diff --git "a/trainer_state (1).json" "b/trainer_state (1).json"
new file mode 100644--- /dev/null
+++ "b/trainer_state (1).json"
@@ -0,0 +1,60388 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 2500,
+  "global_step": 100000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999999999999996e-07,
+      "loss": 10.5991,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.5e-07,
+      "loss": 10.4404,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7e-07,
+      "loss": 10.4079,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.499999999999999e-07,
+      "loss": 10.3433,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2e-06,
+      "loss": 10.2894,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4499999999999999e-06,
+      "loss": 10.2258,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7e-06,
+      "loss": 10.1618,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9499999999999995e-06,
+      "loss": 10.0841,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1999999999999997e-06,
+      "loss": 10.0074,
+      "step": 90
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4499999999999994e-06,
+      "loss": 9.9546,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6999999999999996e-06,
+      "loss": 9.8959,
+      "step": 110
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9499999999999997e-06,
+      "loss": 9.8438,
+      "step": 120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1999999999999994e-06,
+      "loss": 9.7693,
+      "step": 130
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4499999999999996e-06,
+      "loss": 9.6931,
+      "step": 140
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6999999999999997e-06,
+      "loss": 9.6457,
+      "step": 150
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9499999999999995e-06,
+      "loss": 9.5779,
+      "step": 160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2e-06,
+      "loss": 9.5133,
+      "step": 170
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.45e-06,
+      "loss": 9.4415,
+      "step": 180
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.699999999999999e-06,
+      "loss": 9.3703,
+      "step": 190
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.95e-06,
+      "loss": 9.2958,
+      "step": 200
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.199999999999999e-06,
+      "loss": 9.2097,
+      "step": 210
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.4499999999999995e-06,
+      "loss": 9.1562,
+      "step": 220
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.7e-06,
+      "loss": 9.0652,
+      "step": 230
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.95e-06,
+      "loss": 9.0147,
+      "step": 240
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.199999999999999e-06,
+      "loss": 8.917,
+      "step": 250
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.449999999999999e-06,
+      "loss": 8.86,
+      "step": 260
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.699999999999999e-06,
+      "loss": 8.7804,
+      "step": 270
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.949999999999999e-06,
+      "loss": 8.6979,
+      "step": 280
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.2e-06,
+      "loss": 8.6123,
+      "step": 290
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.449999999999999e-06,
+      "loss": 8.5626,
+      "step": 300
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.699999999999999e-06,
+      "loss": 8.4648,
+      "step": 310
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.949999999999998e-06,
+      "loss": 8.4083,
+      "step": 320
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.2e-06,
+      "loss": 8.3222,
+      "step": 330
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.449999999999999e-06,
+      "loss": 8.2542,
+      "step": 340
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.7e-06,
+      "loss": 8.1903,
+      "step": 350
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.949999999999999e-06,
+      "loss": 8.1185,
+      "step": 360
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.199999999999998e-06,
+      "loss": 8.0568,
+      "step": 370
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.45e-06,
+      "loss": 7.9776,
+      "step": 380
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.699999999999999e-06,
+      "loss": 7.9201,
+      "step": 390
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.949999999999998e-06,
+      "loss": 7.84,
+      "step": 400
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.02e-05,
+      "loss": 7.788,
+      "step": 410
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0425e-05,
+      "loss": 7.708,
+      "step": 420
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0675e-05,
+      "loss": 7.6345,
+      "step": 430
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0925e-05,
+      "loss": 7.5975,
+      "step": 440
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1174999999999999e-05,
+      "loss": 7.5093,
+      "step": 450
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1424999999999998e-05,
+      "loss": 7.4443,
+      "step": 460
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1675e-05,
+      "loss": 7.3872,
+      "step": 470
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1924999999999998e-05,
+      "loss": 7.3158,
+      "step": 480
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2175e-05,
+      "loss": 7.2681,
+      "step": 490
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2424999999999999e-05,
+      "loss": 7.2233,
+      "step": 500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2675e-05,
+      "loss": 7.1814,
+      "step": 510
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2924999999999999e-05,
+      "loss": 7.1009,
+      "step": 520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3174999999999998e-05,
+      "loss": 7.0456,
+      "step": 530
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3424999999999998e-05,
+      "loss": 7.0051,
+      "step": 540
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3674999999999997e-05,
+      "loss": 6.9287,
+      "step": 550
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3925e-05,
+      "loss": 6.858,
+      "step": 560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4174999999999999e-05,
+      "loss": 6.8178,
+      "step": 570
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4424999999999998e-05,
+      "loss": 6.8199,
+      "step": 580
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4674999999999998e-05,
+      "loss": 6.7664,
+      "step": 590
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4925e-05,
+      "loss": 6.6982,
+      "step": 600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5175e-05,
+      "loss": 6.6649,
+      "step": 610
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5425e-05,
+      "loss": 6.6193,
+      "step": 620
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5674999999999998e-05,
+      "loss": 6.5672,
+      "step": 630
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5925e-05,
+      "loss": 6.5679,
+      "step": 640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6175e-05,
+      "loss": 6.5134,
+      "step": 650
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6425e-05,
+      "loss": 6.4569,
+      "step": 660
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6675e-05,
+      "loss": 6.4107,
+      "step": 670
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6924999999999998e-05,
+      "loss": 6.396,
+      "step": 680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7175e-05,
+      "loss": 6.3522,
+      "step": 690
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.74e-05,
+      "loss": 6.3442,
+      "step": 700
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.765e-05,
+      "loss": 6.3664,
+      "step": 710
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7899999999999998e-05,
+      "loss": 6.2891,
+      "step": 720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8149999999999997e-05,
+      "loss": 6.2448,
+      "step": 730
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8399999999999997e-05,
+      "loss": 6.1858,
+      "step": 740
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.865e-05,
+      "loss": 6.1695,
+      "step": 750
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.89e-05,
+      "loss": 6.1665,
+      "step": 760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.915e-05,
+      "loss": 6.0991,
+      "step": 770
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9399999999999997e-05,
+      "loss": 6.08,
+      "step": 780
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.965e-05,
+      "loss": 6.0533,
+      "step": 790
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9899999999999996e-05,
+      "loss": 6.0263,
+      "step": 800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.015e-05,
+      "loss": 5.9932,
+      "step": 810
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.04e-05,
+      "loss": 5.962,
+      "step": 820
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0649999999999997e-05,
+      "loss": 5.9409,
+      "step": 830
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.09e-05,
+      "loss": 5.919,
+      "step": 840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1149999999999996e-05,
+      "loss": 5.9305,
+      "step": 850
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.14e-05,
+      "loss": 5.9071,
+      "step": 860
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1649999999999998e-05,
+      "loss": 5.8489,
+      "step": 870
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1899999999999997e-05,
+      "loss": 5.8171,
+      "step": 880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.215e-05,
+      "loss": 5.7521,
+      "step": 890
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.24e-05,
+      "loss": 5.748,
+      "step": 900
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2649999999999998e-05,
+      "loss": 5.7184,
+      "step": 910
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2899999999999998e-05,
+      "loss": 5.7007,
+      "step": 920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3149999999999997e-05,
+      "loss": 5.7113,
+      "step": 930
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.34e-05,
+      "loss": 5.6782,
+      "step": 940
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.365e-05,
+      "loss": 5.6706,
+      "step": 950
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3899999999999998e-05,
+      "loss": 5.6403,
+      "step": 960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4149999999999997e-05,
+      "loss": 5.6065,
+      "step": 970
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4399999999999997e-05,
+      "loss": 5.5808,
+      "step": 980
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4649999999999996e-05,
+      "loss": 5.5483,
+      "step": 990
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.49e-05,
+      "loss": 5.5431,
+      "step": 1000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5149999999999998e-05,
+      "loss": 5.5711,
+      "step": 1010
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5399999999999997e-05,
+      "loss": 5.5249,
+      "step": 1020
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.565e-05,
+      "loss": 5.483,
+      "step": 1030
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5899999999999996e-05,
+      "loss": 5.4368,
+      "step": 1040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.615e-05,
+      "loss": 5.4181,
+      "step": 1050
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6399999999999995e-05,
+      "loss": 5.4194,
+      "step": 1060
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6649999999999997e-05,
+      "loss": 5.3955,
+      "step": 1070
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.69e-05,
+      "loss": 5.4054,
+      "step": 1080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7149999999999996e-05,
+      "loss": 5.356,
+      "step": 1090
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.74e-05,
+      "loss": 5.3373,
+      "step": 1100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7649999999999994e-05,
+      "loss": 5.289,
+      "step": 1110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7899999999999997e-05,
+      "loss": 5.3229,
+      "step": 1120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.815e-05,
+      "loss": 5.2827,
+      "step": 1130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8399999999999996e-05,
+      "loss": 5.2503,
+      "step": 1140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.865e-05,
+      "loss": 5.179,
+      "step": 1150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.89e-05,
+      "loss": 5.1685,
+      "step": 1160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9149999999999997e-05,
+      "loss": 5.2519,
+      "step": 1170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.94e-05,
+      "loss": 5.2074,
+      "step": 1180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9649999999999996e-05,
+      "loss": 5.172,
+      "step": 1190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.99e-05,
+      "loss": 5.1738,
+      "step": 1200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0149999999999998e-05,
+      "loss": 5.1721,
+      "step": 1210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0399999999999997e-05,
+      "loss": 5.1373,
+      "step": 1220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.065e-05,
+      "loss": 5.0675,
+      "step": 1230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.09e-05,
+      "loss": 5.0579,
+      "step": 1240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.115e-05,
+      "loss": 4.9938,
+      "step": 1250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.14e-05,
+      "loss": 5.0288,
+      "step": 1260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.165e-05,
+      "loss": 5.0438,
+      "step": 1270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1899999999999996e-05,
+      "loss": 4.9984,
+      "step": 1280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2149999999999995e-05,
+      "loss": 5.0142,
+      "step": 1290
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2399999999999995e-05,
+      "loss": 4.9647,
+      "step": 1300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.265e-05,
+      "loss": 4.9983,
+      "step": 1310
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.289999999999999e-05,
+      "loss": 4.9675,
+      "step": 1320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.315e-05,
+      "loss": 4.9787,
+      "step": 1330
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.34e-05,
+      "loss": 4.9638,
+      "step": 1340
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.365e-05,
+      "loss": 4.8791,
+      "step": 1350
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.39e-05,
+      "loss": 4.8835,
+      "step": 1360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4149999999999997e-05,
+      "loss": 4.8899,
+      "step": 1370
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4399999999999996e-05,
+      "loss": 4.8351,
+      "step": 1380
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.465e-05,
+      "loss": 4.8162,
+      "step": 1390
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4899999999999995e-05,
+      "loss": 4.8588,
+      "step": 1400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.515e-05,
+      "loss": 4.8019,
+      "step": 1410
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.539999999999999e-05,
+      "loss": 4.8557,
+      "step": 1420
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.565e-05,
+      "loss": 4.8064,
+      "step": 1430
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.59e-05,
+      "loss": 4.7612,
+      "step": 1440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.615e-05,
+      "loss": 4.7071,
+      "step": 1450
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.64e-05,
+      "loss": 4.727,
+      "step": 1460
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6649999999999996e-05,
+      "loss": 4.686,
+      "step": 1470
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6899999999999996e-05,
+      "loss": 4.6832,
+      "step": 1480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7149999999999995e-05,
+      "loss": 4.6862,
+      "step": 1490
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7399999999999994e-05,
+      "loss": 4.6609,
+      "step": 1500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7649999999999994e-05,
+      "loss": 4.6903,
+      "step": 1510
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.789999999999999e-05,
+      "loss": 4.6529,
+      "step": 1520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.815e-05,
+      "loss": 4.6384,
+      "step": 1530
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.84e-05,
+      "loss": 4.6419,
+      "step": 1540
