 0.5726,
+      "step": 38480
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.2938299452687213e-06,
+      "loss": 0.339,
+      "step": 38490
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.29006222155752e-06,
+      "loss": 0.5403,
+      "step": 38500
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.286297037106701e-06,
+      "loss": 0.1399,
+      "step": 38510
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.282534393751369e-06,
+      "loss": 0.6494,
+      "step": 38520
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.2787742933253936e-06,
+      "loss": 0.9051,
+      "step": 38530
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.2750167376613945e-06,
+      "loss": 0.98,
+      "step": 38540
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.2712617285907668e-06,
+      "loss": 0.5566,
+      "step": 38550
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.2675092679436525e-06,
+      "loss": 0.3889,
+      "step": 38560
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.2637593575489562e-06,
+      "loss": 0.2252,
+      "step": 38570
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.260011999234338e-06,
+      "loss": 0.8652,
+      "step": 38580
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.2562671948262134e-06,
+      "loss": 0.5661,
+      "step": 38590
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.2525249461497585e-06,
+      "loss": 0.4515,
+      "step": 38600
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.2487852550288986e-06,
+      "loss": 0.8839,
+      "step": 38610
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.2450481232863144e-06,
+      "loss": 0.8046,
+      "step": 38620
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.2413135527434324e-06,
+      "loss": 0.3477,
+      "step": 38630
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.2375815452204444e-06,
+      "loss": 0.1162,
+      "step": 38640
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.233852102536282e-06,
+      "loss": 0.2483,
+      "step": 38650
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.230125226508632e-06,
+      "loss": 0.2738,
+      "step": 38660
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.226400918953927e-06,
+      "loss": 0.2559,
+      "step": 38670
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.222679181687349e-06,
+      "loss": 0.2962,
+      "step": 38680
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.2189600165228265e-06,
+      "loss": 0.5954,
+      "step": 38690
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.2152434252730397e-06,
+      "loss": 0.546,
+      "step": 38700
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.2115294097494088e-06,
+      "loss": 0.5356,
+      "step": 38710
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.207817971762097e-06,
+      "loss": 0.2988,
+      "step": 38720
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.2041091131200133e-06,
+      "loss": 0.3633,
+      "step": 38730
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.2004028356308146e-06,
+      "loss": 0.7466,
+      "step": 38740
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.196699141100894e-06,
+      "loss": 0.8244,
+      "step": 38750
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.192998031335387e-06,
+      "loss": 0.3217,
+      "step": 38760
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.1892995081381687e-06,
+      "loss": 0.643,
+      "step": 38770
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.185603573311853e-06,
+      "loss": 0.9874,
+      "step": 38780
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.181910228657798e-06,
+      "loss": 0.6318,
+      "step": 38790
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.1782194759760947e-06,
+      "loss": 0.5392,
+      "step": 38800
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.1745313170655672e-06,
+      "loss": 0.3199,
+      "step": 38810
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.170845753723778e-06,
+      "loss": 0.5739,
+      "step": 38820
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.1671627877470326e-06,
+      "loss": 0.6472,
+      "step": 38830
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.16348242093036e-06,
+      "loss": 0.6141,
+      "step": 38840
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.1598046550675273e-06,
+      "loss": 0.4404,
+      "step": 38850
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.156129491951031e-06,
+      "loss": 0.6016,
+      "step": 38860
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.152456933372103e-06,
+      "loss": 0.3857,
+      "step": 38870
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.1487869811207007e-06,
+      "loss": 0.5235,
+      "step": 38880
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.1451196369855212e-06,
+      "loss": 0.7967,
+      "step": 38890
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.1414549027539765e-06,
+      "loss": 0.2731,
+      "step": 38900
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.1377927802122154e-06,
+      "loss": 0.4294,
+      "step": 38910
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.1341332711451104e-06,
+      "loss": 0.7456,
+      "step": 38920
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.130476377336266e-06,
+      "loss": 0.5222,
+      "step": 38930
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.1268221005680057e-06,
+      "loss": 0.5666,
+      "step": 38940
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.1231704426213795e-06,
+      "loss": 0.4093,
+      "step": 38950
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.1195214052761614e-06,
+      "loss": 0.7834,
+      "step": 38960
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.115874990310845e-06,
+      "loss": 0.4512,
+      "step": 38970
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.1122311995026557e-06,
+      "loss": 0.4492,
+      "step": 38980
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.108590034627526e-06,
+      "loss": 0.4992,
+      "step": 38990
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.104951497460118e-06,
+      "loss": 0.3091,
+      "step": 39000
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.1013155897738106e-06,
+      "loss": 0.411,
+      "step": 39010
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.097682313340698e-06,
+      "loss": 0.9523,
+      "step": 39020
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.0940516699316e-06,
+      "loss": 0.3464,
+      "step": 39030
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.0904236613160463e-06,
+      "loss": 0.4488,
+      "step": 39040
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.086798289262284e-06,
+      "loss": 0.4607,
+      "step": 39050
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.0831755555372753e-06,
+      "loss": 0.4364,
+      "step": 39060
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.0795554619066955e-06,
+      "loss": 0.5127,
+      "step": 39070
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.075938010134936e-06,
+      "loss": 0.5928,
+      "step": 39080
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.072323201985098e-06,
+      "loss": 0.431,
+      "step": 39090
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.068711039218996e-06,
+      "loss": 0.4404,
+      "step": 39100
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.06510152359715e-06,
+      "loss": 0.1239,
+      "step": 39110
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.0614946568788e-06,
+      "loss": 0.8922,
+      "step": 39120
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.057890440821886e-06,
+      "loss": 0.5615,
+      "step": 39130
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.0542888771830605e-06,
+      "loss": 0.6164,
+      "step": 39140
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.050689967717681e-06,
+      "loss": 0.3161,
+      "step": 39150
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.0470937141798125e-06,
+      "loss": 0.3372,
+      "step": 39160
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.0435001183222243e-06,
+      "loss": 0.6414,
+      "step": 39170
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.039909181896393e-06,
+      "loss": 0.9572,
+      "step": 39180
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.036320906652495e-06,
+      "loss": 0.6156,
+      "step": 39190
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.032735294339415e-06,
+      "loss": 0.1312,
+      "step": 39200
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.029152346704733e-06,
+      "loss": 0.865,
+      "step": 39210
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.0255720654947383e-06,
+      "loss": 0.4385,
+      "step": 39220
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.021994452454415e-06,
+      "loss": 0.4145,
+      "step": 39230
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.0184195093274506e-06,
+      "loss": 0.4148,
+      "step": 39240
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.0148472378562215e-06,
+      "loss": 0.4605,
+      "step": 39250
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 2.0112776397818165e-06,
+      "loss": 0.4646,
+      "step": 39260
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 2.007710716844012e-06,
+      "loss": 0.8151,
+      "step": 39270
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 2.0041464707812835e-06,
+      "loss": 0.3476,
+      "step": 39280
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 2.0005849033308014e-06,
+      "loss": 0.712,
+      "step": 39290
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 1.9970260162284266e-06,
+      "loss": 0.566,
+      "step": 39300
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 1.993469811208723e-06,
+      "loss": 0.6018,
+      "step": 39310
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 1.9899162900049393e-06,
+      "loss": 0.6175,
+      "step": 39320
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 1.9863654543490208e-06,
+      "loss": 0.6559,
+      "step": 39330
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 1.982817305971595e-06,
+      "loss": 0.336,
+      "step": 39340
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 1.9792718466019927e-06,
+      "loss": 0.7435,
+      "step": 39350
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 1.975729077968226e-06,
+      "loss": 0.3189,
+      "step": 39360
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 1.972189001796997e-06,
+      "loss": 0.7507,
+      "step": 39370
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 1.9686516198136965e-06,
+      "loss": 0.7727,
+      "step": 39380
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 1.9651169337424006e-06,
+      "loss": 0.3345,
+      "step": 39390
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 1.9615849453058707e-06,
+      "loss": 0.193,
+      "step": 39400
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 1.958055656225559e-06,
+      "loss": 0.8061,
+      "step": 39410
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 1.9545290682215987e-06,
+      "loss": 0.4554,
+      "step": 39420
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 1.9510051830128007e-06,
+      "loss": 0.2398,
+      "step": 39430
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 1.947484002316664e-06,
+      "loss": 0.1954,
+      "step": 39440
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 1.943965527849375e-06,
+      "loss": 0.4979,
+      "step": 39450
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 1.9404497613257914e-06,
+      "loss": 0.4122,
+      "step": 39460
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 1.9369367044594567e-06,
+      "loss": 0.501,
+      "step": 39470
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 1.9334263589625904e-06,
+      "loss": 0.4208,
+      "step": 39480
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 1.929918726546092e-06,
+      "loss": 0.4463,
+      "step": 39490
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 1.9264138089195424e-06,
+      "loss": 0.7467,
+      "step": 39500
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 1.922911607791196e-06,
+      "loss": 0.4804,
+      "step": 39510
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 1.9194121248679772e-06,
+      "loss": 0.7707,
+      "step": 39520
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 1.915915361855496e-06,
+      "loss": 0.4711,
+      "step": 39530
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 1.912421320458029e-06,
+      "loss": 0.5653,
+      "step": 39540
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 1.9089300023785338e-06,
+      "loss": 0.473,
+      "step": 39550
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 1.9054414093186343e-06,
+      "loss": 0.4003,
+      "step": 39560
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 1.9019555429786287e-06,
+      "loss": 0.8618,
+      "step": 39570
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 1.8984724050574857e-06,
+      "loss": 0.3744,
+      "step": 39580
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 1.8949919972528412e-06,
+      "loss": 0.3537,
+      "step": 39590
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 1.891514321261012e-06,
+      "loss": 0.6167,
+      "step": 39600
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 1.888039378776968e-06,
+      "loss": 0.5296,
+      "step": 39610
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 1.8845671714943557e-06,
+      "loss": 0.701,
+      "step": 39620
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 1.8810977011054845e-06,
+      "loss": 0.4249,
+      "step": 39630
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 1.877630969301337e-06,
+      "loss": 0.4951,
+      "step": 39640
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 1.8741669777715532e-06,
+      "loss": 0.4541,
+      "step": 39650
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 1.8707057282044417e-06,
+      "loss": 0.7534,
+      "step": 39660
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 1.867247222286973e-06,
+      "loss": 0.2315,
+      "step": 39670
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 1.86379146170478e-06,
+      "loss": 0.6699,
+      "step": 39680
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 1.8603384481421596e-06,
+      "loss": 0.4909,
+      "step": 39690
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 1.856888183282068e-06,
+      "loss": 0.4573,
+      "step": 39700
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 1.8534406688061228e-06,
+      "loss": 0.5582,
+      "step": 39710
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 1.8499959063946007e-06,
+      "loss": 0.4972,
+      "step": 39720
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 1.8465538977264343e-06,
+      "loss": 0.1831,
+      "step": 39730
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 1.843114644479221e-06,
+      "loss": 0.9039,
+      "step": 39740
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 1.83967814832921e-06,
+      "loss": 0.4582,
+      "step": 39750
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 1.8362444109513072e-06,
+      "loss": 0.4638,
+      "step": 39760
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 1.8328134340190738e-06,
+      "loss": 0.4641,
+      "step": 39770
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 1.8293852192047275e-06,
+      "loss": 0.7425,
+      "step": 39780
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 1.8259597681791374e-06,
+      "loss": 0.8255,
+      "step": 39790
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 1.822537082611828e-06,
+      "loss": 0.5125,
+      "step": 39800
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 1.819117164170973e-06,
+      "loss": 0.5191,
+      "step": 39810
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 1.8157000145233987e-06,
+      "loss": 0.4591,
+      "step": 39820
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 1.812285635334586e-06,
+      "loss": 0.2738,
+      "step": 39830
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 1.8088740282686587e-06,
+      "loss": 0.3304,
+      "step": 39840
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 1.8054651949883941e-06,
+      "loss": 0.6303,
+      "step": 39850
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 1.8020591371552175e-06,
+      "loss": 0.7347,
+      "step": 39860
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 1.7986558564291935e-06,
+      "loss": 0.5348,
+      "step": 39870
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 1.7952553544690462e-06,
+      "loss": 0.2213,
+      "step": 39880
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 1.791857632932138e-06,
+      "loss": 0.4525,
+      "step": 39890
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 1.7884626934744759e-06,
+      "loss": 0.3864,
+      "step": 39900
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 1.785070537750712e-06,
+      "loss": 0.5979,
+      "step": 39910
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 1.78168116741414e-06,
+      "loss": 0.3264,
+      "step": 39920
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 1.7782945841167034e-06,
+      "loss": 0.3334,
+      "step": 39930
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 1.7749107895089774e-06,
+      "loss": 0.6249,
+      "step": 39940
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 1.7715297852401858e-06,
+      "loss": 0.3204,
+      "step": 39950
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 1.7681515729581825e-06,
+      "loss": 0.2486,
+      "step": 39960