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.865e-05,
+      "loss": 4.5833,
+      "step": 1550
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8900000000000004e-05,
+      "loss": 4.5477,
+      "step": 1560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.9149999999999996e-05,
+      "loss": 4.6188,
+      "step": 1570
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.9399999999999995e-05,
+      "loss": 4.5747,
+      "step": 1580
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.9649999999999995e-05,
+      "loss": 4.4917,
+      "step": 1590
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.99e-05,
+      "loss": 4.4842,
+      "step": 1600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.014999999999999e-05,
+      "loss": 4.5239,
+      "step": 1610
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.039999999999999e-05,
+      "loss": 4.4849,
+      "step": 1620
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.065e-05,
+      "loss": 4.4855,
+      "step": 1630
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.09e-05,
+      "loss": 4.4695,
+      "step": 1640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.115e-05,
+      "loss": 4.462,
+      "step": 1650
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.14e-05,
+      "loss": 4.4728,
+      "step": 1660
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1649999999999996e-05,
+      "loss": 4.4426,
+      "step": 1670
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1899999999999995e-05,
+      "loss": 4.4123,
+      "step": 1680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.215e-05,
+      "loss": 4.4026,
+      "step": 1690
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.24e-05,
+      "loss": 4.4942,
+      "step": 1700
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.264999999999999e-05,
+      "loss": 4.4393,
+      "step": 1710
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.289999999999999e-05,
+      "loss": 4.3781,
+      "step": 1720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.315e-05,
+      "loss": 4.36,
+      "step": 1730
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.34e-05,
+      "loss": 4.4202,
+      "step": 1740
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.364999999999999e-05,
+      "loss": 4.4418,
+      "step": 1750
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.3899999999999996e-05,
+      "loss": 4.3932,
+      "step": 1760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4149999999999996e-05,
+      "loss": 4.3306,
+      "step": 1770
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4399999999999995e-05,
+      "loss": 4.2984,
+      "step": 1780
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.465e-05,
+      "loss": 4.2832,
+      "step": 1790
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.49e-05,
+      "loss": 4.2793,
+      "step": 1800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.514999999999999e-05,
+      "loss": 4.2818,
+      "step": 1810
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.539999999999999e-05,
+      "loss": 4.2509,
+      "step": 1820
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.565e-05,
+      "loss": 4.2453,
+      "step": 1830
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.59e-05,
+      "loss": 4.1921,
+      "step": 1840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.614999999999999e-05,
+      "loss": 4.2396,
+      "step": 1850
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6399999999999996e-05,
+      "loss": 4.2299,
+      "step": 1860
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6649999999999996e-05,
+      "loss": 4.192,
+      "step": 1870
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6899999999999995e-05,
+      "loss": 4.2031,
+      "step": 1880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.715e-05,
+      "loss": 4.2628,
+      "step": 1890
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.7399999999999993e-05,
+      "loss": 4.2162,
+      "step": 1900
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.764999999999999e-05,
+      "loss": 4.1488,
+      "step": 1910
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.79e-05,
+      "loss": 4.1782,
+      "step": 1920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.815e-05,
+      "loss": 4.164,
+      "step": 1930
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.84e-05,
+      "loss": 4.1491,
+      "step": 1940
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.864999999999999e-05,
+      "loss": 4.1645,
+      "step": 1950
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8899999999999996e-05,
+      "loss": 4.1327,
+      "step": 1960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9149999999999995e-05,
+      "loss": 4.1126,
+      "step": 1970
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9399999999999995e-05,
+      "loss": 4.0764,
+      "step": 1980
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.965e-05,
+      "loss": 4.1517,
+      "step": 1990
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.989999999999999e-05,
+      "loss": 4.1227,
+      "step": 2000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.014999999999999e-05,
+      "loss": 4.0753,
+      "step": 2010
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.04e-05,
+      "loss": 4.0627,
+      "step": 2020
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.065e-05,
+      "loss": 4.0252,
+      "step": 2030
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.089999999999999e-05,
+      "loss": 4.0361,
+      "step": 2040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.1149999999999996e-05,
+      "loss": 4.0053,
+      "step": 2050
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.1399999999999996e-05,
+      "loss": 4.0252,
+      "step": 2060
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.1649999999999995e-05,
+      "loss": 4.001,
+      "step": 2070
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.1899999999999994e-05,
+      "loss": 3.9855,
+      "step": 2080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.215e-05,
+      "loss": 3.9878,
+      "step": 2090
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.239999999999999e-05,
+      "loss": 3.989,
+      "step": 2100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.264999999999999e-05,
+      "loss": 3.9635,
+      "step": 2110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.29e-05,
+      "loss": 4.0035,
+      "step": 2120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.315e-05,
+      "loss": 3.9413,
+      "step": 2130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.339999999999999e-05,
+      "loss": 3.8845,
+      "step": 2140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.3649999999999996e-05,
+      "loss": 3.8886,
+      "step": 2150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.3899999999999996e-05,
+      "loss": 3.9756,
+      "step": 2160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.4149999999999995e-05,
+      "loss": 3.9235,
+      "step": 2170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.44e-05,
+      "loss": 3.9015,
+      "step": 2180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.4649999999999993e-05,
+      "loss": 3.9052,
+      "step": 2190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.489999999999999e-05,
+      "loss": 3.9044,
+      "step": 2200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.514999999999999e-05,
+      "loss": 3.8846,
+      "step": 2210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.54e-05,
+      "loss": 3.902,
+      "step": 2220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.565e-05,
+      "loss": 3.8615,
+      "step": 2230
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.589999999999999e-05,
+      "loss": 3.8604,
+      "step": 2240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.6149999999999996e-05,
+      "loss": 3.8167,
+      "step": 2250
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.6399999999999995e-05,
+      "loss": 3.819,
+      "step": 2260
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.6649999999999995e-05,
+      "loss": 3.7915,
+      "step": 2270
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.69e-05,
+      "loss": 3.7781,
+      "step": 2280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.714999999999999e-05,
+      "loss": 3.7887,
+      "step": 2290
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.739999999999999e-05,
+      "loss": 3.8007,
+      "step": 2300
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.765e-05,
+      "loss": 3.7612,
+      "step": 2310
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.79e-05,
+      "loss": 3.75,
+      "step": 2320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.814999999999999e-05,
+      "loss": 3.7403,
+      "step": 2330
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.839999999999999e-05,
+      "loss": 3.7463,
+      "step": 2340
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.8649999999999996e-05,
+      "loss": 3.7643,
+      "step": 2350
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.8899999999999995e-05,
+      "loss": 3.7407,
+      "step": 2360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9149999999999994e-05,
+      "loss": 3.7601,
+      "step": 2370
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.94e-05,
+      "loss": 3.6923,
+      "step": 2380
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.964999999999999e-05,
+      "loss": 3.7234,
+      "step": 2390
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.989999999999999e-05,
+      "loss": 3.6752,
+      "step": 2400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.015e-05,
+      "loss": 3.6437,
+      "step": 2410
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.04e-05,
+      "loss": 3.6805,
+      "step": 2420
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.064999999999999e-05,
+      "loss": 3.7198,
+      "step": 2430
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.0899999999999996e-05,
+      "loss": 3.6588,
+      "step": 2440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.115e-05,
+      "loss": 3.6113,
+      "step": 2450
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.139999999999999e-05,
+      "loss": 3.6444,
+      "step": 2460
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.165e-05,
+      "loss": 3.6489,
+      "step": 2470
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.19e-05,
+      "loss": 3.5805,
+      "step": 2480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.214999999999999e-05,
+      "loss": 3.5343,
+      "step": 2490
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.239999999999999e-05,
+      "loss": 3.5642,
+      "step": 2500
+    },
+    {
+      "epoch": 0.03,
+      "eval_accuracy": 0.4383129848262597,
+      "eval_loss": 3.9296875,
+      "eval_runtime": 97.6812,
+      "eval_samples_per_second": 818.99,
+      "eval_steps_per_second": 1.607,
+      "step": 2500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.264999999999999e-05,
+      "loss": 3.5807,
+      "step": 2510
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.29e-05,
+      "loss": 3.5057,
+      "step": 2520
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.314999999999999e-05,
+      "loss": 3.5416,
+      "step": 2530
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.34e-05,
+      "loss": 3.5582,
+      "step": 2540
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.365e-05,
+      "loss": 3.5261,
+      "step": 2550
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.39e-05,
+      "loss": 3.5442,
+      "step": 2560
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.414999999999999e-05,
+      "loss": 3.5676,
+      "step": 2570
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.44e-05,
+      "loss": 3.514,
+      "step": 2580
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.465e-05,
+      "loss": 3.5178,
+      "step": 2590
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.489999999999999e-05,
+      "loss": 3.4848,
+      "step": 2600
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.515e-05,
+      "loss": 3.4749,
+      "step": 2610
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.539999999999999e-05,
+      "loss": 3.4313,
+      "step": 2620
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.565e-05,
+      "loss": 3.473,
+      "step": 2630
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.59e-05,
+      "loss": 3.4574,
+      "step": 2640
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.615e-05,
+      "loss": 3.4329,
+      "step": 2650
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.639999999999999e-05,
+      "loss": 3.4348,
+      "step": 2660
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.665e-05,
+      "loss": 3.4441,
+      "step": 2670
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.69e-05,
+      "loss": 3.4589,
+      "step": 2680
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.714999999999999e-05,
+      "loss": 3.4277,
+      "step": 2690
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.739999999999998e-05,
+      "loss": 3.4026,
+      "step": 2700
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.764999999999999e-05,
+      "loss": 3.3649,
+      "step": 2710
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.79e-05,
+      "loss": 3.356,
+      "step": 2720
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.814999999999999e-05,
+      "loss": 3.3409,
+      "step": 2730
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.84e-05,
+      "loss": 3.3401,
+      "step": 2740
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.864999999999999e-05,
+      "loss": 3.3325,
+      "step": 2750
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.89e-05,
+      "loss": 3.3165,
+      "step": 2760
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.915e-05,
+      "loss": 3.3381,
+      "step": 2770
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.939999999999999e-05,
+      "loss": 3.3134,
+      "step": 2780
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.964999999999999e-05,
+      "loss": 3.3009,
+      "step": 2790
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.989999999999999e-05,
+      "loss": 3.289,
+      "step": 2800
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.015e-05,
+      "loss": 3.2767,
+      "step": 2810
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.039999999999999e-05,
+      "loss": 3.2677,
+      "step": 2820
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.065e-05,
+      "loss": 3.2375,
+      "step": 2830
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.09e-05,
+      "loss": 3.2508,
+      "step": 2840
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.115e-05,
+      "loss": 3.2772,
+      "step": 2850
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.139999999999999e-05,
+      "loss": 3.1986,
+      "step": 2860
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.165e-05,
+      "loss": 3.2352,
+      "step": 2870
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.19e-05,
+      "loss": 3.2138,
+      "step": 2880
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.214999999999999e-05,
+      "loss": 3.207,
+      "step": 2890
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.24e-05,
+      "loss": 3.189,
+      "step": 2900
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.264999999999999e-05,
+      "loss": 3.129,
+      "step": 2910
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.29e-05,
+      "loss": 3.149,
+      "step": 2920
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.315e-05,
+      "loss": 3.167,
+      "step": 2930
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.34e-05,
+      "loss": 3.1163,
+      "step": 2940
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.364999999999999e-05,
+      "loss": 3.1226,
+      "step": 2950
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.39e-05,
+      "loss": 3.08,
+      "step": 2960
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.415e-05,
+      "loss": 3.1136,
+      "step": 2970
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.439999999999999e-05,
+      "loss": 3.0943,
+      "step": 2980
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.464999999999998e-05,
+      "loss": 3.0379,
+      "step": 2990
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.489999999999999e-05,
+      "loss": 3.0346,
+      "step": 3000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.515e-05,
+      "loss": 3.0365,
+      "step": 3010
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.54e-05,
+      "loss": 3.0473,
+      "step": 3020
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.564999999999998e-05,
+      "loss": 3.0086,
+      "step": 3030
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.589999999999999e-05,
+      "loss": 3.0474,
+      "step": 3040
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.615e-05,
+      "loss": 3.0107,