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 1.764776154309473e-06,
+      "loss": 0.3564,
+      "step": 39970
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 1.7614035309391938e-06,
+      "loss": 0.6261,
+      "step": 39980
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 1.758033704491121e-06,
+      "loss": 0.5859,
+      "step": 39990
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 1.7546666766076658e-06,
+      "loss": 1.0645,
+      "step": 40000
+    },
+    {
+      "epoch": 19.04,
+      "eval_accuracy": 0.8810526315789474,
+      "eval_f1": 0.8810526315789474,
+      "eval_loss": 0.7618152499198914,
+      "eval_runtime": 771.4019,
+      "eval_samples_per_second": 6.158,
+      "eval_steps_per_second": 1.54,
+      "step": 40000
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 1.751302448929877e-06,
+      "loss": 0.2598,
+      "step": 40010
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 1.7479410230974375e-06,
+      "loss": 0.4122,
+      "step": 40020
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 1.744582400748668e-06,
+      "loss": 0.283,
+      "step": 40030
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 1.741226583520521e-06,
+      "loss": 0.5017,
+      "step": 40040
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 1.7378735730485766e-06,
+      "loss": 0.9872,
+      "step": 40050
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 1.7345233709670513e-06,
+      "loss": 0.5128,
+      "step": 40060
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 1.7311759789087977e-06,
+      "loss": 0.4265,
+      "step": 40070
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 1.7278313985052915e-06,
+      "loss": 0.5808,
+      "step": 40080
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 1.7244896313866404e-06,
+      "loss": 0.3198,
+      "step": 40090
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 1.7211506791815815e-06,
+      "loss": 0.5577,
+      "step": 40100
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 1.7178145435174772e-06,
+      "loss": 0.5626,
+      "step": 40110
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 1.7144812260203234e-06,
+      "loss": 0.3418,
+      "step": 40120
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 1.7111507283147392e-06,
+      "loss": 0.2519,
+      "step": 40130
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 1.7078230520239644e-06,
+      "loss": 0.3026,
+      "step": 40140
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 1.7044981987698672e-06,
+      "loss": 0.3616,
+      "step": 40150
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 1.7011761701729456e-06,
+      "loss": 0.4037,
+      "step": 40160
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 1.6978569678523137e-06,
+      "loss": 0.2516,
+      "step": 40170
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 1.6945405934257107e-06,
+      "loss": 0.3255,
+      "step": 40180
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 1.6912270485094972e-06,
+      "loss": 0.3581,
+      "step": 40190
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 1.6879163347186536e-06,
+      "loss": 0.8509,
+      "step": 40200
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 1.6846084536667816e-06,
+      "loss": 0.6085,
+      "step": 40210
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 1.681303406966106e-06,
+      "loss": 0.3777,
+      "step": 40220
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 1.6780011962274622e-06,
+      "loss": 0.3812,
+      "step": 40230
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 1.6747018230603087e-06,
+      "loss": 0.9465,
+      "step": 40240
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 1.6714052890727182e-06,
+      "loss": 0.4388,
+      "step": 40250
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 1.668111595871385e-06,
+      "loss": 0.6455,
+      "step": 40260
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 1.6648207450616137e-06,
+      "loss": 0.2788,
+      "step": 40270
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 1.661532738247325e-06,
+      "loss": 0.7803,
+      "step": 40280
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 1.6582475770310526e-06,
+      "loss": 0.5762,
+      "step": 40290
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 1.6549652630139425e-06,
+      "loss": 0.5437,
+      "step": 40300
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 1.6516857977957617e-06,
+      "loss": 0.5942,
+      "step": 40310
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 1.6484091829748748e-06,
+      "loss": 0.7837,
+      "step": 40320
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 1.6451354201482654e-06,
+      "loss": 0.6525,
+      "step": 40330
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 1.6418645109115255e-06,
+      "loss": 0.5469,
+      "step": 40340
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 1.6385964568588587e-06,
+      "loss": 0.2445,
+      "step": 40350
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 1.6353312595830746e-06,
+      "loss": 0.3352,
+      "step": 40360
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 1.6320689206755892e-06,
+      "loss": 0.3698,
+      "step": 40370
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 1.6288094417264273e-06,
+      "loss": 0.4792,
+      "step": 40380
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 1.6255528243242198e-06,
+      "loss": 0.5886,
+      "step": 40390
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 1.622299070056202e-06,
+      "loss": 0.4715,
+      "step": 40400
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 1.6190481805082146e-06,
+      "loss": 0.5926,
+      "step": 40410
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 1.615800157264701e-06,
+      "loss": 0.5368,
+      "step": 40420
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 1.612555001908709e-06,
+      "loss": 0.2857,
+      "step": 40430
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 1.609312716021886e-06,
+      "loss": 0.6023,
+      "step": 40440
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 1.6060733011844852e-06,
+      "loss": 0.3041,
+      "step": 40450
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 1.602836758975358e-06,
+      "loss": 0.2443,
+      "step": 40460
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 1.5996030909719554e-06,
+      "loss": 0.4438,
+      "step": 40470
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 1.5963722987503259e-06,
+      "loss": 0.6301,
+      "step": 40480
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 1.5931443838851215e-06,
+      "loss": 0.3789,
+      "step": 40490
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 1.5899193479495858e-06,
+      "loss": 0.7228,
+      "step": 40500
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 1.5866971925155641e-06,
+      "loss": 0.3848,
+      "step": 40510
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 1.583477919153495e-06,
+      "loss": 0.2661,
+      "step": 40520
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 1.5802615294324137e-06,
+      "loss": 0.503,
+      "step": 40530
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 1.577048024919947e-06,
+      "loss": 0.3205,
+      "step": 40540
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 1.5738374071823219e-06,
+      "loss": 0.7299,
+      "step": 40550
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 1.5706296777843525e-06,
+      "loss": 0.4654,
+      "step": 40560
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 1.5674248382894501e-06,
+      "loss": 0.4685,
+      "step": 40570
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 1.5642228902596076e-06,
+      "loss": 0.5656,
+      "step": 40580
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 1.5610238352554214e-06,
+      "loss": 0.3782,
+      "step": 40590
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 1.5578276748360706e-06,
+      "loss": 0.275,
+      "step": 40600
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 1.5546344105593247e-06,
+      "loss": 0.6039,
+      "step": 40610
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 1.5514440439815419e-06,
+      "loss": 0.5187,
+      "step": 40620
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 1.5482565766576661e-06,
+      "loss": 0.4336,
+      "step": 40630
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 1.5450720101412346e-06,
+      "loss": 0.6509,
+      "step": 40640
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 1.5418903459843633e-06,
+      "loss": 0.2785,
+      "step": 40650
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 1.5387115857377602e-06,
+      "loss": 0.355,
+      "step": 40660
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 1.535535730950708e-06,
+      "loss": 0.5541,
+      "step": 40670
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 1.5323627831710843e-06,
+      "loss": 0.2017,
+      "step": 40680
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 1.5291927439453457e-06,
+      "loss": 0.5485,
+      "step": 40690
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 1.526025614818528e-06,
+      "loss": 0.6539,
+      "step": 40700
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 1.5228613973342541e-06,
+      "loss": 1.0763,
+      "step": 40710
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 1.5197000930347237e-06,
+      "loss": 0.388,
+      "step": 40720
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 1.5165417034607169e-06,
+      "loss": 0.534,
+      "step": 40730
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 1.5133862301515977e-06,
+      "loss": 0.3051,
+      "step": 40740
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 1.5102336746453054e-06,
+      "loss": 0.4692,
+      "step": 40750
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 1.5070840384783534e-06,
+      "loss": 0.5556,
+      "step": 40760
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 1.5039373231858372e-06,
+      "loss": 0.5806,
+      "step": 40770
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 1.5007935303014297e-06,
+      "loss": 0.6341,
+      "step": 40780
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 1.4976526613573772e-06,
+      "loss": 0.6556,
+      "step": 40790
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 1.4945147178844997e-06,
+      "loss": 0.4185,
+      "step": 40800
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 1.4913797014121927e-06,
+      "loss": 0.4563,
+      "step": 40810
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 1.4882476134684245e-06,
+      "loss": 0.5987,
+      "step": 40820
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 1.4851184555797397e-06,
+      "loss": 0.3656,
+      "step": 40830
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 1.481992229271252e-06,
+      "loss": 0.556,
+      "step": 40840
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 1.4788689360666423e-06,
+      "loss": 0.7573,
+      "step": 40850
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 1.4757485774881682e-06,
+      "loss": 0.2427,
+      "step": 40860
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 1.4726311550566512e-06,
+      "loss": 0.5579,
+      "step": 40870
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 1.4695166702914903e-06,
+      "loss": 0.4394,
+      "step": 40880
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 1.4664051247106443e-06,
+      "loss": 0.7392,
+      "step": 40890
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 1.4632965198306448e-06,
+      "loss": 0.6134,
+      "step": 40900
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 1.460190857166586e-06,
+      "loss": 0.6297,
+      "step": 40910
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 1.4570881382321298e-06,
+      "loss": 0.5465,
+      "step": 40920
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 1.453988364539508e-06,
+      "loss": 0.503,
+      "step": 40930
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 1.4508915375995082e-06,
+      "loss": 0.3181,
+      "step": 40940
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 1.4477976589214873e-06,
+      "loss": 0.6384,
+      "step": 40950
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 1.4447067300133621e-06,
+      "loss": 0.6659,
+      "step": 40960
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 1.4416187523816186e-06,
+      "loss": 0.605,
+      "step": 40970
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 1.4385337275312969e-06,
+      "loss": 0.5757,
+      "step": 40980
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 1.4354516569660009e-06,
+      "loss": 0.387,
+      "step": 40990
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 1.432372542187895e-06,
+      "loss": 0.5196,
+      "step": 41000
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 1.429296384697701e-06,
+      "loss": 0.3582,
+      "step": 41010
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 1.4262231859947016e-06,
+      "loss": 0.516,
+      "step": 41020
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 1.423152947576736e-06,
+      "loss": 0.3667,
+      "step": 41030
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 1.4200856709402014e-06,
+      "loss": 0.3674,
+      "step": 41040
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 1.4170213575800518e-06,
+      "loss": 0.7149,
+      "step": 41050
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 1.4139600089897933e-06,
+      "loss": 0.2172,
+      "step": 41060
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 1.4109016266614934e-06,
+      "loss": 0.5621,
+      "step": 41070
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 1.4078462120857708e-06,
+      "loss": 0.4305,
+      "step": 41080
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 1.4047937667517954e-06,
+      "loss": 0.4987,
+      "step": 41090
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 1.4017442921472933e-06,
+      "loss": 0.6915,
+      "step": 41100
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 1.3986977897585398e-06,
+      "loss": 0.5445,
+      "step": 41110
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 1.3956542610703654e-06,
+      "loss": 0.4355,
+      "step": 41120
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 1.392613707566147e-06,
+      "loss": 0.7685,
+      "step": 41130
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 1.3895761307278137e-06,
+      "loss": 0.4018,
+      "step": 41140
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 1.3865415320358427e-06,
+      "loss": 0.6319,
+      "step": 41150
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 1.3835099129692637e-06,
+      "loss": 0.6525,
+      "step": 41160
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 1.3804812750056494e-06,
+      "loss": 0.5524,
+      "step": 41170
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 1.3774556196211206e-06,
+      "loss": 0.2176,
+      "step": 41180
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 1.3744329482903447e-06,
+      "loss": 0.4774,
+      "step": 41190
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 1.3714132624865363e-06,
+      "loss": 0.8767,
+      "step": 41200
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 1.368396563681453e-06,
+      "loss": 0.9403,
+      "step": 41210
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 1.3653828533453968e-06,
+      "loss": 0.6222,
+      "step": 41220
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 1.3623721329472139e-06,
+      "loss": 0.8261,
+      "step": 41230
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 1.3593644039542919e-06,
+      "loss": 0.5185,
+      "step": 41240
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 1.3563596678325607e-06,
+      "loss": 0.6334,
+      "step": 41250
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 1.3533579260464956e-06,
+      "loss": 0.3334,
+      "step": 41260
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 1.3503591800591072e-06,
+      "loss": 0.3246,
+      "step": 41270
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 1.3473634313319497e-06,
+      "loss": 0.661,
+      "step": 41280
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 1.3443706813251082e-06,
+      "loss": 0.5927,
+      "step": 41290
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 1.3413809314972197e-06,
+      "loss": 0.2087,
+      "step": 41300
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 1.33839418330545e-06,
+      "loss": 0.3979,
+      "step": 41310
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 1.3354104382055022e-06,
+      "loss": 0.654,
+      "step": 41320
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 1.3324296976516195e-06,
+      "loss": 0.5012,
+      "step": 41330
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 1.3294519630965752e-06,
+      "loss": 0.5119,
+      "step": 41340
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 1.3264772359916855e-06,
+      "loss": 0.2278,
+      "step": 41350
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 1.3235055177867942e-06,
+      "loss": 0.5415,
+      "step": 41360
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 1.3205368099302818e-06,
+      "loss": 0.6114,
+      "step": 41370
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 1.3175711138690582e-06,
+      "loss": 0.3263,
+      "step": 41380
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 1.3146084310485668e-06,
+      "loss": 0.4803,
+      "step": 41390
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 1.3116487629127872e-06,
+      "loss": 0.3701,
+      "step": 41400
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 1.308692110904223e-06,
+      "loss": 0.5197,
+      "step": 41410
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 1.3057384764639107e-06,
+      "loss": 0.7112,
+      "step": 41420
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 1.3027878610314173e-06,
+      "loss": 0.5485,
+      "step": 41430
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 1.2998402660448339e-06,
+      "loss": 0.3948,
+      "step": 41440
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 1.2968956929407854e-06,
+      "loss": 0.9255,
+      "step": 41450
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 1.2939541431544224e-06,
+      "loss": 0.6046,
+      "step": 41460
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 1.2910156181194163e-06,
+      "loss": 0.5436,
+      "step": 41470
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 1.288080119267969e-06,
+      "loss": 0.6322,
+      "step": 41480
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 1.2851476480308092e-06,
+      "loss": 0.8129,
+      "step": 41490
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 1.282218205837188e-06,
+      "loss": 0.2926,
+      "step": 41500
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 1.2792917941148778e-06,
+      "loss": 0.4101,
+      "step": 41510
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 1.2763684142901778e-06,
+      "loss": 0.6326,
+      "step": 41520
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 1.2734480677879066e-06,
+      "loss": 0.7398,
+      "step": 41530
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 1.270530756031404e-06,
+      "loss": 0.5005,
+      "step": 41540
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 1.2676164804425391e-06,
+      "loss": 0.3263,
+      "step": 41550
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 1.2647052424416878e-06,
+      "loss": 0.3796,
+      "step": 41560
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 1.2617970434477532e-06,
+      "loss": 0.491,
+      "step": 41570
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 1.258891884878156e-06,
+      "loss": 0.426,
+      "step": 41580
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 1.2559897681488377e-06,
+      "loss": 0.712,
+      "step": 41590
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 1.2530906946742544e-06,
+      "loss": 0.29,
+      "step": 41600
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 1.2501946658673771e-06,
+      "loss": 0.5593,
+      "step": 41610
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 1.2473016831396962e-06,
+      "loss": 0.5783,
+      "step": 41620
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 1.2444117479012166e-06,
+      "loss": 0.5131,
+      "step": 41630
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 1.2415248615604577e-06,
+      "loss": 0.4384,
+      "step": 41640
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 1.2386410255244518e-06,
+      "loss": 0.307,
+      "step": 41650
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 1.235760241198747e-06,
+      "loss": 0.4889,
+      "step": 41660
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 1.2328825099873995e-06,
+      "loss": 0.3653,
+      "step": 41670
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 1.2300078332929845e-06,
+      "loss": 0.583,
+      "step": 41680
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 1.2271362125165825e-06,
+      "loss": 0.5165,
+      "step": 41690
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 1.224267649057788e-06,
+      "loss": 0.6934,
+      "step": 41700
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 1.2214021443147022e-06,
+      "loss": 0.4905,
+      "step": 41710
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 1.2185396996839376e-06,
+      "loss": 0.6703,
+      "step": 41720
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 1.2156803165606156e-06,
+      "loss": 0.7073,
+      "step": 41730
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 1.2128239963383647e-06,
+      "loss": 0.314,
+      "step": 41740
+    },
+    {
+      "epoch": 20.04,
+      "learning_rate": 1.2099707404093204e-06,
+      "loss": 0.6241,
+      "step": 41750
+    },
+    {
+      "epoch": 20.04,
+      "learning_rate": 1.2071205501641256e-06,
+      "loss": 0.6667,
+      "step": 41760
+    },
+    {
+      "epoch": 20.04,
+      "learning_rate": 1.2042734269919254e-06,
+      "loss": 0.3184,
+      "step": 41770
+    },
+    {
+      "epoch": 20.04,
+      "learning_rate": 1.2014293722803782e-06,
+      "loss": 0.469,
+      "step": 41780
+    },
+    {
+      "epoch": 20.04,
+      "learning_rate": 1.1985883874156386e-06,
+      "loss": 0.161,
+      "step": 41790
+    },
+    {
+      "epoch": 20.04,
+      "learning_rate": 1.1957504737823682e-06,
+      "loss": 0.4946,
+      "step": 41800
+    },
+    {
+      "epoch": 20.04,
+      "learning_rate": 1.1929156327637321e-06,
+      "loss": 0.3183,
+      "step": 41810
+    },
+    {
+      "epoch": 20.04,
+      "learning_rate": 1.1900838657413978e-06,
+      "loss": 0.1142,
+      "step": 41820
+    },
+    {
+      "epoch": 20.04,
+      "learning_rate": 1.187255174095533e-06,
+      "loss": 0.5159,
+      "step": 41830
+    },
+    {
+      "epoch": 20.04,
+      "learning_rate": 1.1844295592048086e-06,
+      "loss": 0.1947,
+      "step": 41840
+    },
+    {
+      "epoch": 20.04,
+      "learning_rate": 1.181607022446394e-06,
+      "loss": 0.3758,
+      "step": 41850
+    },
+    {
+      "epoch": 20.04,
+      "learning_rate": 1.1787875651959606e-06,
+      "loss": 0.4884,
+      "step": 41860
+    },
+    {
+      "epoch": 20.04,
+      "learning_rate": 1.175971188827675e-06,
+      "loss": 0.6049,
+      "step": 41870
+    },
+    {
+      "epoch": 20.04,
+      "learning_rate": 1.173157894714209e-06,
+      "loss": 0.6191,
+      "step": 41880
+    },
+    {
+      "epoch": 20.04,
+      "learning_rate": 1.1703476842267253e-06,
+      "loss": 0.4537,
+      "step": 41890
+    },
+    {
+      "epoch": 20.04,
+      "learning_rate": 1.1675405587348885e-06,
+      "loss": 0.6927,
+      "step": 41900
+    },
+    {
+      "epoch": 20.04,
+      "learning_rate": 1.1647365196068524e-06,
+      "loss": 0.1467,
+      "step": 41910
+    },
+    {
+      "epoch": 20.04,
+      "learning_rate": 1.1619355682092774e-06,
+      "loss": 0.4122,
+      "step": 41920
+    },
+    {
+      "epoch": 20.04,
+      "learning_rate": 1.1591377059073103e-06,
+      "loss": 0.3217,
+      "step": 41930
+    },
+    {
+      "epoch": 20.04,
+      "learning_rate": 1.1563429340645955e-06,
+      "loss": 0.8771,
+      "step": 41940
+    },
+    {
+      "epoch": 20.04,
+      "learning_rate": 1.1535512540432707e-06,
+      "loss": 0.6063,
+      "step": 41950
+    },
+    {
+      "epoch": 20.04,
+      "learning_rate": 1.1507626672039656e-06,
+      "loss": 0.5014,
+      "step": 41960
+    },
+    {
+      "epoch": 20.04,
+      "learning_rate": 1.1479771749058071e-06,
+      "loss": 0.5023,
+      "step": 41970
+    },
+    {
+      "epoch": 20.04,
+      "learning_rate": 1.1451947785064086e-06,
+      "loss": 0.7961,
+      "step": 41980
+    },
+    {
+      "epoch": 20.04,
+      "learning_rate": 1.1424154793618775e-06,
+      "loss": 0.3274,
+      "step": 41990
+    },
+    {
+      "epoch": 20.04,
+      "learning_rate": 1.1396392788268054e-06,
+      "loss": 0.4949,
+      "step": 42000
+    },
+    {
+      "epoch": 20.04,
+      "eval_accuracy": 0.8747368421052631,
+      "eval_f1": 0.8747368421052631,
+      "eval_loss": 0.7881549000740051,
+      "eval_runtime": 742.269,
+      "eval_samples_per_second": 6.399,
+      "eval_steps_per_second": 1.6,
+      "step": 42000
+    },
+    {
+      "epoch": 21.0,
+      "learning_rate": 1.1368661782542842e-06,
+      "loss": 0.5779,
+      "step": 42010
+    },
+    {
+      "epoch": 21.0,
+      "learning_rate": 1.1340961789958867e-06,
+      "loss": 0.7587,
+      "step": 42020
+    },
+    {
+      "epoch": 21.0,
+      "learning_rate": 1.1313292824016783e-06,
+      "loss": 0.7537,
+      "step": 42030
+    },
+    {
+      "epoch": 21.0,
+      "learning_rate": 1.1285654898202107e-06,
+      "loss": 0.2199,
+      "step": 42040
+    },
+    {
+      "epoch": 21.0,
+      "learning_rate": 1.1258048025985219e-06,
+      "loss": 0.4507,
+      "step": 42050
+    },
+    {
+      "epoch": 21.0,
+      "learning_rate": 1.123047222082135e-06,
+      "loss": 0.3394,
+      "step": 42060
+    },
+    {
+      "epoch": 21.0,
+      "learning_rate": 1.120292749615067e-06,
+      "loss": 0.812,
+      "step": 42070
+    },
+    {
+      "epoch": 21.0,
+      "learning_rate": 1.117541386539812e-06,
+      "loss": 0.4893,
+      "step": 42080
+    },
+    {
+      "epoch": 21.0,
+      "learning_rate": 1.1147931341973493e-06,
+      "loss": 0.7984,
+      "step": 42090
+    },
+    {
+      "epoch": 21.0,
+      "learning_rate": 1.1120479939271424e-06,
+      "loss": 0.3631,
+      "step": 42100
+    },
+    {
+      "epoch": 21.0,
+      "learning_rate": 1.1093059670671444e-06,
+      "loss": 0.2635,
+      "step": 42110
+    },
+    {
+      "epoch": 21.0,
+      "learning_rate": 1.1065670549537818e-06,
+      "loss": 0.1696,
+      "step": 42120
+    },
+    {
+      "epoch": 21.0,
+      "learning_rate": 1.1038312589219687e-06,
+      "loss": 0.4704,
+      "step": 42130
+    },
+    {
+      "epoch": 21.0,
+      "learning_rate": 1.1010985803050986e-06,
+      "loss": 0.4442,
+      "step": 42140
+    },
+    {
+      "epoch": 21.0,
+      "learning_rate": 1.0983690204350433e-06,
+      "loss": 0.1498,
+      "step": 42150
+    },
+    {
+      "epoch": 21.0,
+      "learning_rate": 1.0956425806421622e-06,
+      "loss": 0.4469,
+      "step": 42160
+    },
+    {
+      "epoch": 21.0,
+      "learning_rate": 1.0929192622552872e-06,
+      "loss": 0.3605,
+      "step": 42170
+    },
+    {
+      "epoch": 21.0,
+      "learning_rate": 1.0901990666017283e-06,
+      "loss": 1.0167,
+      "step": 42180
+    },
+    {
+      "epoch": 21.0,
+      "learning_rate": 1.087481995007276e-06,
+      "loss": 0.9402,
+      "step": 42190
+    },
+    {
+      "epoch": 21.0,
+      "learning_rate": 1.0847680487962005e-06,
+      "loss": 0.5113,
+      "step": 42200
+    },
+    {
+      "epoch": 21.0,
+      "learning_rate": 1.0820572292912442e-06,
+      "loss": 0.2354,
+      "step": 42210
+    },
+    {
+      "epoch": 21.0,
+      "learning_rate": 1.0793495378136291e-06,
+      "loss": 0.1514,
+      "step": 42220
+    },
+    {
+      "epoch": 21.0,
+      "learning_rate": 1.0766449756830507e-06,
+      "loss": 0.4024,
+      "step": 42230
+    },
+    {
+      "epoch": 21.0,
+      "learning_rate": 1.0739435442176783e-06,
+      "loss": 0.1413,
+      "step": 42240
+    },
+    {
+      "epoch": 21.0,
+      "learning_rate": 1.0712452447341584e-06,
+      "loss": 0.4313,
+      "step": 42250
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 1.0685500785476093e-06,
+      "loss": 0.3361,
+      "step": 42260
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 1.0658580469716214e-06,
+      "loss": 0.5155,
+      "step": 42270
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 1.0631691513182585e-06,
+      "loss": 0.7166,
+      "step": 42280
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 1.0604833928980537e-06,
+      "loss": 0.5803,
+      "step": 42290
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 1.0578007730200167e-06,
+      "loss": 0.2315,
+      "step": 42300
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 1.0551212929916225e-06,
+      "loss": 0.4389,
+      "step": 42310
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 1.0524449541188174e-06,
+      "loss": 0.6883,
+      "step": 42320
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 1.0497717577060153e-06,
+      "loss": 0.5107,
+      "step": 42330
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 1.0471017050560999e-06,
+      "loss": 0.5452,
+      "step": 42340
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 1.0444347974704235e-06,
+      "loss": 0.2522,
+      "step": 42350
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 1.0417710362488031e-06,
+      "loss": 0.2507,
+      "step": 42360
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 1.039110422689525e-06,
+      "loss": 0.2414,
+      "step": 42370
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 1.036452958089338e-06,
+      "loss": 1.171,
+      "step": 42380
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 1.0337986437434584e-06,
+      "loss": 0.3574,
+      "step": 42390
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 1.031147480945569e-06,
+      "loss": 0.3141,
+      "step": 42400
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 1.0284994709878125e-06,
+      "loss": 0.8145,
+      "step": 42410
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 1.0258546151607978e-06,
+      "loss": 0.5917,
+      "step": 42420
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 1.0232129147535943e-06,
+      "loss": 0.5349,
+      "step": 42430
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 1.0205743710537355e-06,
+      "loss": 0.5481,
+      "step": 42440
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 1.0179389853472152e-06,
+      "loss": 0.3898,
+      "step": 42450
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 1.0153067589184893e-06,
+      "loss": 0.4142,
+      "step": 42460
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 1.0126776930504706e-06,
+      "loss": 0.4881,
+      "step": 42470
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 1.010051789024535e-06,
+      "loss": 0.4907,
+      "step": 42480
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 1.0074290481205182e-06,
+      "loss": 0.5862,
+      "step": 42490
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 1.0048094716167097e-06,
+      "loss": 0.2149,
+      "step": 42500
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 1.0021930607898611e-06,
+      "loss": 0.5522,
+      "step": 42510
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 9.995798169151788e-07,
+      "loss": 0.5393,
+      "step": 42520
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 9.969697412663247e-07,
+      "loss": 0.4814,
+      "step": 42530
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 9.9436283511542e-07,
+      "loss": 0.364,
+      "step": 42540
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 9.917590997330377e-07,
+      "loss": 0.5711,
+      "step": 42550
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 9.891585363882066e-07,
+      "loss": 0.5678,
+      "step": 42560
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 9.865611463484108e-07,
+      "loss": 0.1107,
+      "step": 42570
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 9.839669308795846e-07,
+      "loss": 0.6555,
+      "step": 42580
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 9.813758912461204e-07,
+      "loss": 0.987,
+      "step": 42590
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 9.787880287108574e-07,
+      "loss": 0.3986,
+      "step": 42600
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 9.762033445350901e-07,
+      "loss": 0.498,
+      "step": 42610
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 9.736218399785573e-07,
+      "loss": 0.5059,
+      "step": 42620
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 9.710435162994585e-07,
+      "loss": 0.4711,
+      "step": 42630
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 9.68468374754435e-07,
+      "loss": 0.2611,
+      "step": 42640
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 9.658964165985798e-07,
+      "loss": 0.4665,
+      "step": 42650
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 9.63327643085434e-07,
+      "loss": 1.166,
+      "step": 42660
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 9.607620554669846e-07,
+      "loss": 0.3244,
+      "step": 42670
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 9.581996549936721e-07,
+      "loss": 0.8249,
+      "step": 42680
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 9.55640442914376e-07,
+      "loss": 0.5912,
+      "step": 42690
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 9.530844204764286e-07,
+      "loss": 0.4654,
+      "step": 42700
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 9.505315889256005e-07,
+      "loss": 0.5017,
+      "step": 42710
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 9.479819495061102e-07,
+      "loss": 0.2483,
+      "step": 42720
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 9.454355034606241e-07,
+      "loss": 0.2872,
+      "step": 42730
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 9.428922520302479e-07,
+      "loss": 0.3335,
+      "step": 42740
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 9.40352196454532e-07,
+      "loss": 0.6583,
+      "step": 42750
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 9.378153379714682e-07,
+      "loss": 0.6831,
+      "step": 42760
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 9.352816778174878e-07,
+      "loss": 0.1119,
+      "step": 42770
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 9.327512172274711e-07,
+      "loss": 0.6276,
+      "step": 42780
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 9.302239574347323e-07,
+      "loss": 0.1037,
+      "step": 42790
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 9.276998996710248e-07,
+      "loss": 0.432,
+      "step": 42800
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 9.251790451665426e-07,
+      "loss": 0.5301,
+      "step": 42810
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 9.226613951499246e-07,
+      "loss": 0.2378,
+      "step": 42820