+      "step": 3050
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.639999999999999e-05,
+      "loss": 2.9755,
+      "step": 3060
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.664999999999999e-05,
+      "loss": 2.9917,
+      "step": 3070
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.69e-05,
+      "loss": 2.952,
+      "step": 3080
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.714999999999999e-05,
+      "loss": 2.9553,
+      "step": 3090
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.74e-05,
+      "loss": 2.9484,
+      "step": 3100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.765e-05,
+      "loss": 3.0178,
+      "step": 3110
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.79e-05,
+      "loss": 2.9621,
+      "step": 3120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.815e-05,
+      "loss": 2.9268,
+      "step": 3130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.839999999999998e-05,
+      "loss": 2.9683,
+      "step": 3140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.864999999999999e-05,
+      "loss": 2.9026,
+      "step": 3150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.89e-05,
+      "loss": 2.8892,
+      "step": 3160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.914999999999999e-05,
+      "loss": 2.8845,
+      "step": 3170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.939999999999999e-05,
+      "loss": 2.8826,
+      "step": 3180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.965e-05,
+      "loss": 2.8888,
+      "step": 3190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.989999999999999e-05,
+      "loss": 2.8068,
+      "step": 3200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.015e-05,
+      "loss": 2.8483,
+      "step": 3210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.04e-05,
+      "loss": 2.8422,
+      "step": 3220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.064999999999998e-05,
+      "loss": 2.8175,
+      "step": 3230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.089999999999999e-05,
+      "loss": 2.8425,
+      "step": 3240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.115e-05,
+      "loss": 2.826,
+      "step": 3250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.139999999999999e-05,
+      "loss": 2.7938,
+      "step": 3260
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.164999999999999e-05,
+      "loss": 2.7908,
+      "step": 3270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.19e-05,
+      "loss": 2.7734,
+      "step": 3280
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.214999999999999e-05,
+      "loss": 2.8051,
+      "step": 3290
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.24e-05,
+      "loss": 2.7474,
+      "step": 3300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.265e-05,
+      "loss": 2.7743,
+      "step": 3310
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.289999999999998e-05,
+      "loss": 2.7415,
+      "step": 3320
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.314999999999999e-05,
+      "loss": 2.7359,
+      "step": 3330
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.34e-05,
+      "loss": 2.7306,
+      "step": 3340
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.364999999999999e-05,
+      "loss": 2.7382,
+      "step": 3350
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.389999999999999e-05,
+      "loss": 2.7887,
+      "step": 3360
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.415e-05,
+      "loss": 2.7417,
+      "step": 3370
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.439999999999999e-05,
+      "loss": 2.7036,
+      "step": 3380
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.465e-05,
+      "loss": 2.6724,
+      "step": 3390
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.489999999999999e-05,
+      "loss": 2.6508,
+      "step": 3400
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.515e-05,
+      "loss": 2.6735,
+      "step": 3410
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.54e-05,
+      "loss": 2.652,
+      "step": 3420
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.564999999999998e-05,
+      "loss": 2.6687,
+      "step": 3430
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.589999999999999e-05,
+      "loss": 2.6581,
+      "step": 3440
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.615e-05,
+      "loss": 2.6346,
+      "step": 3450
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.639999999999999e-05,
+      "loss": 2.6429,
+      "step": 3460
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.664999999999999e-05,
+      "loss": 2.6137,
+      "step": 3470
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.69e-05,
+      "loss": 2.6249,
+      "step": 3480
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.714999999999999e-05,
+      "loss": 2.6009,
+      "step": 3490
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.74e-05,
+      "loss": 2.6004,
+      "step": 3500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.765e-05,
+      "loss": 2.5755,
+      "step": 3510
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.789999999999998e-05,
+      "loss": 2.5508,
+      "step": 3520
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.814999999999999e-05,
+      "loss": 2.5676,
+      "step": 3530
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.84e-05,
+      "loss": 2.5596,
+      "step": 3540
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.864999999999999e-05,
+      "loss": 2.5754,
+      "step": 3550
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.889999999999999e-05,
+      "loss": 2.5696,
+      "step": 3560
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.915e-05,
+      "loss": 2.542,
+      "step": 3570
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.939999999999999e-05,
+      "loss": 2.5561,
+      "step": 3580
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.965e-05,
+      "loss": 2.5059,
+      "step": 3590
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.99e-05,
+      "loss": 2.5288,
+      "step": 3600
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.014999999999998e-05,
+      "loss": 2.5039,
+      "step": 3610
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.039999999999999e-05,
+      "loss": 2.5243,
+      "step": 3620
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.064999999999998e-05,
+      "loss": 2.4772,
+      "step": 3630
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.089999999999999e-05,
+      "loss": 2.4543,
+      "step": 3640
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.114999999999999e-05,
+      "loss": 2.4743,
+      "step": 3650
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.139999999999999e-05,
+      "loss": 2.488,
+      "step": 3660
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.164999999999999e-05,
+      "loss": 2.4918,
+      "step": 3670
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.19e-05,
+      "loss": 2.4866,
+      "step": 3680
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.214999999999999e-05,
+      "loss": 2.4649,
+      "step": 3690
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.24e-05,
+      "loss": 2.4613,
+      "step": 3700
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.265e-05,
+      "loss": 2.4337,
+      "step": 3710
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.289999999999998e-05,
+      "loss": 2.4073,
+      "step": 3720
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.314999999999999e-05,
+      "loss": 2.4314,
+      "step": 3730
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.34e-05,
+      "loss": 2.4311,
+      "step": 3740
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.364999999999999e-05,
+      "loss": 2.4553,
+      "step": 3750
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.389999999999999e-05,
+      "loss": 2.4242,
+      "step": 3760
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.415e-05,
+      "loss": 2.3964,
+      "step": 3770
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.439999999999999e-05,
+      "loss": 2.3865,
+      "step": 3780
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.465e-05,
+      "loss": 2.4134,
+      "step": 3790
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.49e-05,
+      "loss": 2.3987,
+      "step": 3800
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.514999999999998e-05,
+      "loss": 2.3879,
+      "step": 3810
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.539999999999999e-05,
+      "loss": 2.3892,
+      "step": 3820
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.565e-05,
+      "loss": 2.3583,
+      "step": 3830
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.589999999999999e-05,
+      "loss": 2.3354,
+      "step": 3840
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.614999999999999e-05,
+      "loss": 2.3411,
+      "step": 3850
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.64e-05,
+      "loss": 2.3319,
+      "step": 3860
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.664999999999999e-05,
+      "loss": 2.339,
+      "step": 3870
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.69e-05,
+      "loss": 2.3017,
+      "step": 3880
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.714999999999999e-05,
+      "loss": 2.3514,
+      "step": 3890
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.74e-05,
+      "loss": 2.3212,
+      "step": 3900
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.764999999999999e-05,
+      "loss": 2.3247,
+      "step": 3910
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.789999999999998e-05,
+      "loss": 2.3187,
+      "step": 3920
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.814999999999999e-05,
+      "loss": 2.302,
+      "step": 3930
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.839999999999999e-05,
+      "loss": 2.3278,
+      "step": 3940
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.864999999999999e-05,
+      "loss": 2.321,
+      "step": 3950
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.889999999999999e-05,
+      "loss": 2.2785,
+      "step": 3960
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.915e-05,
+      "loss": 2.2715,
+      "step": 3970
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.939999999999999e-05,
+      "loss": 2.2599,
+      "step": 3980
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.965e-05,
+      "loss": 2.2777,
+      "step": 3990
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99e-05,
+      "loss": 2.2513,
+      "step": 4000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010014999999999998,
+      "loss": 2.2255,
+      "step": 4010
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010039999999999999,
+      "loss": 2.2582,
+      "step": 4020
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010065,
+      "loss": 2.2871,
+      "step": 4030
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010089999999999999,
+      "loss": 2.2631,
+      "step": 4040
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010114999999999999,
+      "loss": 2.2364,
+      "step": 4050
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001014,
+      "loss": 2.2495,
+      "step": 4060
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010164999999999999,
+      "loss": 2.218,
+      "step": 4070
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001019,
+      "loss": 2.2098,
+      "step": 4080
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010215,
+      "loss": 2.2088,
+      "step": 4090
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010239999999999998,
+      "loss": 2.1903,
+      "step": 4100
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010264999999999999,
+      "loss": 2.2031,
+      "step": 4110
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001029,
+      "loss": 2.1972,
+      "step": 4120
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010314999999999999,
+      "loss": 2.1614,
+      "step": 4130
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010339999999999999,
+      "loss": 2.1714,
+      "step": 4140
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010364999999999999,
+      "loss": 2.1924,
+      "step": 4150
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010389999999999999,
+      "loss": 2.1667,
+      "step": 4160
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010415,
+      "loss": 2.1696,
+      "step": 4170
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010439999999999999,
+      "loss": 2.1879,
+      "step": 4180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010465,
+      "loss": 2.1699,
+      "step": 4190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001049,
+      "loss": 2.1627,
+      "step": 4200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010514999999999998,
+      "loss": 2.1355,
+      "step": 4210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010539999999999999,
+      "loss": 2.1718,
+      "step": 4220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010564999999999999,
+      "loss": 2.1131,
+      "step": 4230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010589999999999999,
+      "loss": 2.099,
+      "step": 4240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010614999999999999,
+      "loss": 2.138,
+      "step": 4250
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001064,
+      "loss": 2.1346,
+      "step": 4260
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010664999999999999,
+      "loss": 2.1164,
+      "step": 4270
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001069,
+      "loss": 2.1027,
+      "step": 4280
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010715,
+      "loss": 2.1362,
+      "step": 4290
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010739999999999998,
+      "loss": 2.1061,
+      "step": 4300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010764999999999999,
+      "loss": 2.097,
+      "step": 4310
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001079,
+      "loss": 2.0938,
+      "step": 4320
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010814999999999999,
+      "loss": 2.081,
+      "step": 4330
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010839999999999999,
+      "loss": 2.0956,
+      "step": 4340
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010865,
+      "loss": 2.0812,
+      "step": 4350
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010889999999999999,
+      "loss": 2.0977,
+      "step": 4360
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010915,
+      "loss": 2.0734,
+      "step": 4370
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010939999999999998,
+      "loss": 2.0834,
+      "step": 4380
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010964999999999998,
+      "loss": 2.0577,
+      "step": 4390
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010989999999999999,
+      "loss": 2.0629,
+      "step": 4400
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00011014999999999998,
+      "loss": 2.0315,
+      "step": 4410
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00011039999999999999,
+      "loss": 2.0288,
+      "step": 4420
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00011064999999999999,
+      "loss": 2.0043,
+      "step": 4430
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00011089999999999999,
+      "loss": 2.0263,
+      "step": 4440
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00011114999999999999,
+      "loss": 2.0275,
+      "step": 4450
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001114,
+      "loss": 2.0508,
+      "step": 4460
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00011164999999999999,
+      "loss": 2.0482,
+      "step": 4470
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001119,
+      "loss": 2.0245,
+      "step": 4480
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00011215,
+      "loss": 2.0276,
+      "step": 4490
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00011239999999999998,
+      "loss": 2.0157,
+      "step": 4500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00011264999999999999,
+      "loss": 2.0158,
+      "step": 4510
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00011289999999999999,
+      "loss": 1.9956,
+      "step": 4520
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00011314999999999999,
+      "loss": 2.0053,
+      "step": 4530
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00011339999999999999,
+      "loss": 1.9749,
+      "step": 4540
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00011365,
+      "loss": 1.9494,
+      "step": 4550
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00011389999999999999,
+      "loss": 1.9653,
+      "step": 4560
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00011415,
+      "loss": 1.9768,
+      "step": 4570
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001144,
+      "loss": 1.9863,
+      "step": 4580
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00011464999999999998,
+      "loss": 1.9441,
+      "step": 4590
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00011489999999999999,
+      "loss": 1.9764,
+      "step": 4600
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00011515,
+      "loss": 1.9818,
+      "step": 4610
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00011539999999999999,
+      "loss": 1.9608,
+      "step": 4620
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00011564999999999999,
+      "loss": 1.9741,
+      "step": 4630
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00011589999999999998,
+      "loss": 1.9437,
+      "step": 4640
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00011614999999999999,
+      "loss": 1.9227,
+      "step": 4650
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001164,
+      "loss": 1.9289,
+      "step": 4660
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00011664999999999998,
+      "loss": 1.9195,
+      "step": 4670
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00011689999999999998,
+      "loss": 1.9182,
+      "step": 4680
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00011714999999999999,
+      "loss": 1.8692,
+      "step": 4690
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00011739999999999998,
+      "loss": 1.909,
+      "step": 4700
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00011762499999999999,
+      "loss": 1.931,
+      "step": 4710
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00011787499999999999,
+      "loss": 1.8903,
+      "step": 4720
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00011812499999999998,
+      "loss": 1.9148,
+      "step": 4730
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00011837499999999999,
+      "loss": 1.9032,
+      "step": 4740
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000118625,
+      "loss": 1.8977,
+      "step": 4750
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00011887499999999999,
+      "loss": 1.8749,
+      "step": 4760
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000119125,
+      "loss": 1.8801,
+      "step": 4770
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00011937499999999999,
+      "loss": 1.8636,
+      "step": 4780
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000119625,
+      "loss": 1.8592,
+      "step": 4790
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000119875,
+      "loss": 1.8598,
+      "step": 4800
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00012012499999999998,
+      "loss": 1.8642,
+      "step": 4810
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00012037499999999999,
+      "loss": 1.8885,
+      "step": 4820
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00012062499999999999,
+      "loss": 1.8736,
+      "step": 4830
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00012087499999999998,
+      "loss": 1.8383,
+      "step": 4840
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00012112499999999999,
+      "loss": 1.861,
+      "step": 4850
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000121375,
+      "loss": 1.8416,
+      "step": 4860
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00012162499999999999,
+      "loss": 1.8399,
+      "step": 4870
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000121875,
+      "loss": 1.8497,
+      "step": 4880
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000122125,
+      "loss": 1.8374,
+      "step": 4890
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00012237499999999998,
+      "loss": 1.8468,
+      "step": 4900
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00012262499999999999,
+      "loss": 1.814,
+      "step": 4910
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000122875,
+      "loss": 1.8219,
+      "step": 4920
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000123125,
+      "loss": 1.8261,
+      "step": 4930
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000123375,
+      "loss": 1.8272,
+      "step": 4940
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00012362499999999998,
+      "loss": 1.822,
+      "step": 4950
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000123875,
+      "loss": 1.8071,
+      "step": 4960
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000124125,
+      "loss": 1.825,
+      "step": 4970
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000124375,
+      "loss": 1.8145,
+      "step": 4980
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00012462499999999998,
+      "loss": 1.8154,
+      "step": 4990
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00012487499999999999,
+      "loss": 1.7897,
+      "step": 5000
+    },
+    {
+      "epoch": 0.05,
+      "eval_accuracy": 0.6654779469588099,
+      "eval_loss": 2.021484375,
+      "eval_runtime": 98.4067,
+      "eval_samples_per_second": 812.953,
+      "eval_steps_per_second": 1.595,
+      "step": 5000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000125125,
+      "loss": 1.7843,
+      "step": 5010
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000125375,
+      "loss": 1.7975,
+      "step": 5020
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000125625,
+      "loss": 1.7791,
+      "step": 5030
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00012587499999999998,
+      "loss": 1.8006,
+      "step": 5040
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000126125,
+      "loss": 1.7863,
+      "step": 5050
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000126375,
+      "loss": 1.7956,
+      "step": 5060
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00012662499999999997,
+      "loss": 1.7842,
+      "step": 5070
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00012687499999999998,
+      "loss": 1.7654,
+      "step": 5080
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000127125,
+      "loss": 1.7536,
+      "step": 5090
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000127375,
+      "loss": 1.7645,
+      "step": 5100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000127625,
+      "loss": 1.7458,
+      "step": 5110
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000127875,
+      "loss": 1.7508,
+      "step": 5120
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00012812499999999998,
+      "loss": 1.7588,
+      "step": 5130
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000128375,
+      "loss": 1.7645,
+      "step": 5140
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000128625,
+      "loss": 1.7593,
+      "step": 5150
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00012887499999999997,
+      "loss": 1.7307,
+      "step": 5160
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00012912499999999998,
+      "loss": 1.7578,
+      "step": 5170
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000129375,
+      "loss": 1.7509,
+      "step": 5180
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000129625,
+      "loss": 1.7421,
+      "step": 5190
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000129875,
+      "loss": 1.7378,
+      "step": 5200
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000130125,
+      "loss": 1.7253,
+      "step": 5210
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00013037499999999998,
+      "loss": 1.7089,
+      "step": 5220
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000130625,
+      "loss": 1.6957,
+      "step": 5230
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000130875,
+      "loss": 1.6984,
+      "step": 5240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00013112499999999998,
+      "loss": 1.7067,
+      "step": 5250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00013137499999999998,
+      "loss": 1.7153,
+      "step": 5260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000131625,
+      "loss": 1.7003,
+      "step": 5270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000131875,
+      "loss": 1.7009,
+      "step": 5280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000132125,
+      "loss": 1.7018,
+      "step": 5290
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00013237499999999998,
+      "loss": 1.7049,
+      "step": 5300
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00013262499999999998,
+      "loss": 1.6905,
+      "step": 5310
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000132875,
+      "loss": 1.6874,
+      "step": 5320
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00013312499999999997,
+      "loss": 1.6798,
+      "step": 5330
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00013337499999999998,
+      "loss": 1.681,
+      "step": 5340
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00013362499999999998,
+      "loss": 1.6508,
+      "step": 5350
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000133875,
+      "loss": 1.6663,
+      "step": 5360
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000134125,
+      "loss": 1.6725,
+      "step": 5370
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000134375,
+      "loss": 1.6529,
+      "step": 5380
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00013462499999999998,
+      "loss": 1.6667,
+      "step": 5390
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00013487499999999999,
+      "loss": 1.6653,
+      "step": 5400
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000135125,
+      "loss": 1.6569,
+      "step": 5410
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00013537499999999997,
+      "loss": 1.667,
+      "step": 5420
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00013562499999999998,
+      "loss": 1.6484,
+      "step": 5430
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00013587499999999998,
+      "loss": 1.621,
+      "step": 5440
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000136125,
+      "loss": 1.6436,
+      "step": 5450
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000136375,
+      "loss": 1.633,
+      "step": 5460
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000136625,
+      "loss": 1.6295,
+      "step": 5470
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00013687499999999998,
+      "loss": 1.628,
+      "step": 5480
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00013712499999999999,
+      "loss": 1.6232,
+      "step": 5490
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000137375,
+      "loss": 1.6255,
+      "step": 5500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000137625,
+      "loss": 1.6432,
+      "step": 5510
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000137875,
+      "loss": 1.6279,
+      "step": 5520
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00013812499999999998,
+      "loss": 1.6173,
+      "step": 5530
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000138375,
+      "loss": 1.614,
+      "step": 5540
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000138625,
+      "loss": 1.613,
+      "step": 5550
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00013887499999999997,
+      "loss": 1.6166,
+      "step": 5560
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00013912499999999998,
+      "loss": 1.6158,
+      "step": 5570
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00013937499999999999,
+      "loss": 1.6133,
+      "step": 5580
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000139625,
+      "loss": 1.5946,
+      "step": 5590
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000139875,
+      "loss": 1.6178,
+      "step": 5600
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000140125,
+      "loss": 1.6214,
+      "step": 5610
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00014037499999999998,
+      "loss": 1.6187,
+      "step": 5620
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000140625,
+      "loss": 1.6119,
+      "step": 5630
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000140875,
+      "loss": 1.5932,
+      "step": 5640
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00014112499999999997,
+      "loss": 1.6099,
+      "step": 5650
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00014137499999999998,
+      "loss": 1.5841,
+      "step": 5660
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000141625,
+      "loss": 1.5941,
+      "step": 5670
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000141875,
+      "loss": 1.5714,
+      "step": 5680
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000142125,
+      "loss": 1.5596,
+      "step": 5690
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000142375,
+      "loss": 1.5825,
+      "step": 5700
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001426,
+      "loss": 1.5837,
+      "step": 5710
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00014285,
+      "loss": 1.5929,
+      "step": 5720
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00014309999999999998,
+      "loss": 1.5859,
+      "step": 5730
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00014335,
+      "loss": 1.5909,
+      "step": 5740
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001436,
+      "loss": 1.5944,
+      "step": 5750
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00014384999999999997,
+      "loss": 1.5592,
+      "step": 5760
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00014409999999999998,
+      "loss": 1.5771,
+      "step": 5770
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00014434999999999999,
+      "loss": 1.5739,
+      "step": 5780
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001446,
+      "loss": 1.5645,
+      "step": 5790
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00014485,
+      "loss": 1.5484,
+      "step": 5800
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001451,
+      "loss": 1.5657,
+      "step": 5810
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00014534999999999998,
+      "loss": 1.5617,
+      "step": 5820
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001456,
+      "loss": 1.5707,
+      "step": 5830
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00014585,
+      "loss": 1.5512,
+      "step": 5840
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00014609999999999997,
+      "loss": 1.5669,
+      "step": 5850
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00014634999999999998,
+      "loss": 1.5415,
+      "step": 5860
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00014659999999999999,
+      "loss": 1.5404,
+      "step": 5870
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00014685,
+      "loss": 1.515,
+      "step": 5880
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001471,
+      "loss": 1.5422,
+      "step": 5890
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00014734999999999998,
+      "loss": 1.496,
+      "step": 5900
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00014759999999999998,
+      "loss": 1.4912,
+      "step": 5910
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00014785,
+      "loss": 1.5192,
+      "step": 5920
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00014809999999999997,
+      "loss": 1.5174,
+      "step": 5930
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00014834999999999997,
+      "loss": 1.5201,
+      "step": 5940
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00014859999999999998,
+      "loss": 1.5123,
+      "step": 5950
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00014884999999999999,
+      "loss": 1.5244,
+      "step": 5960
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001491,
+      "loss": 1.5144,
+      "step": 5970
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00014935,
+      "loss": 1.514,
+      "step": 5980
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00014959999999999998,
+      "loss": 1.4987,
+      "step": 5990
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00014984999999999998,
+      "loss": 1.5266,
+      "step": 6000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001501,
+      "loss": 1.5074,
+      "step": 6010
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00015034999999999997,
+      "loss": 1.4996,
+      "step": 6020
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00015059999999999997,
+      "loss": 1.5075,
+      "step": 6030
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00015084999999999998,
+      "loss": 1.4872,
+      "step": 6040
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001511,
+      "loss": 1.5025,
+      "step": 6050
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00015134999999999997,
+      "loss": 1.4719,
+      "step": 6060
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00015159999999999997,
+      "loss": 1.502,
+      "step": 6070
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00015184999999999998,
+      "loss": 1.5157,
+      "step": 6080
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00015209999999999998,