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 9.201469508482394e-07,
+      "loss": 0.6031,
+      "step": 42830
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 9.176357134869981e-07,
+      "loss": 0.5935,
+      "step": 42840
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 9.15127684290146e-07,
+      "loss": 0.752,
+      "step": 42850
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 9.126228644800669e-07,
+      "loss": 0.4204,
+      "step": 42860
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 9.101212552775793e-07,
+      "loss": 0.1586,
+      "step": 42870
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 9.076228579019377e-07,
+      "loss": 0.7899,
+      "step": 42880
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 9.051276735708292e-07,
+      "loss": 0.1739,
+      "step": 42890
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 9.026357035003774e-07,
+      "loss": 0.1888,
+      "step": 42900
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 9.001469489051367e-07,
+      "loss": 0.1448,
+      "step": 42910
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 8.976614109980985e-07,
+      "loss": 0.4078,
+      "step": 42920
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 8.951790909906829e-07,
+      "loss": 0.628,
+      "step": 42930
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 8.926999900927413e-07,
+      "loss": 0.6041,
+      "step": 42940
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 8.902241095125588e-07,
+      "loss": 0.4989,
+      "step": 42950
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 8.877514504568484e-07,
+      "loss": 0.4466,
+      "step": 42960
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 8.852820141307555e-07,
+      "loss": 0.4655,
+      "step": 42970
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 8.828158017378512e-07,
+      "loss": 0.7187,
+      "step": 42980
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 8.803528144801393e-07,
+      "loss": 0.8563,
+      "step": 42990
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 8.778930535580476e-07,
+      "loss": 0.516,
+      "step": 43000
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 8.754365201704373e-07,
+      "loss": 0.6609,
+      "step": 43010
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 8.729832155145897e-07,
+      "loss": 0.4027,
+      "step": 43020
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 8.705331407862185e-07,
+      "loss": 0.2548,
+      "step": 43030
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 8.680862971794575e-07,
+      "loss": 0.5615,
+      "step": 43040
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 8.656426858868705e-07,
+      "loss": 0.4992,
+      "step": 43050
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 8.632023080994436e-07,
+      "loss": 0.388,
+      "step": 43060
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 8.607651650065865e-07,
+      "loss": 0.5331,
+      "step": 43070
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 8.583312577961339e-07,
+      "loss": 0.4323,
+      "step": 43080
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 8.559005876543427e-07,
+      "loss": 0.4964,
+      "step": 43090
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 8.534731557658895e-07,
+      "loss": 0.3992,
+      "step": 43100
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 8.510489633138788e-07,
+      "loss": 0.5065,
+      "step": 43110
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 8.486280114798311e-07,
+      "loss": 0.5718,
+      "step": 43120
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 8.462103014436887e-07,
+      "loss": 0.3075,
+      "step": 43130
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 8.437958343838146e-07,
+      "loss": 0.3258,
+      "step": 43140
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 8.413846114769907e-07,
+      "loss": 0.5287,
+      "step": 43150
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 8.389766338984165e-07,
+      "loss": 0.8632,
+      "step": 43160
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 8.365719028217128e-07,
+      "loss": 0.372,
+      "step": 43170
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 8.341704194189148e-07,
+      "loss": 0.4632,
+      "step": 43180
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 8.317721848604754e-07,
+      "loss": 0.2295,
+      "step": 43190
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 8.293772003152686e-07,
+      "loss": 0.5518,
+      "step": 43200
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 8.269854669505783e-07,
+      "loss": 0.6363,
+      "step": 43210
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 8.245969859321062e-07,
+      "loss": 0.4671,
+      "step": 43220
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 8.222117584239719e-07,
+      "loss": 0.5977,
+      "step": 43230
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 8.198297855887005e-07,
+      "loss": 0.6138,
+      "step": 43240
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 8.174510685872417e-07,
+      "loss": 0.6577,
+      "step": 43250
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 8.150756085789512e-07,
+      "loss": 0.4194,
+      "step": 43260
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 8.127034067215999e-07,
+      "loss": 0.3516,
+      "step": 43270
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 8.103344641713695e-07,
+      "loss": 0.3407,
+      "step": 43280
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 8.079687820828538e-07,
+      "loss": 0.5712,
+      "step": 43290
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 8.056063616090581e-07,
+      "loss": 0.6257,
+      "step": 43300
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 8.032472039013988e-07,
+      "loss": 0.3842,
+      "step": 43310
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 8.008913101096996e-07,
+      "loss": 0.684,
+      "step": 43320
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 7.985386813821918e-07,
+      "loss": 0.7516,
+      "step": 43330
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 7.961893188655217e-07,
+      "loss": 0.3816,
+      "step": 43340
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 7.938432237047392e-07,
+      "loss": 0.5124,
+      "step": 43350
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 7.915003970433018e-07,
+      "loss": 0.6431,
+      "step": 43360
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 7.891608400230749e-07,
+      "loss": 0.4608,
+      "step": 43370
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 7.868245537843311e-07,
+      "loss": 0.5772,
+      "step": 43380
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 7.844915394657445e-07,
+      "loss": 0.6119,
+      "step": 43390
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 7.821617982044033e-07,
+      "loss": 0.1979,
+      "step": 43400
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 7.798353311357931e-07,
+      "loss": 0.4184,
+      "step": 43410
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 7.775121393938045e-07,
+      "loss": 0.6045,
+      "step": 43420
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 7.751922241107309e-07,
+      "loss": 0.6312,
+      "step": 43430
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 7.728755864172754e-07,
+      "loss": 0.3845,
+      "step": 43440
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 7.705622274425372e-07,
+      "loss": 0.2585,
+      "step": 43450
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 7.682521483140187e-07,
+      "loss": 0.874,
+      "step": 43460
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 7.659453501576258e-07,
+      "loss": 0.1983,
+      "step": 43470
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 7.636418340976609e-07,
+      "loss": 0.5186,
+      "step": 43480
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 7.613416012568349e-07,
+      "loss": 0.5401,
+      "step": 43490
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 7.59044652756249e-07,
+      "loss": 0.3666,
+      "step": 43500
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 7.567509897154088e-07,
+      "loss": 0.2822,
+      "step": 43510
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 7.54460613252217e-07,
+      "loss": 0.3534,
+      "step": 43520
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 7.52173524482978e-07,
+      "loss": 0.7955,
+      "step": 43530
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 7.498897245223904e-07,
+      "loss": 0.6602,
+      "step": 43540
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 7.476092144835487e-07,
+      "loss": 0.2675,
+      "step": 43550
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 7.453319954779478e-07,
+      "loss": 0.526,
+      "step": 43560
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 7.430580686154751e-07,
+      "loss": 0.5293,
+      "step": 43570
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 7.407874350044155e-07,
+      "loss": 0.8674,
+      "step": 43580
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 7.38520095751449e-07,
+      "loss": 0.8319,
+      "step": 43590
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 7.362560519616474e-07,
+      "loss": 0.3474,
+      "step": 43600
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 7.339953047384795e-07,
+      "loss": 0.7456,
+      "step": 43610
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 7.31737855183805e-07,
+      "loss": 0.3838,
+      "step": 43620
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 7.294837043978786e-07,
+      "loss": 0.359,
+      "step": 43630
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 7.272328534793465e-07,
+      "loss": 0.4701,
+      "step": 43640
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 7.249853035252437e-07,
+      "loss": 0.468,
+      "step": 43650
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 7.227410556310016e-07,
+      "loss": 0.468,
+      "step": 43660
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 7.205001108904375e-07,
+      "loss": 0.4356,
+      "step": 43670
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 7.182624703957603e-07,
+      "loss": 0.1461,
+      "step": 43680
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 7.1602813523757e-07,
+      "loss": 0.5284,
+      "step": 43690
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 7.137971065048549e-07,
+      "loss": 0.6941,
+      "step": 43700
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 7.115693852849891e-07,
+      "loss": 0.7175,
+      "step": 43710
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 7.093449726637369e-07,
+      "loss": 0.4247,
+      "step": 43720
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 7.071238697252533e-07,
+      "loss": 0.3809,
+      "step": 43730
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 7.049060775520741e-07,
+      "loss": 0.6679,
+      "step": 43740
+    },
+    {
+      "epoch": 21.04,
+      "learning_rate": 7.026915972251254e-07,
+      "loss": 0.5361,
+      "step": 43750
+    },
+    {
+      "epoch": 21.04,
+      "learning_rate": 7.004804298237175e-07,
+      "loss": 0.7257,
+      "step": 43760
+    },
+    {
+      "epoch": 21.04,
+      "learning_rate": 6.982725764255463e-07,
+      "loss": 0.3651,
+      "step": 43770
+    },
+    {
+      "epoch": 21.04,
+      "learning_rate": 6.960680381066936e-07,
+      "loss": 0.4422,
+      "step": 43780
+    },
+    {
+      "epoch": 21.04,
+      "learning_rate": 6.938668159416234e-07,
+      "loss": 0.1049,
+      "step": 43790
+    },
+    {
+      "epoch": 21.04,
+      "learning_rate": 6.916689110031857e-07,
+      "loss": 0.4155,
+      "step": 43800
+    },
+    {
+      "epoch": 21.04,
+      "learning_rate": 6.894743243626112e-07,
+      "loss": 0.4926,
+      "step": 43810
+    },
+    {
+      "epoch": 21.04,
+      "learning_rate": 6.872830570895167e-07,
+      "loss": 0.3949,
+      "step": 43820
+    },
+    {
+      "epoch": 21.04,
+      "learning_rate": 6.850951102518979e-07,
+      "loss": 0.2546,
+      "step": 43830
+    },
+    {
+      "epoch": 21.04,
+      "learning_rate": 6.829104849161344e-07,
+      "loss": 0.4445,
+      "step": 43840
+    },
+    {
+      "epoch": 21.04,
+      "learning_rate": 6.80729182146986e-07,
+      "loss": 0.5695,
+      "step": 43850
+    },
+    {
+      "epoch": 21.04,
+      "learning_rate": 6.785512030075925e-07,
+      "loss": 0.6643,
+      "step": 43860
+    },
+    {
+      "epoch": 21.04,
+      "learning_rate": 6.763765485594744e-07,
+      "loss": 0.7855,
+      "step": 43870
+    },
+    {
+      "epoch": 21.04,
+      "learning_rate": 6.742052198625326e-07,
+      "loss": 0.5017,
+      "step": 43880
+    },
+    {
+      "epoch": 21.04,
+      "learning_rate": 6.720372179750461e-07,
+      "loss": 0.3665,
+      "step": 43890
+    },
+    {
+      "epoch": 21.04,
+      "learning_rate": 6.698725439536723e-07,
+      "loss": 0.4931,
+      "step": 43900
+    },
+    {
+      "epoch": 21.04,
+      "learning_rate": 6.677111988534465e-07,
+      "loss": 0.3041,
+      "step": 43910
+    },
+    {
+      "epoch": 21.04,
+      "learning_rate": 6.655531837277834e-07,
+      "loss": 0.2887,
+      "step": 43920
+    },
+    {
+      "epoch": 21.04,
+      "learning_rate": 6.633984996284725e-07,
+      "loss": 0.2932,
+      "step": 43930
+    },
+    {
+      "epoch": 21.04,
+      "learning_rate": 6.61247147605683e-07,
+      "loss": 0.4111,
+      "step": 43940
+    },
+    {
+      "epoch": 21.04,
+      "learning_rate": 6.590991287079529e-07,
+      "loss": 0.4439,
+      "step": 43950
+    },
+    {
+      "epoch": 21.04,
+      "learning_rate": 6.569544439822045e-07,
+      "loss": 0.604,
+      "step": 43960
+    },
+    {
+      "epoch": 21.04,
+      "learning_rate": 6.548130944737294e-07,
+      "loss": 0.4894,
+      "step": 43970
+    },
+    {
+      "epoch": 21.04,
+      "learning_rate": 6.526750812261958e-07,
+      "loss": 0.3146,
+      "step": 43980
+    },
+    {
+      "epoch": 21.04,
+      "learning_rate": 6.505404052816455e-07,
+      "loss": 0.2673,
+      "step": 43990
+    },
+    {
+      "epoch": 21.04,
+      "learning_rate": 6.484090676804927e-07,
+      "loss": 0.6131,
+      "step": 44000
+    },
+    {
+      "epoch": 21.04,
+      "eval_accuracy": 0.8705263157894737,
+      "eval_f1": 0.8705263157894737,
+      "eval_loss": 0.796357274055481,
+      "eval_runtime": 747.1435,
+      "eval_samples_per_second": 6.358,
+      "eval_steps_per_second": 1.59,
+      "step": 44000
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 6.462810694615273e-07,
+      "loss": 0.6104,
+      "step": 44010
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 6.441564116619089e-07,
+      "loss": 0.671,
+      "step": 44020
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 6.420350953171708e-07,
+      "loss": 0.466,
+      "step": 44030
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 6.399171214612126e-07,
+      "loss": 0.3888,
+      "step": 44040
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 6.378024911263144e-07,
+      "loss": 0.428,
+      "step": 44050
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 6.356912053431185e-07,
+      "loss": 0.177,
+      "step": 44060
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 6.33583265140641e-07,
+      "loss": 0.6163,
+      "step": 44070
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 6.314786715462656e-07,
+      "loss": 0.3802,
+      "step": 44080
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 6.293774255857463e-07,
+      "loss": 0.5927,
+      "step": 44090
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 6.272795282832028e-07,
+      "loss": 0.2755,
+      "step": 44100
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 6.25184980661129e-07,
+      "loss": 0.4401,
+      "step": 44110
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 6.230937837403783e-07,
+      "loss": 0.3299,
+      "step": 44120
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 6.210059385401754e-07,
+      "loss": 0.2617,
+      "step": 44130
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 6.189214460781104e-07,
+      "loss": 0.3985,
+      "step": 44140
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 6.168403073701415e-07,
+      "loss": 0.512,
+      "step": 44150
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 6.14762523430589e-07,
+      "loss": 0.4434,
+      "step": 44160
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 6.126880952721403e-07,
+      "loss": 0.4754,
+      "step": 44170
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 6.10617023905845e-07,
+      "loss": 0.5834,
+      "step": 44180
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 6.085493103411183e-07,
+      "loss": 0.2878,
+      "step": 44190
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 6.064849555857421e-07,
+      "loss": 0.3629,
+      "step": 44200
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 6.044239606458543e-07,
+      "loss": 0.5338,
+      "step": 44210
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 6.023663265259597e-07,
+      "loss": 0.5599,
+      "step": 44220
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 6.003120542289234e-07,
+      "loss": 0.5573,
+      "step": 44230
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 5.982611447559722e-07,
+      "loss": 0.3172,
+      "step": 44240
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 5.962135991066972e-07,
+      "loss": 0.7887,
+      "step": 44250
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 5.941694182790461e-07,
+      "loss": 0.4521,