+      "loss": 1.4907,
+      "step": 6090
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00015235,
+      "loss": 1.4747,
+      "step": 6100
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001526,
+      "loss": 1.4808,
+      "step": 6110
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00015284999999999997,
+      "loss": 1.4805,
+      "step": 6120
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00015309999999999998,
+      "loss": 1.4615,
+      "step": 6130
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00015335,
+      "loss": 1.4887,
+      "step": 6140
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001536,
+      "loss": 1.453,
+      "step": 6150
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00015385,
+      "loss": 1.463,
+      "step": 6160
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001541,
+      "loss": 1.4306,
+      "step": 6170
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00015434999999999998,
+      "loss": 1.469,
+      "step": 6180
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001546,
+      "loss": 1.4942,
+      "step": 6190
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00015485,
+      "loss": 1.4675,
+      "step": 6200
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001551,
+      "loss": 1.4585,
+      "step": 6210
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00015535,
+      "loss": 1.467,
+      "step": 6220
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00015560000000000001,
+      "loss": 1.4841,
+      "step": 6230
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00015584999999999997,
+      "loss": 1.4545,
+      "step": 6240
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00015609999999999997,
+      "loss": 1.4503,
+      "step": 6250
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00015634999999999998,
+      "loss": 1.447,
+      "step": 6260
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00015659999999999998,
+      "loss": 1.4531,
+      "step": 6270
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00015685,
+      "loss": 1.4416,
+      "step": 6280
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00015709999999999997,
+      "loss": 1.4706,
+      "step": 6290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00015734999999999998,
+      "loss": 1.4469,
+      "step": 6300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00015759999999999998,
+      "loss": 1.4523,
+      "step": 6310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00015785,
+      "loss": 1.4449,
+      "step": 6320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001581,
+      "loss": 1.4422,
+      "step": 6330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00015835,
+      "loss": 1.4359,
+      "step": 6340
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00015859999999999998,
+      "loss": 1.4429,
+      "step": 6350
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00015884999999999999,
+      "loss": 1.4271,
+      "step": 6360
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001591,
+      "loss": 1.4198,
+      "step": 6370
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00015935,
+      "loss": 1.4261,
+      "step": 6380
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001596,
+      "loss": 1.4375,
+      "step": 6390
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00015985,
+      "loss": 1.4319,
+      "step": 6400
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00016009999999999996,
+      "loss": 1.4446,
+      "step": 6410
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00016034999999999997,
+      "loss": 1.4272,
+      "step": 6420
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00016059999999999997,
+      "loss": 1.4311,
+      "step": 6430
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00016084999999999998,
+      "loss": 1.4489,
+      "step": 6440
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00016109999999999999,
+      "loss": 1.4271,
+      "step": 6450
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00016135,
+      "loss": 1.4196,
+      "step": 6460
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00016159999999999997,
+      "loss": 1.3898,
+      "step": 6470
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00016184999999999998,
+      "loss": 1.4023,
+      "step": 6480
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00016209999999999998,
+      "loss": 1.4047,
+      "step": 6490
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016235,
+      "loss": 1.3904,
+      "step": 6500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001626,
+      "loss": 1.4102,
+      "step": 6510
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016284999999999997,
+      "loss": 1.3983,
+      "step": 6520
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016309999999999998,
+      "loss": 1.4131,
+      "step": 6530
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016334999999999999,
+      "loss": 1.4131,
+      "step": 6540
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001636,
+      "loss": 1.4097,
+      "step": 6550
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016385,
+      "loss": 1.3955,
+      "step": 6560
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001641,
+      "loss": 1.3888,
+      "step": 6570
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016434999999999998,
+      "loss": 1.3908,
+      "step": 6580
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001646,
+      "loss": 1.3946,
+      "step": 6590
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016485,
+      "loss": 1.3861,
+      "step": 6600
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016509999999999997,
+      "loss": 1.389,
+      "step": 6610
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016534999999999998,
+      "loss": 1.3817,
+      "step": 6620
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001656,
+      "loss": 1.3813,
+      "step": 6630
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016584999999999997,
+      "loss": 1.402,
+      "step": 6640
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016609999999999997,
+      "loss": 1.3934,
+      "step": 6650
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016634999999999998,
+      "loss": 1.3689,
+      "step": 6660
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016659999999999998,
+      "loss": 1.3701,
+      "step": 6670
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016685,
+      "loss": 1.365,
+      "step": 6680
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001671,
+      "loss": 1.3718,
+      "step": 6690
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016734999999999997,
+      "loss": 1.3546,
+      "step": 6700
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000167575,
+      "loss": 1.3649,
+      "step": 6710
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000167825,
+      "loss": 1.3595,
+      "step": 6720
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000168075,
+      "loss": 1.3644,
+      "step": 6730
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000168325,
+      "loss": 1.3749,
+      "step": 6740
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016857499999999996,
+      "loss": 1.3693,
+      "step": 6750
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016882499999999997,
+      "loss": 1.3468,
+      "step": 6760
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016907499999999998,
+      "loss": 1.3447,
+      "step": 6770
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016932499999999998,
+      "loss": 1.3418,
+      "step": 6780
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000169575,
+      "loss": 1.3477,
+      "step": 6790
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000169825,
+      "loss": 1.3519,
+      "step": 6800
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00017007499999999997,
+      "loss": 1.3188,
+      "step": 6810
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00017032499999999998,
+      "loss": 1.3433,
+      "step": 6820
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00017057499999999999,
+      "loss": 1.3405,
+      "step": 6830
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000170825,
+      "loss": 1.3375,
+      "step": 6840
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000171075,
+      "loss": 1.3299,
+      "step": 6850
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000171325,
+      "loss": 1.3495,
+      "step": 6860
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00017157499999999998,
+      "loss": 1.3485,
+      "step": 6870
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000171825,
+      "loss": 1.3383,
+      "step": 6880
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000172075,
+      "loss": 1.3346,
+      "step": 6890
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000172325,
+      "loss": 1.3392,
+      "step": 6900
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000172575,
+      "loss": 1.3336,
+      "step": 6910
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00017282499999999996,
+      "loss": 1.3228,
+      "step": 6920
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00017307499999999996,
+      "loss": 1.3324,
+      "step": 6930
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00017332499999999997,
+      "loss": 1.3237,
+      "step": 6940
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00017357499999999998,
+      "loss": 1.3331,
+      "step": 6950
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00017382499999999998,
+      "loss": 1.3495,
+      "step": 6960
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000174075,
+      "loss": 1.3498,
+      "step": 6970
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00017432499999999997,
+      "loss": 1.3428,
+      "step": 6980
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00017457499999999997,
+      "loss": 1.3347,
+      "step": 6990
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00017482499999999998,
+      "loss": 1.3182,
+      "step": 7000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000175075,
+      "loss": 1.3255,
+      "step": 7010
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000175325,
+      "loss": 1.3057,
+      "step": 7020
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000175575,
+      "loss": 1.3166,
+      "step": 7030
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00017582499999999998,
+      "loss": 1.3268,
+      "step": 7040
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00017607499999999998,
+      "loss": 1.3121,
+      "step": 7050
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000176325,
+      "loss": 1.301,
+      "step": 7060
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000176575,
+      "loss": 1.3167,
+      "step": 7070
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000176825,
+      "loss": 1.3102,
+      "step": 7080
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000177075,
+      "loss": 1.329,
+      "step": 7090
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00017732499999999996,
+      "loss": 1.3241,
+      "step": 7100
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00017757499999999997,
+      "loss": 1.3216,
+      "step": 7110
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00017782499999999997,
+      "loss": 1.3041,
+      "step": 7120
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00017807499999999998,
+      "loss": 1.2956,
+      "step": 7130
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00017832499999999998,
+      "loss": 1.3219,
+      "step": 7140
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00017857499999999996,
+      "loss": 1.2878,
+      "step": 7150
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00017882499999999997,
+      "loss": 1.2961,
+      "step": 7160
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00017907499999999998,
+      "loss": 1.2876,
+      "step": 7170
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00017932499999999998,
+      "loss": 1.3088,
+      "step": 7180
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000179575,
+      "loss": 1.296,
+      "step": 7190
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000179825,
+      "loss": 1.3052,
+      "step": 7200
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00018007499999999997,
+      "loss": 1.3145,
+      "step": 7210
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00018032499999999998,
+      "loss": 1.2777,
+      "step": 7220
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00018057499999999998,
+      "loss": 1.2916,
+      "step": 7230
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000180825,
+      "loss": 1.2826,
+      "step": 7240
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000181075,
+      "loss": 1.2731,
+      "step": 7250
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000181325,
+      "loss": 1.2718,
+      "step": 7260
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00018157499999999998,
+      "loss": 1.2597,
+      "step": 7270
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000181825,
+      "loss": 1.2753,
+      "step": 7280
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000182075,
+      "loss": 1.2755,
+      "step": 7290
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000182325,
+      "loss": 1.2735,
+      "step": 7300
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000182575,
+      "loss": 1.269,
+      "step": 7310
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000182825,
+      "loss": 1.2967,
+      "step": 7320
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00018307499999999996,
+      "loss": 1.283,
+      "step": 7330
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00018332499999999997,
+      "loss": 1.2773,
+      "step": 7340
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00018357499999999998,
+      "loss": 1.2836,
+      "step": 7350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00018382499999999998,
+      "loss": 1.272,
+      "step": 7360
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000184075,
+      "loss": 1.2588,
+      "step": 7370
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000184325,
+      "loss": 1.2535,
+      "step": 7380
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00018457499999999997,
+      "loss": 1.2585,
+      "step": 7390
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00018482499999999998,
+      "loss": 1.2544,
+      "step": 7400
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00018507499999999999,
+      "loss": 1.236,
+      "step": 7410
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000185325,
+      "loss": 1.26,
+      "step": 7420
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000185575,
+      "loss": 1.2725,
+      "step": 7430
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00018582499999999998,
+      "loss": 1.2621,
+      "step": 7440
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00018607499999999998,
+      "loss": 1.2566,
+      "step": 7450
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000186325,
+      "loss": 1.2642,
+      "step": 7460
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000186575,
+      "loss": 1.2521,
+      "step": 7470
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000186825,
+      "loss": 1.2418,
+      "step": 7480
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000187075,
+      "loss": 1.2386,
+      "step": 7490
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00018732499999999996,
+      "loss": 1.2496,
+      "step": 7500
+    },
+    {
+      "epoch": 0.07,
+      "eval_accuracy": 0.7488450887376084,
+      "eval_loss": 1.373046875,
+      "eval_runtime": 97.1795,
+      "eval_samples_per_second": 823.219,
+      "eval_steps_per_second": 1.616,
+      "step": 7500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00018757499999999997,
+      "loss": 1.2656,
+      "step": 7510
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00018782499999999997,
+      "loss": 1.2877,
+      "step": 7520
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00018807499999999998,
+      "loss": 1.2559,
+      "step": 7530
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00018832499999999998,
+      "loss": 1.289,
+      "step": 7540
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000188575,
+      "loss": 1.2772,
+      "step": 7550
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00018882499999999997,
+      "loss": 1.2847,
+      "step": 7560
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00018907499999999997,
+      "loss": 1.2564,
+      "step": 7570
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00018932499999999998,
+      "loss": 1.276,
+      "step": 7580
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000189575,
+      "loss": 1.2677,
+      "step": 7590
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000189825,
+      "loss": 1.2607,
+      "step": 7600
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000190075,
+      "loss": 1.2511,
+      "step": 7610
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019032499999999998,
+      "loss": 1.2928,
+      "step": 7620
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019057499999999998,
+      "loss": 1.2494,
+      "step": 7630
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000190825,
+      "loss": 1.2637,
+      "step": 7640
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000191075,
+      "loss": 1.2602,
+      "step": 7650
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000191325,
+      "loss": 1.2572,
+      "step": 7660
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019157499999999995,
+      "loss": 1.2685,
+      "step": 7670
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019182499999999996,
+      "loss": 1.2509,
+      "step": 7680
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019207499999999997,
+      "loss": 1.2515,
+      "step": 7690
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019232499999999997,
+      "loss": 1.2698,
+      "step": 7700
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019255,
+      "loss": 1.2357,
+      "step": 7710
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001928,
+      "loss": 1.2434,
+      "step": 7720
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019304999999999998,
+      "loss": 1.2472,
+      "step": 7730
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019329999999999998,
+      "loss": 1.2485,
+      "step": 7740
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019355,
+      "loss": 1.2525,
+      "step": 7750
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001938,
+      "loss": 1.2505,
+      "step": 7760
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019405,
+      "loss": 1.2413,
+      "step": 7770
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019429999999999998,
+      "loss": 1.2561,
+      "step": 7780
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019454999999999999,
+      "loss": 1.2504,
+      "step": 7790
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001948,
+      "loss": 1.2537,
+      "step": 7800
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019504999999999997,
+      "loss": 1.2451,
+      "step": 7810
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019529999999999998,
+      "loss": 1.2397,
+      "step": 7820
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019554999999999998,
+      "loss": 1.2262,
+      "step": 7830
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019579999999999996,
+      "loss": 1.2322,
+      "step": 7840
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019604999999999997,
+      "loss": 1.2229,
+      "step": 7850
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019629999999999997,
+      "loss": 1.2553,
+      "step": 7860
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019654999999999998,
+      "loss": 1.2222,
+      "step": 7870
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019679999999999999,
+      "loss": 1.2358,
+      "step": 7880
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019705,
+      "loss": 1.2192,
+      "step": 7890
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019729999999999997,
+      "loss": 1.2226,
+      "step": 7900
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019754999999999998,
+      "loss": 1.2095,
+      "step": 7910
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019779999999999998,
+      "loss": 1.2142,
+      "step": 7920
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019805,
+      "loss": 1.216,
+      "step": 7930
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001983,
+      "loss": 1.2335,
+      "step": 7940
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019855,
+      "loss": 1.2481,
+      "step": 7950
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019879999999999998,
+      "loss": 1.2291,
+      "step": 7960
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019905,
+      "loss": 1.2168,
+      "step": 7970
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001993,
+      "loss": 1.2517,
+      "step": 7980
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019955,
+      "loss": 1.2335,
+      "step": 7990
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001998,
+      "loss": 1.2429,
+      "step": 8000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00020004999999999996,
+      "loss": 1.2454,
+      "step": 8010
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00020029999999999996,
+      "loss": 1.2418,
+      "step": 8020
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00020054999999999997,
+      "loss": 1.2321,
+      "step": 8030
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00020079999999999997,
+      "loss": 1.2369,
+      "step": 8040
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00020104999999999998,
+      "loss": 1.2323,
+      "step": 8050
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002013,
+      "loss": 1.2338,
+      "step": 8060
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00020154999999999997,
+      "loss": 1.2371,
+      "step": 8070
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00020179999999999997,
+      "loss": 1.2068,
+      "step": 8080
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00020204999999999998,
+      "loss": 1.2246,
+      "step": 8090
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00020229999999999998,
+      "loss": 1.2138,
+      "step": 8100
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00020255,
+      "loss": 1.2173,
+      "step": 8110
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002028,
+      "loss": 1.2031,
+      "step": 8120
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00020304999999999998,
+      "loss": 1.222,
+      "step": 8130
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00020329999999999998,
+      "loss": 1.212,
+      "step": 8140
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00020355,
+      "loss": 1.2242,
+      "step": 8150
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002038,
+      "loss": 1.2087,
+      "step": 8160
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00020405,
+      "loss": 1.216,
+      "step": 8170
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002043,
+      "loss": 1.2334,
+      "step": 8180
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00020454999999999996,
+      "loss": 1.2154,
+      "step": 8190
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00020479999999999996,
+      "loss": 1.2098,
+      "step": 8200
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00020504999999999997,
+      "loss": 1.1788,
+      "step": 8210
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00020529999999999998,
+      "loss": 1.1994,
+      "step": 8220
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00020554999999999998,
+      "loss": 1.2037,
+      "step": 8230
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002058,
+      "loss": 1.2082,
+      "step": 8240
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00020604999999999997,
+      "loss": 1.2125,
+      "step": 8250
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00020629999999999997,
+      "loss": 1.1954,
+      "step": 8260
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00020654999999999998,
+      "loss": 1.2014,
+      "step": 8270
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00020679999999999999,
+      "loss": 1.2146,
+      "step": 8280
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00020705,
+      "loss": 1.2013,
+      "step": 8290
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00020729999999999997,
+      "loss": 1.2174,
+      "step": 8300
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00020754999999999998,
+      "loss": 1.2043,
+      "step": 8310
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00020779999999999998,
+      "loss": 1.1967,
+      "step": 8320
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00020805,
+      "loss": 1.1849,
+      "step": 8330
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002083,
+      "loss": 1.1729,
+      "step": 8340
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00020855,
+      "loss": 1.1918,
+      "step": 8350
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00020879999999999998,
+      "loss": 1.1862,
+      "step": 8360
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00020904999999999999,
+      "loss": 1.1838,
+      "step": 8370
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002093,
+      "loss": 1.1633,
+      "step": 8380
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00020955,
+      "loss": 1.175,
+      "step": 8390
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002098,
+      "loss": 1.1767,
+      "step": 8400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00021004999999999998,
+      "loss": 1.184,
+      "step": 8410
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00021029999999999996,
+      "loss": 1.1896,
+      "step": 8420
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00021054999999999997,
+      "loss": 1.2001,
+      "step": 8430
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00021079999999999997,
+      "loss": 1.1948,
+      "step": 8440
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00021104999999999998,
+      "loss": 1.1864,
+      "step": 8450
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00021129999999999999,
+      "loss": 1.1713,
+      "step": 8460
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00021155,
+      "loss": 1.1674,
+      "step": 8470
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00021179999999999997,
+      "loss": 1.1793,
+      "step": 8480
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00021204999999999998,
+      "loss": 1.1904,
+      "step": 8490
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00021229999999999998,
+      "loss": 1.1735,
+      "step": 8500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00021255,
+      "loss": 1.1594,
+      "step": 8510
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002128,
+      "loss": 1.1885,
+      "step": 8520
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00021304999999999997,
+      "loss": 1.1641,
+      "step": 8530
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00021329999999999998,
+      "loss": 1.1783,
+      "step": 8540
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00021355,
+      "loss": 1.1782,
+      "step": 8550
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002138,
+      "loss": 1.1854,
+      "step": 8560
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00021405,
+      "loss": 1.1445,
+      "step": 8570
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002143,
+      "loss": 1.1713,
+      "step": 8580
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00021454999999999996,
+      "loss": 1.1643,
+      "step": 8590
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00021479999999999996,
+      "loss": 1.1834,
+      "step": 8600
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00021504999999999997,
+      "loss": 1.1736,
+      "step": 8610
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00021529999999999997,
+      "loss": 1.1749,
+      "step": 8620
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00021554999999999998,
+      "loss": 1.1803,
+      "step": 8630
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002158,
+      "loss": 1.1583,
+      "step": 8640
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00021604999999999997,
+      "loss": 1.1511,
+      "step": 8650
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00021629999999999997,
+      "loss": 1.1896,
+      "step": 8660
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00021654999999999998,
+      "loss": 1.1664,
+      "step": 8670
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00021679999999999998,
+      "loss": 1.1734,
+      "step": 8680
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00021705,
+      "loss": 1.1613,
+      "step": 8690
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002173,
+      "loss": 1.1422,
+      "step": 8700
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00021752499999999996,
+      "loss": 1.1381,
+      "step": 8710
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00021777499999999997,
+      "loss": 1.1172,
+      "step": 8720
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00021802499999999997,
+      "loss": 1.131,
+      "step": 8730
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00021827499999999998,
+      "loss": 1.1344,
+      "step": 8740
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00021852499999999999,
+      "loss": 1.1266,
+      "step": 8750
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00021877499999999996,
+      "loss": 1.1072,
+      "step": 8760
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00021902499999999997,
+      "loss": 1.1199,
+      "step": 8770
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00021927499999999998,
+      "loss": 1.1395,
+      "step": 8780
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00021952499999999998,
+      "loss": 1.1471,
+      "step": 8790
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000219775,
+      "loss": 1.1389,
+      "step": 8800
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000220025,
+      "loss": 1.1376,
+      "step": 8810
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00022027499999999997,
+      "loss": 1.1415,
+      "step": 8820
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00022052499999999998,
+      "loss": 1.158,
+      "step": 8830
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00022077499999999999,
+      "loss": 1.1549,
+      "step": 8840
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000221025,
+      "loss": 1.1279,
+      "step": 8850
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000221275,
+      "loss": 1.1354,
+      "step": 8860
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000221525,
+      "loss": 1.1481,
+      "step": 8870
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00022177499999999996,
+      "loss": 1.1304,
+      "step": 8880
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00022202499999999996,
+      "loss": 1.1206,
+      "step": 8890
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00022227499999999997,
+      "loss": 1.1287,
+      "step": 8900
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00022252499999999997,
+      "loss": 1.1263,
+      "step": 8910
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00022277499999999998,
+      "loss": 1.1377,
+      "step": 8920
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00022302499999999996,
+      "loss": 1.1323,
+      "step": 8930
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00022327499999999997,
+      "loss": 1.1399,
+      "step": 8940
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00022352499999999997,
+      "loss": 1.1158,
+      "step": 8950
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00022377499999999998,
+      "loss": 1.119,
+      "step": 8960
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00022402499999999998,
+      "loss": 1.1351,
+      "step": 8970
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000224275,
+      "loss": 1.1323,
+      "step": 8980
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00022452499999999997,
+      "loss": 1.128,
+      "step": 8990
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00022477499999999997,
+      "loss": 1.1033,
+      "step": 9000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00022502499999999998,
+      "loss": 1.1259,
+      "step": 9010
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000225275,
+      "loss": 1.1012,
+      "step": 9020
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000225525,
+      "loss": 1.1288,
+      "step": 9030
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000225775,
+      "loss": 1.1358,
+      "step": 9040
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00022602499999999998,
+      "loss": 1.1197,
+      "step": 9050
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00022627499999999998,
+      "loss": 1.1229,
+      "step": 9060
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000226525,
+      "loss": 1.1221,
+      "step": 9070
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000226775,
+      "loss": 1.1059,
+      "step": 9080
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000227025,
+      "loss": 1.1354,
+      "step": 9090
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000227275,
+      "loss": 1.1306,
+      "step": 9100
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00022752499999999996,
+      "loss": 1.1264,
+      "step": 9110
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00022777499999999997,