+      "step": 44260
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 5.921286032693277e-07,
+      "loss": 0.5678,
+      "step": 44270
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 5.900911550722105e-07,
+      "loss": 0.3744,
+      "step": 44280
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 5.88057074680721e-07,
+      "loss": 0.576,
+      "step": 44290
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 5.86026363086247e-07,
+      "loss": 0.5443,
+      "step": 44300
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 5.83999021278531e-07,
+      "loss": 0.8581,
+      "step": 44310
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 5.819750502456758e-07,
+      "loss": 0.2708,
+      "step": 44320
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 5.799544509741384e-07,
+      "loss": 0.7194,
+      "step": 44330
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 5.779372244487374e-07,
+      "loss": 0.4304,
+      "step": 44340
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 5.759233716526433e-07,
+      "loss": 0.3599,
+      "step": 44350
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 5.739128935673832e-07,
+      "loss": 0.259,
+      "step": 44360
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 5.719057911728392e-07,
+      "loss": 0.2699,
+      "step": 44370
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 5.699020654472497e-07,
+      "loss": 0.5559,
+      "step": 44380
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 5.679017173672052e-07,
+      "loss": 0.3535,
+      "step": 44390
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 5.659047479076524e-07,
+      "loss": 0.3878,
+      "step": 44400
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 5.639111580418885e-07,
+      "loss": 0.2821,
+      "step": 44410
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 5.619209487415655e-07,
+      "loss": 0.1983,
+      "step": 44420
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 5.599341209766867e-07,
+      "loss": 0.5107,
+      "step": 44430
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 5.579506757156091e-07,
+      "loss": 0.5185,
+      "step": 44440
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 5.559706139250395e-07,
+      "loss": 0.4973,
+      "step": 44450
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 5.539939365700358e-07,
+      "loss": 0.3193,
+      "step": 44460
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 5.520206446140058e-07,
+      "loss": 0.495,
+      "step": 44470
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 5.500507390187093e-07,
+      "loss": 0.6467,
+      "step": 44480
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 5.480842207442533e-07,
+      "loss": 0.6714,
+      "step": 44490
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 5.461210907490952e-07,
+      "loss": 0.3219,
+      "step": 44500
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 5.441613499900405e-07,
+      "loss": 0.4564,
+      "step": 44510
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 5.422049994222427e-07,
+      "loss": 0.4278,
+      "step": 44520
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 5.40252039999205e-07,
+      "loss": 0.3553,
+      "step": 44530
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 5.38302472672775e-07,
+      "loss": 0.5166,
+      "step": 44540
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 5.363562983931486e-07,
+      "loss": 0.318,
+      "step": 44550
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 5.344135181088677e-07,
+      "loss": 0.6389,
+      "step": 44560
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 5.324741327668168e-07,
+      "loss": 0.635,
+      "step": 44570
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 5.305381433122328e-07,
+      "loss": 0.6101,
+      "step": 44580
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 5.286055506886911e-07,
+      "loss": 0.3987,
+      "step": 44590
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 5.266763558381149e-07,
+      "loss": 0.8255,
+      "step": 44600
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 5.247505597007704e-07,
+      "loss": 0.2846,
+      "step": 44610
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 5.228281632152651e-07,
+      "loss": 0.6096,
+      "step": 44620
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 5.209091673185546e-07,
+      "loss": 0.667,
+      "step": 44630
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 5.18993572945933e-07,
+      "loss": 0.3633,
+      "step": 44640
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 5.170813810310388e-07,
+      "loss": 0.4713,
+      "step": 44650
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 5.151725925058478e-07,
+      "loss": 0.2222,
+      "step": 44660
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 5.132672083006836e-07,
+      "loss": 0.459,
+      "step": 44670
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 5.113652293442067e-07,
+      "loss": 0.4248,
+      "step": 44680
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 5.094666565634178e-07,
+      "loss": 0.602,
+      "step": 44690
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 5.075714908836587e-07,
+      "loss": 0.4219,
+      "step": 44700
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 5.056797332286103e-07,
+      "loss": 0.3305,
+      "step": 44710
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 5.037913845202901e-07,
+      "loss": 0.2996,
+      "step": 44720
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 5.019064456790612e-07,
+      "loss": 0.2826,
+      "step": 44730
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 5.000249176236151e-07,
+      "loss": 0.4705,
+      "step": 44740
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 4.981468012709877e-07,
+      "loss": 0.6573,
+      "step": 44750
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 4.962720975365492e-07,
+      "loss": 0.4925,
+      "step": 44760
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 4.944008073340086e-07,
+      "loss": 0.3959,
+      "step": 44770
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 4.925329315754099e-07,
+      "loss": 0.2424,
+      "step": 44780
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 4.906684711711328e-07,
+      "loss": 0.3226,
+      "step": 44790
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 4.888074270298917e-07,
+      "loss": 0.4597,
+      "step": 44800
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 4.869498000587375e-07,
+      "loss": 0.3997,
+      "step": 44810
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 4.850955911630576e-07,
+      "loss": 0.3794,
+      "step": 44820
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 4.832448012465665e-07,
+      "loss": 0.5885,
+      "step": 44830
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 4.813974312113195e-07,
+      "loss": 0.5058,
+      "step": 44840
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 4.795534819577005e-07,
+      "loss": 0.5762,
+      "step": 44850
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 4.777129543844297e-07,
+      "loss": 0.5572,
+      "step": 44860
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 4.7587584938855786e-07,
+      "loss": 0.9591,
+      "step": 44870
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 4.7404216786546634e-07,
+      "loss": 0.6308,
+      "step": 44880
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 4.7221191070887013e-07,
+      "loss": 0.3762,
+      "step": 44890
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 4.703850788108141e-07,
+      "loss": 0.7058,
+      "step": 44900
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 4.685616730616743e-07,
+      "loss": 0.167,
+      "step": 44910
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 4.667416943501557e-07,
+      "loss": 0.4422,
+      "step": 44920
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 4.649251435632945e-07,
+      "loss": 0.2814,
+      "step": 44930
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 4.631120215864551e-07,
+      "loss": 0.4397,
+      "step": 44940
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 4.6130232930333056e-07,
+      "loss": 0.6745,
+      "step": 44950
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 4.5949606759594446e-07,
+      "loss": 0.0493,
+      "step": 44960
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 4.57693237344646e-07,
+      "loss": 0.5869,
+      "step": 44970
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 4.5589383942811315e-07,
+      "loss": 0.5944,
+      "step": 44980
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 4.540978747233493e-07,
+      "loss": 0.3901,
+      "step": 44990
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 4.5230534410568764e-07,
+      "loss": 0.572,
+      "step": 45000
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 4.505162484487843e-07,
+      "loss": 0.3358,
+      "step": 45010
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 4.4873058862462347e-07,
+      "loss": 0.349,
+      "step": 45020
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 4.469483655035148e-07,
+      "loss": 0.5895,
+      "step": 45030
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 4.451695799540892e-07,
+      "loss": 0.2153,
+      "step": 45040
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 4.433942328433091e-07,
+      "loss": 0.6343,
+      "step": 45050
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 4.4162232503645484e-07,
+      "loss": 0.2725,
+      "step": 45060
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 4.3985385739713306e-07,
+      "loss": 0.5323,
+      "step": 45070
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 4.3808883078727437e-07,
+      "loss": 0.3967,
+      "step": 45080
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 4.363272460671297e-07,
+      "loss": 0.7175,
+      "step": 45090
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 4.3456910409527483e-07,
+      "loss": 1.0338,
+      "step": 45100
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 4.328144057286068e-07,
+      "loss": 0.352,
+      "step": 45110
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 4.3106315182234484e-07,
+      "loss": 0.4248,
+      "step": 45120
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 4.2931534323002787e-07,
+      "loss": 0.5194,
+      "step": 45130
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 4.2757098080351606e-07,
+      "loss": 0.497,
+      "step": 45140
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 4.258300653929928e-07,
+      "loss": 0.5801,
+      "step": 45150
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 4.2409259784695675e-07,
+      "loss": 0.2365,
+      "step": 45160
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 4.223585790122306e-07,
+      "loss": 0.3726,
+      "step": 45170
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 4.2062800973395157e-07,
+      "loss": 0.6571,
+      "step": 45180
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 4.1890089085557996e-07,
+      "loss": 0.4289,
+      "step": 45190
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 4.171772232188925e-07,
+      "loss": 0.9237,
+      "step": 45200
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 4.1545700766398297e-07,
+      "loss": 0.2547,
+      "step": 45210
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 4.137402450292641e-07,
+      "loss": 0.3113,
+      "step": 45220
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 4.1202693615146476e-07,
+      "loss": 0.4242,
+      "step": 45230
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 4.103170818656296e-07,
+      "loss": 0.433,
+      "step": 45240
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 4.0861068300512364e-07,
+      "loss": 0.5028,
+      "step": 45250
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 4.0690774040162416e-07,
+      "loss": 0.6273,
+      "step": 45260
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 4.0520825488512463e-07,
+      "loss": 0.5756,
+      "step": 45270
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 4.0351222728393087e-07,
+      "loss": 0.3489,
+      "step": 45280
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 4.0181965842467e-07,
+      "loss": 0.9094,
+      "step": 45290
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 4.001305491322779e-07,
+      "loss": 0.4944,
+      "step": 45300
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 3.984449002300053e-07,
+      "loss": 0.7408,
+      "step": 45310
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 3.9676271253941827e-07,
+      "loss": 0.7202,
+      "step": 45320
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 3.9508398688039267e-07,
+      "loss": 0.6042,
+      "step": 45330
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 3.9340872407112234e-07,
+      "loss": 0.5301,
+      "step": 45340
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 3.917369249281083e-07,
+      "loss": 0.437,
+      "step": 45350
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 3.900685902661638e-07,
+      "loss": 0.4105,
+      "step": 45360
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 3.884037208984151e-07,
+      "loss": 0.2409,
+      "step": 45370
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 3.8674231763630134e-07,
+      "loss": 0.3427,
+      "step": 45380
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 3.850843812895691e-07,
+      "loss": 0.5633,
+      "step": 45390
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 3.8342991266627605e-07,
+      "loss": 0.6687,
+      "step": 45400
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 3.8177891257279136e-07,
+      "loss": 0.4543,
+      "step": 45410
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 3.8013138181379054e-07,
+      "loss": 0.5662,
+      "step": 45420
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 3.784873211922604e-07,
+      "loss": 0.9155,
+      "step": 45430
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 3.768467315094992e-07,
+      "loss": 0.4995,
+      "step": 45440
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 3.752096135651073e-07,
+      "loss": 0.4798,
+      "step": 45450
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 3.7357596815699557e-07,
+      "loss": 0.7374,
+      "step": 45460
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 3.719457960813846e-07,
+      "loss": 0.3294,
+      "step": 45470
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 3.7031909813279974e-07,
+      "loss": 0.5075,
+      "step": 45480
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 3.68695875104075e-07,
+      "loss": 0.3049,
+      "step": 45490
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 3.6707612778634855e-07,
+      "loss": 0.7697,
+      "step": 45500
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 3.6545985696906535e-07,
+      "loss": 0.4896,
+      "step": 45510
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 3.6384706343997544e-07,
+      "loss": 0.531,
+      "step": 45520
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 3.6223774798513825e-07,
+      "loss": 0.5829,
+      "step": 45530
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 3.6063191138891144e-07,
+      "loss": 0.7269,
+      "step": 45540
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 3.590295544339606e-07,
+      "loss": 0.3007,
+      "step": 45550
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 3.574306779012551e-07,
+      "loss": 0.6541,
+      "step": 45560
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 3.5583528257006755e-07,
+      "loss": 0.6501,
+      "step": 45570
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 3.542433692179767e-07,
+      "loss": 0.2508,
+      "step": 45580
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 3.526549386208594e-07,
+      "loss": 0.3447,
+      "step": 45590
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 3.510699915528981e-07,
+      "loss": 0.2587,
+      "step": 45600
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 3.494885287865765e-07,
+      "loss": 0.4874,
+      "step": 45610
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 3.4791055109268053e-07,
+      "loss": 0.6196,
+      "step": 45620
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 3.4633605924029754e-07,
+      "loss": 0.7277,
+      "step": 45630
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 3.4476505399681615e-07,
+      "loss": 0.754,
+      "step": 45640
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 3.431975361279238e-07,
+      "loss": 0.6357,
+      "step": 45650
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 3.416335063976103e-07,
+      "loss": 0.1892,
+      "step": 45660
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 3.400729655681659e-07,
+      "loss": 0.5099,
+      "step": 45670
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 3.3851591440017785e-07,
+      "loss": 0.6402,
+      "step": 45680
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 3.369623536525343e-07,
+      "loss": 0.4794,
+      "step": 45690
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 3.354122840824228e-07,
+      "loss": 0.4491,
+      "step": 45700
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 3.3386570644532736e-07,
+      "loss": 0.4113,
+      "step": 45710
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 3.323226214950309e-07,
+      "loss": 0.278,
+      "step": 45720
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 3.307830299836151e-07,
+      "loss": 0.4852,
+      "step": 45730
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 3.2924693266145804e-07,
+      "loss": 0.7552,
+      "step": 45740
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 3.2771433027723425e-07,
+      "loss": 0.5251,
+      "step": 45750
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 3.261852235779153e-07,
+      "loss": 0.3029,
+      "step": 45760
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 3.246596133087709e-07,
+      "loss": 0.442,
+      "step": 45770
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 3.2313750021336464e-07,
+      "loss": 0.7183,
+      "step": 45780
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 3.2161888503355386e-07,
+      "loss": 0.7049,
+      "step": 45790
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 3.201037685094957e-07,
+      "loss": 0.7686,
+      "step": 45800
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 3.1859215137963775e-07,
+      "loss": 0.3101,
+      "step": 45810
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 3.170840343807249e-07,
+      "loss": 0.6807,
+      "step": 45820
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 3.1557941824779417e-07,
+      "loss": 0.3282,
+      "step": 45830
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 3.140783037141781e-07,
+      "loss": 0.5177,
+      "step": 45840
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 3.1258069151150067e-07,
+      "loss": 0.6931,
+      "step": 45850
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 3.110865823696804e-07,
+      "loss": 0.4378,
+      "step": 45860
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 3.0959597701692905e-07,
+      "loss": 0.683,
+      "step": 45870
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 3.0810887617974793e-07,
+      "loss": 0.3099,
+      "step": 45880
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 3.066252805829323e-07,
+      "loss": 0.7326,
+      "step": 45890
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 3.0514519094956864e-07,
+      "loss": 0.6025,
+      "step": 45900
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 3.0366860800103507e-07,
+      "loss": 0.6772,
+      "step": 45910
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 3.02195532457e-07,
+      "loss": 0.3919,
+      "step": 45920
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 3.0072596503542244e-07,
+      "loss": 0.2697,
+      "step": 45930
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 2.99259906452552e-07,
+      "loss": 0.6047,
+      "step": 45940
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 2.97797357422927e-07,
+      "loss": 0.3856,
+      "step": 45950
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 2.9633831865937656e-07,
+      "loss": 0.4051,
+      "step": 45960
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 2.9488279087302004e-07,
+      "loss": 0.4755,
+      "step": 45970
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 2.9343077477326173e-07,
+      "loss": 0.3595,
+      "step": 45980
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 2.9198227106779634e-07,
+      "loss": 0.6075,
+      "step": 45990
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 2.905372804626083e-07,
+      "loss": 0.628,
+      "step": 46000
+    },
+    {
+      "epoch": 22.04,
+      "eval_accuracy": 0.8747368421052631,
+      "eval_f1": 0.8747368421052631,
+      "eval_loss": 0.8088645339012146,
+      "eval_runtime": 748.0599,
+      "eval_samples_per_second": 6.35,
+      "eval_steps_per_second": 1.588,
+      "step": 46000
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 2.8909580366196926e-07,
+      "loss": 0.4376,
+      "step": 46010
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 2.8765784136843554e-07,
+      "loss": 0.7477,
+      "step": 46020
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 2.8622339428285406e-07,
+      "loss": 0.294,
+      "step": 46030
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 2.847924631043555e-07,
+      "loss": 0.4043,
+      "step": 46040
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 2.833650485303596e-07,
+      "loss": 0.4888,
+      "step": 46050
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 2.8194115125657146e-07,
+      "loss": 0.7153,
+      "step": 46060
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 2.8052077197697925e-07,
+      "loss": 0.3605,
+      "step": 46070
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 2.791039113838592e-07,
+      "loss": 0.2711,
+      "step": 46080
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 2.7769057016777234e-07,
+      "loss": 0.3246,
+      "step": 46090
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 2.7628074901756423e-07,
+      "loss": 0.696,
+      "step": 46100
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 2.748744486203644e-07,
+      "loss": 0.6081,
+      "step": 46110
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 2.734716696615863e-07,
+      "loss": 0.2118,
+      "step": 46120
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 2.720724128249272e-07,
+      "loss": 0.617,
+      "step": 46130
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 2.7067667879236815e-07,
+      "loss": 0.6095,
+      "step": 46140
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 2.6928446824417503e-07,
+      "loss": 0.5906,
+      "step": 46150
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 2.678957818588909e-07,
+      "loss": 0.7001,
+      "step": 46160
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 2.6651062031334607e-07,
+      "loss": 0.2159,
+      "step": 46170
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 2.651289842826504e-07,
+      "loss": 0.3471,
+      "step": 46180
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 2.637508744401987e-07,
+      "loss": 0.4117,
+      "step": 46190
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 2.623762914576644e-07,
+      "loss": 0.6562,
+      "step": 46200
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 2.6100523600500163e-07,
+      "loss": 0.5202,
+      "step": 46210
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 2.596377087504459e-07,
+      "loss": 0.6931,
+      "step": 46220
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 2.5827371036051486e-07,
+      "loss": 0.4239,
+      "step": 46230
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 2.5691324150000333e-07,
+      "loss": 0.8724,
+      "step": 46240
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 2.555563028319885e-07,
+      "loss": 0.8134,
+      "step": 46250
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 2.542028950178246e-07,
+      "loss": 0.6479,
+      "step": 46260
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 2.528530187171474e-07,
+      "loss": 0.3903,
+      "step": 46270
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 2.5150667458786804e-07,
+      "loss": 0.3458,
+      "step": 46280
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 2.5016386328618077e-07,
+      "loss": 0.4394,
+      "step": 46290
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 2.488245854665547e-07,
+      "loss": 0.5828,
+      "step": 46300
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 2.4748884178173754e-07,
+      "loss": 0.6674,
+      "step": 46310
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 2.461566328827536e-07,
+      "loss": 0.3149,
+      "step": 46320
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 2.4482795941890677e-07,
+      "loss": 0.6931,
+      "step": 46330
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 2.435028220377758e-07,
+      "loss": 0.3118,
+      "step": 46340
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 2.421812213852165e-07,
+      "loss": 0.6017,
+      "step": 46350
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 2.408631581053602e-07,
+      "loss": 0.4137,
+      "step": 46360
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 2.3954863284061573e-07,
+      "loss": 0.395,
+      "step": 46370
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 2.3823764623166712e-07,
+      "loss": 0.5066,
+      "step": 46380
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 2.3693019891747241e-07,
+      "loss": 0.4814,
+      "step": 46390
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 2.3562629153526693e-07,
+      "loss": 0.5267,
+      "step": 46400
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 2.3432592472055732e-07,
+      "loss": 0.8471,
+      "step": 46410
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 2.3302909910712823e-07,
+      "loss": 0.3451,
+      "step": 46420
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 2.3173581532703574e-07,
+      "loss": 0.6432,
+      "step": 46430
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 2.3044607401061152e-07,
+      "loss": 0.2905,
+      "step": 46440
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 2.2915987578645852e-07,
+      "loss": 0.4183,
+      "step": 46450
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 2.2787722128145534e-07,
+      "loss": 0.4307,
+      "step": 46460
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 2.26598111120751e-07,
+      "loss": 0.4788,
+      "step": 46470
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 2.2532254592777025e-07,
+      "loss": 0.2118,
+      "step": 46480
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 2.2405052632420652e-07,
+      "loss": 0.3335,
+      "step": 46490
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 2.2278205293002645e-07,
+      "loss": 0.1742,
+      "step": 46500
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 2.2151712636346966e-07,
+      "loss": 0.5771,
+      "step": 46510
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 2.2025574724104548e-07,
+      "loss": 0.2051,
+      "step": 46520
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 2.189979161775346e-07,
+      "loss": 0.3724,
+      "step": 46530
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 2.1774363378598838e-07,
+      "loss": 0.3743,
+      "step": 46540
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 2.1649290067773026e-07,
+      "loss": 0.5481,
+      "step": 46550
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 2.152457174623501e-07,
+      "loss": 0.6652,
+      "step": 46560
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 2.1400208474771087e-07,
+      "loss": 0.4334,
+      "step": 46570
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 2.1276200313994522e-07,
+      "loss": 0.339,
+      "step": 46580
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 2.115254732434538e-07,
+      "loss": 0.4181,
+      "step": 46590
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 2.102924956609037e-07,
+      "loss": 0.3678,
+      "step": 46600
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 2.0906307099323508e-07,
+      "loss": 0.4751,
+      "step": 46610
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 2.0783719983965443e-07,
+      "loss": 0.4232,
+      "step": 46620
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 2.0661488279763634e-07,
+      "loss": 0.3166,
+      "step": 46630
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 2.0539612046292344e-07,
+      "loss": 0.3519,
+      "step": 46640
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 2.0418091342952638e-07,
+      "loss": 0.4472,
+      "step": 46650
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 2.0296926228971973e-07,
+      "loss": 0.6069,
+      "step": 46660
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 2.0176116763405027e-07,
+      "loss": 0.5496,
+      "step": 46670
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 2.0055663005132868e-07,
+      "loss": 0.5399,
+      "step": 46680
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 1.9935565012863032e-07,
+      "loss": 0.2832,
+      "step": 46690
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 1.9815822845129783e-07,
+      "loss": 0.5219,
+      "step": 46700
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 1.9696436560294184e-07,
+      "loss": 0.4706,
+      "step": 46710
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 1.9577406216543607e-07,
+      "loss": 0.3866,
+      "step": 46720
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 1.9458731871891899e-07,
+      "loss": 0.4084,
+      "step": 46730
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 1.9340413584179623e-07,
+      "loss": 0.2977,
+      "step": 46740
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.9222451411073648e-07,
+      "loss": 0.6459,
+      "step": 46750
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.9104845410067072e-07,
+      "loss": 0.5983,
+      "step": 46760
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.8987595638480042e-07,
+      "loss": 0.5445,
+      "step": 46770
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.887070215345843e-07,
+      "loss": 0.5333,
+      "step": 46780
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.875416501197466e-07,
+      "loss": 0.2043,
+      "step": 46790
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.8637984270827552e-07,
+      "loss": 0.6342,
+      "step": 46800
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.8522159986642306e-07,
+      "loss": 0.6586,
+      "step": 46810
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.8406692215870185e-07,
+      "loss": 0.3052,
+      "step": 46820
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.8291581014788755e-07,
+      "loss": 0.9379,
+      "step": 46830
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.8176826439501887e-07,
+      "loss": 0.26,
+      "step": 46840
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.8062428545939424e-07,
+      "loss": 0.3086,
+      "step": 46850
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.7948387389857767e-07,
+      "loss": 0.1077,
+      "step": 46860
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.7834703026838957e-07,
+      "loss": 0.3269,
+      "step": 46870
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.7721375512291417e-07,
+      "loss": 0.3113,
+      "step": 46880
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.7608404901449548e-07,
+      "loss": 0.2717,
+      "step": 46890
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.749579124937406e-07,
+      "loss": 0.6674,
+      "step": 46900
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.738353461095138e-07,
+      "loss": 0.2997,
+      "step": 46910
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.7271635040893908e-07,
+      "loss": 0.6477,
+      "step": 46920
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.716009259374035e-07,
+      "loss": 0.3548,
+      "step": 46930
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.704890732385489e-07,
+      "loss": 0.3934,
+      "step": 46940
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.6938079285428094e-07,
+      "loss": 0.3057,
+      "step": 46950
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.6827608532476168e-07,
+      "loss": 0.6208,
+      "step": 46960
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.6717495118841048e-07,
+      "loss": 0.3769,
+      "step": 46970
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.6607739098190883e-07,
+      "loss": 0.331,
+      "step": 46980
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.6498340524019218e-07,
+      "loss": 0.6539,
+      "step": 46990
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.6389299449645734e-07,
+      "loss": 0.6753,
+      "step": 47000
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.6280615928215753e-07,
+      "loss": 0.1063,
+      "step": 47010
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.6172290012700235e-07,
+      "loss": 0.7285,
+      "step": 47020
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.606432175589595e-07,
+      "loss": 0.5733,
+      "step": 47030
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.5956711210425218e-07,
+      "loss": 0.4015,
+      "step": 47040
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.584945842873625e-07,
+      "loss": 0.6097,
+      "step": 47050
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.5742563463102654e-07,
+      "loss": 0.326,
+      "step": 47060
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.5636026365623833e-07,
+      "loss": 0.8333,
+      "step": 47070
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.5529847188224504e-07,
+      "loss": 0.4393,
+      "step": 47080
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.5424025982655187e-07,
+      "loss": 0.9822,
+      "step": 47090
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.5318562800491957e-07,
+      "loss": 0.6041,
+      "step": 47100
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.5213457693136201e-07,
+      "loss": 0.3749,
+      "step": 47110
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.5108710711814855e-07,
+      "loss": 0.4134,
+      "step": 47120
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.5004321907580332e-07,
+      "loss": 0.4031,
+      "step": 47130
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.4900291331310518e-07,
+      "loss": 0.35,
+      "step": 47140
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.4796619033708686e-07,
+      "loss": 0.3272,
+      "step": 47150
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.4693305065303252e-07,
+      "loss": 0.247,
+      "step": 47160
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.459034947644844e-07,
+      "loss": 0.7903,
+      "step": 47170
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.4487752317323272e-07,
+      "loss": 0.3523,
+      "step": 47180
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.438551363793267e-07,
+      "loss": 0.7924,