+      "loss": 1.1259,
+      "step": 9120
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00022802499999999997,
+      "loss": 1.1077,
+      "step": 9130
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00022827499999999998,
+      "loss": 1.1097,
+      "step": 9140
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00022852499999999998,
+      "loss": 1.1277,
+      "step": 9150
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00022877499999999996,
+      "loss": 1.1337,
+      "step": 9160
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00022902499999999997,
+      "loss": 1.1278,
+      "step": 9170
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00022927499999999998,
+      "loss": 1.1213,
+      "step": 9180
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00022952499999999998,
+      "loss": 1.1104,
+      "step": 9190
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000229775,
+      "loss": 1.0953,
+      "step": 9200
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000230025,
+      "loss": 1.1256,
+      "step": 9210
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00023027499999999997,
+      "loss": 1.1117,
+      "step": 9220
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00023052499999999998,
+      "loss": 1.1406,
+      "step": 9230
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00023077499999999999,
+      "loss": 1.1138,
+      "step": 9240
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000231025,
+      "loss": 1.1336,
+      "step": 9250
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000231275,
+      "loss": 1.1208,
+      "step": 9260
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000231525,
+      "loss": 1.1263,
+      "step": 9270
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00023177499999999996,
+      "loss": 1.1306,
+      "step": 9280
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00023202499999999996,
+      "loss": 1.1261,
+      "step": 9290
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00023227499999999997,
+      "loss": 1.1191,
+      "step": 9300
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00023252499999999997,
+      "loss": 1.1023,
+      "step": 9310
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00023277499999999998,
+      "loss": 1.0853,
+      "step": 9320
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00023302499999999999,
+      "loss": 1.1104,
+      "step": 9330
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00023327499999999996,
+      "loss": 1.1077,
+      "step": 9340
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00023352499999999997,
+      "loss": 1.1262,
+      "step": 9350
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00023377499999999998,
+      "loss": 1.125,
+      "step": 9360
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00023402499999999998,
+      "loss": 1.1072,
+      "step": 9370
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000234275,
+      "loss": 1.11,
+      "step": 9380
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00023452499999999997,
+      "loss": 1.0846,
+      "step": 9390
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00023477499999999997,
+      "loss": 1.0895,
+      "step": 9400
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00023502499999999998,
+      "loss": 1.0858,
+      "step": 9410
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00023527499999999999,
+      "loss": 1.0886,
+      "step": 9420
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000235525,
+      "loss": 1.0858,
+      "step": 9430
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000235775,
+      "loss": 1.0684,
+      "step": 9440
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00023602499999999998,
+      "loss": 1.0701,
+      "step": 9450
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00023627499999999998,
+      "loss": 1.0703,
+      "step": 9460
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00023652499999999996,
+      "loss": 1.0928,
+      "step": 9470
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00023677499999999997,
+      "loss": 1.0823,
+      "step": 9480
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00023702499999999997,
+      "loss": 1.0958,
+      "step": 9490
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00023727499999999998,
+      "loss": 1.0629,
+      "step": 9500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00023752499999999996,
+      "loss": 1.0654,
+      "step": 9510
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00023777499999999997,
+      "loss": 1.063,
+      "step": 9520
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00023802499999999997,
+      "loss": 1.0727,
+      "step": 9530
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00023827499999999998,
+      "loss": 1.0725,
+      "step": 9540
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00023852499999999998,
+      "loss": 1.0759,
+      "step": 9550
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000238775,
+      "loss": 1.0493,
+      "step": 9560
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00023902499999999997,
+      "loss": 1.0511,
+      "step": 9570
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00023927499999999997,
+      "loss": 1.0333,
+      "step": 9580
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00023952499999999998,
+      "loss": 1.0718,
+      "step": 9590
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000239775,
+      "loss": 1.0631,
+      "step": 9600
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000240025,
+      "loss": 1.0662,
+      "step": 9610
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000240275,
+      "loss": 1.0533,
+      "step": 9620
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00024052499999999998,
+      "loss": 1.0558,
+      "step": 9630
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00024077499999999998,
+      "loss": 1.0694,
+      "step": 9640
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000241025,
+      "loss": 1.1064,
+      "step": 9650
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000241275,
+      "loss": 1.0828,
+      "step": 9660
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000241525,
+      "loss": 1.0639,
+      "step": 9670
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00024177499999999995,
+      "loss": 1.0705,
+      "step": 9680
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00024202499999999996,
+      "loss": 1.0866,
+      "step": 9690
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00024227499999999997,
+      "loss": 1.0813,
+      "step": 9700
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00024252499999999997,
+      "loss": 1.081,
+      "step": 9710
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00024275,
+      "loss": 1.103,
+      "step": 9720
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000243,
+      "loss": 1.0906,
+      "step": 9730
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00024324999999999998,
+      "loss": 1.0728,
+      "step": 9740
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00024349999999999998,
+      "loss": 1.0843,
+      "step": 9750
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00024375,
+      "loss": 1.0672,
+      "step": 9760
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000244,
+      "loss": 1.0846,
+      "step": 9770
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00024425,
+      "loss": 1.0973,
+      "step": 9780
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002445,
+      "loss": 1.0933,
+      "step": 9790
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00024474999999999996,
+      "loss": 1.0767,
+      "step": 9800
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000245,
+      "loss": 1.0875,
+      "step": 9810
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00024524999999999997,
+      "loss": 1.0775,
+      "step": 9820
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002455,
+      "loss": 1.0833,
+      "step": 9830
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00024575,
+      "loss": 1.0829,
+      "step": 9840
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00024599999999999996,
+      "loss": 1.0804,
+      "step": 9850
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00024625,
+      "loss": 1.0755,
+      "step": 9860
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002465,
+      "loss": 1.0638,
+      "step": 9870
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00024675,
+      "loss": 1.0426,
+      "step": 9880
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000247,
+      "loss": 1.0552,
+      "step": 9890
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00024724999999999997,
+      "loss": 1.0444,
+      "step": 9900
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00024749999999999994,
+      "loss": 1.07,
+      "step": 9910
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00024775,
+      "loss": 1.078,
+      "step": 9920
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00024799999999999996,
+      "loss": 1.0591,
+      "step": 9930
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00024825,
+      "loss": 1.0609,
+      "step": 9940
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00024849999999999997,
+      "loss": 1.0778,
+      "step": 9950
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00024875,
+      "loss": 1.0898,
+      "step": 9960
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000249,
+      "loss": 1.0609,
+      "step": 9970
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00024924999999999996,
+      "loss": 1.0747,
+      "step": 9980
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002495,
+      "loss": 1.0724,
+      "step": 9990
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00024974999999999997,
+      "loss": 1.0477,
+      "step": 10000
+    },
+    {
+      "epoch": 0.1,
+      "eval_accuracy": 0.7842036631094624,
+      "eval_loss": 1.115234375,
+      "eval_runtime": 96.9888,
+      "eval_samples_per_second": 824.838,
+      "eval_steps_per_second": 1.619,
+      "step": 10000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00025,
+      "loss": 1.0264,
+      "step": 10010
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00025025,
+      "loss": 1.0462,
+      "step": 10020
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00025049999999999996,
+      "loss": 1.0428,
+      "step": 10030
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00025075,
+      "loss": 1.0627,
+      "step": 10040
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000251,
+      "loss": 1.049,
+      "step": 10050
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00025125,
+      "loss": 1.0465,
+      "step": 10060
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002515,
+      "loss": 1.0318,
+      "step": 10070
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00025174999999999997,
+      "loss": 1.0378,
+      "step": 10080
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00025199999999999995,
+      "loss": 1.0406,
+      "step": 10090
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00025225,
+      "loss": 1.0562,
+      "step": 10100
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00025249999999999996,
+      "loss": 1.0432,
+      "step": 10110
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00025275,
+      "loss": 1.0308,
+      "step": 10120
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00025299999999999997,
+      "loss": 1.0178,
+      "step": 10130
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00025324999999999995,
+      "loss": 1.0565,
+      "step": 10140
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002535,
+      "loss": 1.0464,
+      "step": 10150
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00025374999999999996,
+      "loss": 1.0578,
+      "step": 10160
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000254,
+      "loss": 1.0467,
+      "step": 10170
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00025425,
+      "loss": 1.0362,
+      "step": 10180
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002545,
+      "loss": 1.0415,
+      "step": 10190
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00025475,
+      "loss": 1.0466,
+      "step": 10200
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00025499999999999996,
+      "loss": 1.0281,
+      "step": 10210
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00025525,
+      "loss": 1.0507,
+      "step": 10220
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002555,
+      "loss": 1.0262,
+      "step": 10230
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00025575,
+      "loss": 1.0221,
+      "step": 10240
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000256,
+      "loss": 1.0043,
+      "step": 10250
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00025624999999999997,
+      "loss": 1.0257,
+      "step": 10260
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00025649999999999995,
+      "loss": 1.0277,
+      "step": 10270
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00025675,
+      "loss": 1.0356,
+      "step": 10280
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00025699999999999996,
+      "loss": 1.0249,
+      "step": 10290
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00025725,
+      "loss": 1.0259,
+      "step": 10300
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00025749999999999997,
+      "loss": 0.9973,
+      "step": 10310
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00025774999999999995,
+      "loss": 1.0064,
+      "step": 10320
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000258,
+      "loss": 1.0049,
+      "step": 10330
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00025824999999999996,
+      "loss": 1.0566,
+      "step": 10340
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002585,
+      "loss": 1.0659,
+      "step": 10350
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00025875,
+      "loss": 1.0536,
+      "step": 10360
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00025899999999999995,
+      "loss": 1.0511,
+      "step": 10370
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00025925,
+      "loss": 1.0522,
+      "step": 10380
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00025949999999999997,
+      "loss": 1.0431,
+      "step": 10390
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00025975,
+      "loss": 1.0517,
+      "step": 10400
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00026,
+      "loss": 1.0646,
+      "step": 10410
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00026025,
+      "loss": 1.0405,
+      "step": 10420
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002605,
+      "loss": 1.0417,
+      "step": 10430
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00026074999999999997,
+      "loss": 1.0508,
+      "step": 10440
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000261,
+      "loss": 1.0434,
+      "step": 10450
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00026125,
+      "loss": 1.0584,
+      "step": 10460
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002615,
+      "loss": 1.069,
+      "step": 10470
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00026175,
+      "loss": 1.0468,
+      "step": 10480
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00026199999999999997,
+      "loss": 1.0446,
+      "step": 10490
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00026224999999999995,
+      "loss": 1.0502,
+      "step": 10500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002625,
+      "loss": 1.0355,
+      "step": 10510
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00026274999999999996,
+      "loss": 1.0702,
+      "step": 10520
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000263,
+      "loss": 1.0659,
+      "step": 10530
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00026325,
+      "loss": 1.0434,
+      "step": 10540
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00026349999999999995,
+      "loss": 1.0538,
+      "step": 10550
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00026375,
+      "loss": 1.0319,
+      "step": 10560
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00026399999999999997,
+      "loss": 1.0367,
+      "step": 10570
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00026425,
+      "loss": 1.0461,
+      "step": 10580
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002645,
+      "loss": 1.0409,
+      "step": 10590
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00026474999999999996,
+      "loss": 1.0372,
+      "step": 10600
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000265,
+      "loss": 1.048,
+      "step": 10610
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00026524999999999997,
+      "loss": 1.0323,