+      "step": 47190
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.4283633488106352e-07,
+      "loss": 0.3725,
+      "step": 47200
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.4182111917499517e-07,
+      "loss": 0.9139,
+      "step": 47210
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.4080948975592423e-07,
+      "loss": 0.2654,
+      "step": 47220
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.398014471169079e-07,
+      "loss": 0.4064,
+      "step": 47230
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.387969917492532e-07,
+      "loss": 0.3291,
+      "step": 47240
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.3779612414252017e-07,
+      "loss": 0.5799,
+      "step": 47250
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 1.3679884478451948e-07,
+      "loss": 0.423,
+      "step": 47260
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 1.358051541613131e-07,
+      "loss": 0.6288,
+      "step": 47270
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 1.3481505275721202e-07,
+      "loss": 0.6378,
+      "step": 47280
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 1.3382854105478099e-07,
+      "loss": 0.7306,
+      "step": 47290
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 1.3284561953483541e-07,
+      "loss": 0.2512,
+      "step": 47300
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 1.3186628867643713e-07,
+      "loss": 0.4269,
+      "step": 47310
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 1.308905489568993e-07,
+      "loss": 0.2313,
+      "step": 47320
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 1.2991840085178814e-07,
+      "loss": 0.7018,
+      "step": 47330
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 1.2894984483491468e-07,
+      "loss": 0.3318,
+      "step": 47340
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 1.279848813783438e-07,
+      "loss": 0.2508,
+      "step": 47350
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 1.270235109523843e-07,
+      "loss": 0.3462,
+      "step": 47360
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 1.2606573402559718e-07,
+      "loss": 0.1962,
+      "step": 47370
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 1.251115510647932e-07,
+      "loss": 0.6131,
+      "step": 47380
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 1.241609625350279e-07,
+      "loss": 0.4011,
+      "step": 47390
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 1.2321396889960645e-07,
+      "loss": 0.597,
+      "step": 47400
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 1.2227057062008217e-07,
+      "loss": 0.3469,
+      "step": 47410
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 1.213307681562556e-07,
+      "loss": 0.5855,
+      "step": 47420
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 1.2039456196617616e-07,
+      "loss": 0.3312,
+      "step": 47430
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 1.1946195250613885e-07,
+      "loss": 0.3626,
+      "step": 47440
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 1.1853294023068589e-07,
+      "loss": 0.3461,
+      "step": 47450
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 1.1760752559260679e-07,
+      "loss": 0.4803,
+      "step": 47460
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 1.1668570904293657e-07,
+      "loss": 0.4819,
+      "step": 47470
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 1.1576749103096002e-07,
+      "loss": 0.9285,
+      "step": 47480
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 1.1485287200420246e-07,
+      "loss": 0.473,
+      "step": 47490
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 1.1394185240843985e-07,
+      "loss": 0.271,
+      "step": 47500
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 1.1303443268769114e-07,
+      "loss": 0.6521,
+      "step": 47510
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 1.1213061328422175e-07,
+      "loss": 0.6992,
+      "step": 47520
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 1.1123039463854345e-07,
+      "loss": 0.4707,
+      "step": 47530
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 1.1033377718941112e-07,
+      "loss": 0.6397,
+      "step": 47540
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 1.0944076137382436e-07,
+      "loss": 0.3328,
+      "step": 47550
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 1.0855134762702917e-07,
+      "loss": 0.2288,
+      "step": 47560
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 1.0766553638251464e-07,
+      "loss": 0.3954,
+      "step": 47570
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 1.0678332807201541e-07,
+      "loss": 0.4067,
+      "step": 47580
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 1.0590472312550753e-07,
+      "loss": 0.5821,
+      "step": 47590
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 1.0502972197121347e-07,
+      "loss": 0.6503,
+      "step": 47600
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 1.0415832503559708e-07,
+      "loss": 0.5509,
+      "step": 47610
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 1.032905327433678e-07,
+      "loss": 0.9522,
+      "step": 47620
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 1.0242634551747731e-07,
+      "loss": 0.6986,
+      "step": 47630
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 1.0156576377911869e-07,
+      "loss": 0.4621,
+      "step": 47640
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 1.0070878794772975e-07,
+      "loss": 0.6139,
+      "step": 47650
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 9.985541844098972e-08,
+      "loss": 0.5695,
+      "step": 47660
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 9.900565567482089e-08,
+      "loss": 0.5113,
+      "step": 47670
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 9.815950006338697e-08,
+      "loss": 0.3426,
+      "step": 47680
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 9.731695201909474e-08,
+      "loss": 0.4428,
+      "step": 47690
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 9.647801195259071e-08,
+      "loss": 0.4442,
+      "step": 47700
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 9.564268027276446e-08,
+      "loss": 0.3995,
+      "step": 47710
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 9.4810957386747e-08,
+      "loss": 0.8283,
+      "step": 47720
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 9.398284369990989e-08,
+      "loss": 0.438,
+      "step": 47730
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 9.315833961586612e-08,
+      "loss": 0.5626,
+      "step": 47740
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 9.233744553646756e-08,
+      "loss": 0.4854,
+      "step": 47750
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 9.152016186180834e-08,
+      "loss": 0.4819,
+      "step": 47760
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 9.070648899022315e-08,
+      "loss": 0.769,
+      "step": 47770
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 8.989642731828646e-08,
+      "loss": 0.4348,
+      "step": 47780
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 8.908997724081241e-08,
+      "loss": 0.5948,
+      "step": 47790
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 8.828713915085412e-08,
+      "loss": 0.1465,
+      "step": 47800
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 8.748791343970524e-08,
+      "loss": 0.3815,
+      "step": 47810
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 8.669230049690003e-08,
+      "loss": 0.3987,
+      "step": 47820
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 8.590030071020994e-08,
+      "loss": 0.316,
+      "step": 47830
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 8.511191446564537e-08,
+      "loss": 0.2271,
+      "step": 47840
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 8.432714214745646e-08,
+      "loss": 0.3009,
+      "step": 47850
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 8.354598413813141e-08,
+      "loss": 0.2945,
+      "step": 47860
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 8.276844081839819e-08,
+      "loss": 0.2503,
+      "step": 47870
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 8.199451256722196e-08,
+      "loss": 0.5661,
+      "step": 47880
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 8.122419976180434e-08,
+      "loss": 0.6049,
+      "step": 47890
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 8.045750277758751e-08,
+      "loss": 0.4348,
+      "step": 47900
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 7.969442198825089e-08,
+      "loss": 0.8783,
+      "step": 47910
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 7.893495776570947e-08,
+      "loss": 0.4905,
+      "step": 47920
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 7.8179110480118e-08,
+      "loss": 0.3333,
+      "step": 47930
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 7.742688049986596e-08,
+      "loss": 0.4873,
+      "step": 47940
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 7.667826819158257e-08,
+      "loss": 0.4427,
+      "step": 47950
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 7.593327392013183e-08,
+      "loss": 0.4111,
+      "step": 47960
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 7.519189804861493e-08,
+      "loss": 0.6091,
+      "step": 47970
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 7.44541409383695e-08,
+      "loss": 0.2827,
+      "step": 47980
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 7.372000294896792e-08,
+      "loss": 0.2771,
+      "step": 47990
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 7.298948443822229e-08,
+      "loss": 0.5693,
+      "step": 48000
+    },
+    {
+      "epoch": 23.04,
+      "eval_accuracy": 0.8747368421052631,
+      "eval_f1": 0.8747368421052631,
+      "eval_loss": 0.8009958863258362,
+      "eval_runtime": 732.1449,
+      "eval_samples_per_second": 6.488,
+      "eval_steps_per_second": 1.623,
+      "step": 48000
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 7.226258576217865e-08,
+      "loss": 0.3753,
+      "step": 48010
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 7.15393072751161e-08,
+      "loss": 0.4069,
+      "step": 48020
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 7.081964932955349e-08,
+      "loss": 0.4612,
+      "step": 48030
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 7.010361227624357e-08,
+      "loss": 0.3692,
+      "step": 48040
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 6.939119646417302e-08,
+      "loss": 0.3168,
+      "step": 48050
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 6.868240224056577e-08,
+      "loss": 0.5174,
+      "step": 48060
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 6.79772299508788e-08,
+      "loss": 0.5508,
+      "step": 48070
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 6.727567993880468e-08,
+      "loss": 0.5056,
+      "step": 48080
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 6.657775254626991e-08,
+      "loss": 0.4838,
+      "step": 48090
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 6.588344811343738e-08,
+      "loss": 0.3053,
+      "step": 48100
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 6.519276697870142e-08,
+      "loss": 0.4235,
+      "step": 48110
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 6.450570947869106e-08,
+      "loss": 0.5998,
+      "step": 48120
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 6.382227594827012e-08,
+      "loss": 0.5087,
+      "step": 48130
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 6.314246672053715e-08,
+      "loss": 0.4317,
+      "step": 48140
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 6.246628212682209e-08,
+      "loss": 0.8639,
+      "step": 48150
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 6.179372249668802e-08,
+      "loss": 0.3694,
+      "step": 48160
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 6.112478815793437e-08,
+      "loss": 0.2338,
+      "step": 48170
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 6.045947943658953e-08,
+      "loss": 0.2262,
+      "step": 48180
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 5.979779665691826e-08,
+      "loss": 0.5833,
+      "step": 48190
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 5.9139740141416765e-08,
+      "loss": 0.3248,
+      "step": 48200
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 5.848531021081266e-08,
+      "loss": 1.0088,
+      "step": 48210
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 5.7834507184067466e-08,
+      "loss": 0.3743,
+      "step": 48220
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 5.718733137837578e-08,
+      "loss": 0.4842,
+      "step": 48230
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 5.6543783109161974e-08,
+      "loss": 0.2676,
+      "step": 48240
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 5.5903862690085125e-08,
+      "loss": 0.3801,
+      "step": 48250
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 5.526757043303243e-08,
+      "loss": 0.7997,
+      "step": 48260
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 5.463490664812748e-08,
+      "loss": 0.7286,
+      "step": 48270
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 5.4005871643721114e-08,
+      "loss": 0.2438,
+      "step": 48280
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 5.3380465726398096e-08,
+      "loss": 0.6707,
+      "step": 48290
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 5.275868920097293e-08,
+      "loss": 0.5388,
+      "step": 48300
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 5.214054237049321e-08,
+      "loss": 0.3983,
+      "step": 48310
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 5.152602553623459e-08,
+      "loss": 0.3292,
+      "step": 48320
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 5.091513899770667e-08,
+      "loss": 0.5818,
+      "step": 48330
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 5.0307883052647944e-08,
+      "loss": 0.4123,
+      "step": 48340
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 4.970425799702666e-08,
+      "loss": 0.389,
+      "step": 48350
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 4.910426412504332e-08,
+      "loss": 0.258,
+      "step": 48360
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 4.85079017291265e-08,
+      "loss": 0.6119,
+      "step": 48370
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 4.791517109993704e-08,
+      "loss": 0.5102,
+      "step": 48380
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 4.732607252636384e-08,
+      "loss": 0.5128,
+      "step": 48390
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 4.6740606295527236e-08,
+      "loss": 0.1333,
+      "step": 48400
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 4.615877269277563e-08,
+      "loss": 0.3087,
+      "step": 48410
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 4.558057200168802e-08,
+      "loss": 0.6449,
+      "step": 48420
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 4.500600450407233e-08,
+      "loss": 0.3746,
+      "step": 48430
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 4.4435070479965366e-08,
+      "loss": 0.5859,
+      "step": 48440
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 4.386777020763455e-08,
+      "loss": 0.3587,
+      "step": 48450
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 4.330410396357371e-08,
+      "loss": 0.2082,
+      "step": 48460
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 4.274407202250807e-08,
+      "loss": 0.3468,
+      "step": 48470
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 4.2187674657390154e-08,
+      "loss": 0.8718,
+      "step": 48480
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 4.1634912139400514e-08,
+      "loss": 0.4234,
+      "step": 48490
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 4.108578473795033e-08,
+      "loss": 0.6711,
+      "step": 48500
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 4.0540292720675495e-08,
+      "loss": 0.4326,
+      "step": 48510
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 3.999843635344419e-08,
+      "loss": 0.6955,
+      "step": 48520
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 3.946021590035015e-08,
+      "loss": 0.3644,
+      "step": 48530
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 3.892563162371521e-08,
+      "loss": 0.1796,
+      "step": 48540
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 3.839468378408845e-08,
+      "loss": 0.4076,
+      "step": 48550
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 3.7867372640248697e-08,
+      "loss": 0.6121,
+      "step": 48560
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 3.734369844920038e-08,
+      "loss": 0.446,
+      "step": 48570
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 3.6823661466176825e-08,
+      "loss": 0.4861,
+      "step": 48580
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 3.6307261944636126e-08,
+      "loss": 0.4031,
+      "step": 48590
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 3.579450013626612e-08,
+      "loss": 0.3043,
+      "step": 48600
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 3.528537629098022e-08,
+      "loss": 0.3778,
+      "step": 48610
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 3.477989065692078e-08,
+      "loss": 0.4012,
+      "step": 48620
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 3.4278043480453216e-08,
+      "loss": 0.3413,
+      "step": 48630
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 3.377983500617271e-08,
+      "loss": 0.6273,
+      "step": 48640
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 3.328526547690003e-08,
+      "loss": 0.7842,
+      "step": 48650
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 3.279433513368235e-08,
+      "loss": 0.462,
+      "step": 48660
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 3.230704421579328e-08,
+      "loss": 0.6631,
+      "step": 48670
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 3.1823392960732e-08,
+      "loss": 0.3991,
+      "step": 48680
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 3.134338160422412e-08,
+      "loss": 0.7023,
+      "step": 48690
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 3.086701038022249e-08,
+      "loss": 0.2972,
+      "step": 48700
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 3.039427952090307e-08,
+      "loss": 0.4932,
+      "step": 48710
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 2.992518925667154e-08,
+      "loss": 0.3013,
+      "step": 48720
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 2.9459739816154186e-08,
+      "loss": 0.3875,
+      "step": 48730
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 2.8997931426206214e-08,
+      "loss": 0.4445,
+      "step": 48740
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 2.8539764311908412e-08,
+      "loss": 0.5529,
+      "step": 48750
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 2.8085238696565485e-08,
+      "loss": 0.6926,
+      "step": 48760
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 2.7634354801706896e-08,
+      "loss": 0.3991,
+      "step": 48770
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 2.7187112847087693e-08,
+      "loss": 0.9109,
+      "step": 48780
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 2.6743513050690172e-08,
+      "loss": 0.7538,
+      "step": 48790
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 2.6303555628717225e-08,
+      "loss": 0.4482,
+      "step": 48800
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 2.5867240795600656e-08,
+      "loss": 0.3986,
+      "step": 48810
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 2.5434568763993692e-08,
+      "loss": 0.6185,
+      "step": 48820
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 2.5005539744775984e-08,
+      "loss": 0.6925,
+      "step": 48830
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 2.45801539470511e-08,
+      "loss": 0.4379,
+      "step": 48840
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 2.4158411578146523e-08,
+      "loss": 0.4391,
+      "step": 48850
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 2.3740312843614497e-08,
+      "loss": 0.89,
+      "step": 48860
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 2.332585794723119e-08,
+      "loss": 0.3428,
+      "step": 48870
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 2.291504709099751e-08,
+      "loss": 0.7952,
+      "step": 48880
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 2.2507880475136634e-08,
+      "loss": 0.6414,
+      "step": 48890
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 2.2104358298098147e-08,
+      "loss": 0.3109,
+      "step": 48900
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 2.1704480756552237e-08,
+      "loss": 0.159,
+      "step": 48910
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 2.1308248045395494e-08,
+      "loss": 0.4725,
+      "step": 48920
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 2.0915660357746778e-08,
+      "loss": 0.7352,
+      "step": 48930
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 2.05267178849472e-08,
+      "loss": 0.6739,
+      "step": 48940
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 2.0141420816564282e-08,
+      "loss": 0.8071,
+      "step": 48950
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 1.9759769340386148e-08,
+      "loss": 0.5585,
+      "step": 48960
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 1.9381763642425665e-08,
+      "loss": 0.6838,
+      "step": 48970
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 1.9007403906918797e-08,
+      "loss": 0.6336,
+      "step": 48980
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 1.8636690316322092e-08,
+      "loss": 0.5289,
+      "step": 48990
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 1.8269623051318517e-08,
+      "loss": 0.7095,
+      "step": 49000
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 1.7906202290810803e-08,
+      "loss": 0.4653,
+      "step": 49010
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 1.7546428211927257e-08,
+      "loss": 0.2675,
+      "step": 49020
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 1.7190300990016784e-08,
+      "loss": 0.4124,
+      "step": 49030
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 1.6837820798650538e-08,
+      "loss": 0.6305,
+      "step": 49040
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 1.6488987809625268e-08,
+      "loss": 0.6233,
+      "step": 49050
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 1.6143802192955805e-08,
+      "loss": 0.694,
+      "step": 49060
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 1.580226411688257e-08,
+      "loss": 0.252,
+      "step": 49070
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 1.5464373747866577e-08,
+      "loss": 0.4817,
+      "step": 49080
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 1.5130131250591093e-08,
+      "loss": 0.1697,
+      "step": 49090
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 1.4799536787963308e-08,
+      "loss": 0.2944,
+      "step": 49100
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 1.4472590521110162e-08,
+      "loss": 0.7365,
+      "step": 49110
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 1.4149292609380027e-08,
+      "loss": 0.5104,
+      "step": 49120
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 1.3829643210346854e-08,
+      "loss": 0.7771,
+      "step": 49130
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 1.3513642479801857e-08,
+      "loss": 0.5946,
+      "step": 49140
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 1.3201290571760172e-08,
+      "loss": 0.5192,
+      "step": 49150
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 1.2892587638460018e-08,
+      "loss": 0.4447,
+      "step": 49160
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 1.258753383035771e-08,
+      "loss": 0.5295,
+      "step": 49170
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 1.2286129296132653e-08,
+      "loss": 0.3215,
+      "step": 49180
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 1.1988374182687334e-08,
+      "loss": 0.8657,
+      "step": 49190
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 1.1694268635142335e-08,
+      "loss": 0.4053,
+      "step": 49200
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 1.1403812796842161e-08,
+      "loss": 0.5778,
+      "step": 49210
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 1.1117006809351072e-08,
+      "loss": 0.4498,
+      "step": 49220
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 1.0833850812455581e-08,
+      "loss": 0.4014,
+      "step": 49230
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 1.0554344944161132e-08,
+      "loss": 0.3483,
+      "step": 49240
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 1.027848934069625e-08,
+      "loss": 0.2246,
+      "step": 49250
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 1.0006284136509224e-08,
+      "loss": 0.4773,
+      "step": 49260
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 9.737729464269762e-09,
+      "loss": 0.4115,
+      "step": 49270
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 9.472825454868995e-09,
+      "loss": 0.5762,
+      "step": 49280
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 9.211572237416145e-09,
+      "loss": 0.7074,
+      "step": 49290
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 8.953969939245188e-09,
+      "loss": 0.3273,
+      "step": 49300
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 8.700018685905697e-09,
+      "loss": 0.2681,
+      "step": 49310
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 8.449718601171997e-09,
+      "loss": 0.2451,
+      "step": 49320
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 8.203069807038165e-09,
+      "loss": 0.2841,
+      "step": 49330
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 7.960072423715547e-09,
+      "loss": 0.3003,
+      "step": 49340
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 7.720726569640235e-09,
+      "loss": 0.3248,
+      "step": 49350
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 7.48503236146475e-09,
+      "loss": 0.2326,
+      "step": 49360
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 7.252989914064701e-09,
+      "loss": 0.2678,
+      "step": 49370
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 7.024599340534621e-09,
+      "loss": 0.3889,
+      "step": 49380
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 6.7998607521888026e-09,
+      "loss": 0.6104,
+      "step": 49390
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 6.5787742585621256e-09,
+      "loss": 0.3541,
+      "step": 49400
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 6.361339967410895e-09,
+      "loss": 0.4186,
+      "step": 49410
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 6.147557984707841e-09,
+      "loss": 0.5533,
+      "step": 49420
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 5.937428414648782e-09,
+      "loss": 0.4036,
+      "step": 49430
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 5.730951359648462e-09,
+      "loss": 0.8842,
+      "step": 49440
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 5.528126920341381e-09,
+      "loss": 0.4078,
+      "step": 49450
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 5.3289551955809645e-09,
+      "loss": 0.3735,
+      "step": 49460
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 5.133436282441228e-09,
+      "loss": 0.5025,
+      "step": 49470
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 4.941570276215945e-09,
+      "loss": 0.2423,
+      "step": 49480
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 4.753357270418646e-09,
+      "loss": 0.5151,
+      "step": 49490
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 4.568797356781784e-09,
+      "loss": 0.4716,
+      "step": 49500
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 4.387890625257574e-09,
+      "loss": 0.689,
+      "step": 49510
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 4.210637164017983e-09,
+      "loss": 0.4343,
+      "step": 49520
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 4.037037059453908e-09,
+      "loss": 0.1581,
+      "step": 49530
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 3.8670903961751655e-09,
+      "loss": 0.5421,
+      "step": 49540
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 3.700797257013e-09,
+      "loss": 0.4423,
+      "step": 49550
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 3.5381577230167437e-09,
+      "loss": 0.4451,
+      "step": 49560
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 3.3791718734538235e-09,
+      "loss": 0.4525,
+      "step": 49570
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 3.2238397858122546e-09,
+      "loss": 0.3258,
+      "step": 49580
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 3.072161535799811e-09,
+      "loss": 0.4485,
+      "step": 49590
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 2.924137197342358e-09,
+      "loss": 0.6284,
+      "step": 49600
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 2.7797668425846857e-09,
+      "loss": 0.3722,
+      "step": 49610
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 2.6390505418913413e-09,
+      "loss": 0.6805,
+      "step": 49620
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 2.5019883638457973e-09,
+      "loss": 0.9199,
+      "step": 49630
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 2.368580375250451e-09,
+      "loss": 0.2866,
+      "step": 49640
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 2.2388266411266234e-09,
+      "loss": 0.4129,
+      "step": 49650
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 2.1127272247145614e-09,
+      "loss": 0.718,
+      "step": 49660
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 1.9902821874742684e-09,
+      "loss": 0.339,
+      "step": 49670
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 1.8714915890838404e-09,
+      "loss": 0.6156,
+      "step": 49680
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 1.7563554874402975e-09,
+      "loss": 0.3388,
+      "step": 49690
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 1.644873938658753e-09,
+      "loss": 0.5114,
+      "step": 49700
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 1.537046997074909e-09,
+      "loss": 0.8165,
+      "step": 49710
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 1.4328747152417277e-09,
+      "loss": 0.5025,
+      "step": 49720
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 1.332357143932761e-09,
+      "loss": 0.2093,
+      "step": 49730
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 1.2354943321371548e-09,
+      "loss": 0.6465,
+      "step": 49740
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 1.1422863270654781e-09,
+      "loss": 0.7074,
+      "step": 49750
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 1.0527331741472247e-09,
+      "loss": 0.3695,
+      "step": 49760
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 9.668349170274814e-10,
+      "loss": 0.6652,
+      "step": 49770
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 8.845915975735919e-10,
+      "loss": 0.4753,
+      "step": 49780
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 8.060032558693253e-10,
+      "loss": 0.3091,
+      "step": 49790
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 7.310699302182089e-10,
+      "loss": 0.4409,
+      "step": 49800
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 6.597916571418617e-10,
+      "loss": 0.2497,
+      "step": 49810
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 5.92168471379162e-10,
+      "loss": 0.5315,
+      "step": 49820
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 5.282004058895784e-10,
+      "loss": 0.5753,
+      "step": 49830
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 4.678874918515041e-10,
+      "loss": 0.5933,
+      "step": 49840
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 4.112297586589264e-10,
+      "loss": 0.6205,
+      "step": 49850
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 3.582272339272552e-10,
+      "loss": 0.3334,
+      "step": 49860
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 3.088799434891598e-10,
+      "loss": 0.5659,
+      "step": 49870
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 2.631879113954017e-10,
+      "loss": 0.6555,
+      "step": 49880
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 2.2115115991566682e-10,
+      "loss": 0.7503,
+      "step": 49890
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 1.8276970953939875e-10,
+      "loss": 0.4174,
+      "step": 49900
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 1.48043578971635e-10,
+      "loss": 0.234,
+      "step": 49910
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 1.1697278513800313e-10,
+      "loss": 0.521,
+      "step": 49920
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 8.955734318305542e-11,
+      "loss": 0.5629,
+      "step": 49930
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 6.579726646777085e-11,
+      "loss": 0.6017,
+      "step": 49940
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 4.5692566572053116e-11,
+      "loss": 0.6729,
+      "step": 49950
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 2.924325329556332e-11,
+      "loss": 0.2276,
+      "step": 49960
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 1.6449334655221914e-11,
+      "loss": 0.4924,
+      "step": 49970
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 7.310816886874072e-12,
+      "loss": 0.5423,
+      "step": 49980
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 1.8277044444570124e-12,
+      "loss": 0.3546,
+      "step": 49990
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 0.0,
+      "loss": 0.4764,
+      "step": 50000
+    },
+    {
+      "epoch": 24.04,
+      "eval_accuracy": 0.8789473684210526,
+      "eval_f1": 0.8789473684210526,
+      "eval_loss": 0.8117492198944092,
+      "eval_runtime": 766.7022,
+      "eval_samples_per_second": 6.195,
+      "eval_steps_per_second": 1.549,
+      "step": 50000
+    },
+    {
+      "epoch": 24.04,
+      "step": 50000,
+      "total_flos": 2.49219585441792e+20,
+      "train_loss": 0.7113624122858048,
+      "train_runtime": 65187.9107,
+      "train_samples_per_second": 3.068,
+      "train_steps_per_second": 0.767
+    },
+    {
+      "epoch": 24.04,
+      "eval_accuracy": 0.8757894736842106,
+      "eval_f1": 0.8757894736842106,
+      "eval_loss": 0.7742094993591309,
+      "eval_runtime": 766.7571,
+      "eval_samples_per_second": 6.195,
+      "eval_steps_per_second": 1.549,
+      "step": 50000
+    },
+    {
+      "epoch": 24.04,
+      "eval_accuracy": 0.870116156282999,
+      "eval_f1": 0.870116156282999,
+      "eval_loss": 0.7973663210868835,
+      "eval_runtime": 771.4511,
+      "eval_samples_per_second": 6.138,
+      "eval_steps_per_second": 1.535,
+      "step": 50000
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 50000,
+  "num_train_epochs": 9223372036854775807,
+  "save_steps": 500,
+  "total_flos": 2.49219585441792e+20,
+  "trial_name": null,
+  "trial_params": null
+}