+      "step": 10620
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002655,
+      "loss": 1.0342,
+      "step": 10630
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00026575,
+      "loss": 1.0125,
+      "step": 10640
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000266,
+      "loss": 1.0315,
+      "step": 10650
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00026624999999999994,
+      "loss": 1.0279,
+      "step": 10660
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00026649999999999997,
+      "loss": 1.0302,
+      "step": 10670
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00026674999999999995,
+      "loss": 1.0211,
+      "step": 10680
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000267,
+      "loss": 1.0223,
+      "step": 10690
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00026724999999999996,
+      "loss": 1.0169,
+      "step": 10700
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002675,
+      "loss": 1.036,
+      "step": 10710
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00026772499999999996,
+      "loss": 1.0318,
+      "step": 10720
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000267975,
+      "loss": 1.0206,
+      "step": 10730
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000268225,
+      "loss": 0.999,
+      "step": 10740
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000268475,
+      "loss": 0.9801,
+      "step": 10750
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000268725,
+      "loss": 0.9834,
+      "step": 10760
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00026897499999999996,
+      "loss": 1.0261,
+      "step": 10770
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00026922499999999994,
+      "loss": 1.0063,
+      "step": 10780
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000269475,
+      "loss": 1.0153,
+      "step": 10790
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00026972499999999996,
+      "loss": 0.9967,
+      "step": 10800
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000269975,
+      "loss": 0.9976,
+      "step": 10810
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00027022499999999997,
+      "loss": 1.0015,
+      "step": 10820
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00027047499999999995,
+      "loss": 1.0215,
+      "step": 10830
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000270725,
+      "loss": 1.0095,
+      "step": 10840
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00027097499999999996,
+      "loss": 1.0341,
+      "step": 10850
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000271225,
+      "loss": 1.0016,
+      "step": 10860
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00027147499999999997,
+      "loss": 1.0226,
+      "step": 10870
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000271725,
+      "loss": 0.9989,
+      "step": 10880
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000271975,
+      "loss": 1.006,
+      "step": 10890
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00027222499999999996,
+      "loss": 0.9861,
+      "step": 10900
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000272475,
+      "loss": 1.0227,
+      "step": 10910
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000272725,
+      "loss": 0.9842,
+      "step": 10920
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000272975,
+      "loss": 0.9882,
+      "step": 10930
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000273225,
+      "loss": 0.9849,
+      "step": 10940
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00027347499999999997,
+      "loss": 1.003,
+      "step": 10950
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000273725,
+      "loss": 1.0014,
+      "step": 10960
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000273975,
+      "loss": 1.0029,
+      "step": 10970
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00027422499999999996,
+      "loss": 1.0124,
+      "step": 10980
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000274475,
+      "loss": 1.0175,
+      "step": 10990
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00027472499999999997,
+      "loss": 0.9947,
+      "step": 11000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00027497499999999995,
+      "loss": 0.9935,
+      "step": 11010
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000275225,
+      "loss": 1.0061,
+      "step": 11020
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00027547499999999996,
+      "loss": 1.0129,
+      "step": 11030
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000275725,
+      "loss": 1.0046,
+      "step": 11040
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00027597499999999997,
+      "loss": 0.9996,
+      "step": 11050
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00027622499999999995,
+      "loss": 0.9892,
+      "step": 11060
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000276475,
+      "loss": 0.9932,
+      "step": 11070
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00027672499999999996,
+      "loss": 1.021,
+      "step": 11080
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000276975,
+      "loss": 1.0211,
+      "step": 11090
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000277225,
+      "loss": 1.0129,
+      "step": 11100
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000277475,
+      "loss": 0.9703,
+      "step": 11110
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000277725,
+      "loss": 0.9861,
+      "step": 11120
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00027797499999999997,
+      "loss": 0.9778,
+      "step": 11130
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000278225,
+      "loss": 0.9914,
+      "step": 11140
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000278475,
+      "loss": 0.9898,
+      "step": 11150
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000278725,
+      "loss": 1.0094,
+      "step": 11160
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00027897499999999994,
+      "loss": 0.9753,
+      "step": 11170
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00027922499999999997,
+      "loss": 0.9997,
+      "step": 11180
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00027947499999999995,
+      "loss": 0.9668,
+      "step": 11190
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000279725,
+      "loss": 1.0121,
+      "step": 11200
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00027997499999999996,
+      "loss": 1.0003,
+      "step": 11210
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000280225,
+      "loss": 1.0057,
+      "step": 11220
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000280475,
+      "loss": 1.0012,
+      "step": 11230
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00028072499999999995,
+      "loss": 0.9905,
+      "step": 11240
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000280975,
+      "loss": 0.9816,
+      "step": 11250
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00028122499999999996,
+      "loss": 1.0025,
+      "step": 11260
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000281475,
+      "loss": 0.9853,
+      "step": 11270
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000281725,
+      "loss": 0.9964,
+      "step": 11280
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00028197499999999996,
+      "loss": 0.9941,
+      "step": 11290
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000282225,
+      "loss": 0.9888,
+      "step": 11300
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00028247499999999997,
+      "loss": 0.9934,
+      "step": 11310
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000282725,
+      "loss": 1.0022,
+      "step": 11320
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000282975,
+      "loss": 0.9831,
+      "step": 11330
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000283225,
+      "loss": 1.0045,
+      "step": 11340
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00028347499999999994,
+      "loss": 0.9923,
+      "step": 11350
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00028372499999999997,
+      "loss": 1.0049,
+      "step": 11360
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00028397499999999995,
+      "loss": 0.9778,
+      "step": 11370
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000284225,
+      "loss": 0.9896,
+      "step": 11380
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00028447499999999996,
+      "loss": 0.9914,
+      "step": 11390
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00028472499999999994,
+      "loss": 1.005,
+      "step": 11400
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000284975,
+      "loss": 1.0014,
+      "step": 11410
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00028522499999999995,
+      "loss": 1.0103,
+      "step": 11420
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000285475,
+      "loss": 0.9762,
+      "step": 11430
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00028572499999999997,
+      "loss": 1.002,
+      "step": 11440
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000285975,
+      "loss": 0.9845,
+      "step": 11450
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000286225,
+      "loss": 0.9944,
+      "step": 11460
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00028647499999999996,
+      "loss": 0.9813,
+      "step": 11470
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000286725,
+      "loss": 0.9881,
+      "step": 11480
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00028697499999999997,
+      "loss": 0.9478,
+      "step": 11490
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000287225,
+      "loss": 0.976,
+      "step": 11500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000287475,
+      "loss": 0.9837,
+      "step": 11510
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00028772499999999996,
+      "loss": 1.0044,
+      "step": 11520
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000287975,
+      "loss": 0.9704,
+      "step": 11530
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00028822499999999997,
+      "loss": 0.9819,
+      "step": 11540
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000288475,
+      "loss": 0.9807,
+      "step": 11550
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000288725,
+      "loss": 0.9934,
+      "step": 11560
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00028897499999999996,
+      "loss": 0.9772,
+      "step": 11570
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00028922499999999994,
+      "loss": 0.9869,
+      "step": 11580
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000289475,
+      "loss": 1.0183,
+      "step": 11590
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00028972499999999995,
+      "loss": 1.0031,
+      "step": 11600
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000289975,
+      "loss": 1.005,
+      "step": 11610
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029022499999999997,
+      "loss": 0.9948,
+      "step": 11620
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000290475,
+      "loss": 1.0125,
+      "step": 11630
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000290725,
+      "loss": 1.0066,
+      "step": 11640
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029097499999999996,
+      "loss": 1.0183,
+      "step": 11650
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000291225,
+      "loss": 1.0012,
+      "step": 11660
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029147499999999997,
+      "loss": 0.9977,
+      "step": 11670
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000291725,
+      "loss": 0.991,
+      "step": 11680
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000291975,
+      "loss": 0.996,
+      "step": 11690
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029222499999999996,
+      "loss": 1.0012,
+      "step": 11700
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000292475,
+      "loss": 1.0169,
+      "step": 11710
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029269999999999996,
+      "loss": 1.0089,
+      "step": 11720
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029295,
+      "loss": 0.9878,
+      "step": 11730
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029319999999999997,
+      "loss": 0.9738,
+      "step": 11740
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029344999999999995,
+      "loss": 1.0032,
+      "step": 11750
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002937,
+      "loss": 0.994,
+      "step": 11760
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029394999999999996,
+      "loss": 1.005,
+      "step": 11770
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002942,
+      "loss": 0.9683,
+      "step": 11780
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029445,
+      "loss": 0.983,
+      "step": 11790
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029469999999999995,
+      "loss": 0.9743,
+      "step": 11800
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029495,
+      "loss": 0.9866,
+      "step": 11810
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029519999999999997,
+      "loss": 0.9976,
+      "step": 11820
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029545,
+      "loss": 1.0009,
+      "step": 11830
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002957,
+      "loss": 1.0055,
+      "step": 11840
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029595,
+      "loss": 0.9906,
+      "step": 11850
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029619999999999994,
+      "loss": 0.9845,
+      "step": 11860
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029644999999999997,
+      "loss": 0.9647,
+      "step": 11870
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029669999999999995,
+      "loss": 1.0115,
+      "step": 11880
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029695,
+      "loss": 1.0021,
+      "step": 11890
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029719999999999996,
+      "loss": 1.0008,
+      "step": 11900
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029745,
+      "loss": 0.9766,
+      "step": 11910
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029769999999999997,
+      "loss": 0.966,
+      "step": 11920
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029794999999999995,
+      "loss": 0.991,
+      "step": 11930
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002982,
+      "loss": 0.9696,
+      "step": 11940
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029844999999999996,
+      "loss": 0.9896,
+      "step": 11950
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002987,
+      "loss": 0.9878,
+      "step": 11960
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029895,
+      "loss": 0.9877,
+      "step": 11970
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029919999999999995,
+      "loss": 0.9899,
+      "step": 11980
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029945,
+      "loss": 0.9921,
+      "step": 11990
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029969999999999997,
+      "loss": 0.9874,
+      "step": 12000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029995,
+      "loss": 1.0079,
+      "step": 12010
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003002,
+      "loss": 0.9987,
+      "step": 12020
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00030045,
+      "loss": 0.9862,
+      "step": 12030
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00030069999999999994,
+      "loss": 0.9898,
+      "step": 12040
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00030095,
+      "loss": 0.9914,
+      "step": 12050
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00030119999999999995,
+      "loss": 0.9903,
+      "step": 12060
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00030144999999999993,
+      "loss": 0.9978,
+      "step": 12070
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00030169999999999996,
+      "loss": 0.9808,
+      "step": 12080
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00030194999999999994,
+      "loss": 1.0049,
+      "step": 12090
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003022,
+      "loss": 0.9987,
+      "step": 12100
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00030244999999999995,
+      "loss": 0.9894,
+      "step": 12110
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00030269999999999993,
+      "loss": 0.9732,
+      "step": 12120
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00030294999999999996,
+      "loss": 0.9937,
+      "step": 12130
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00030319999999999994,
+      "loss": 0.994,
+      "step": 12140
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00030345,
+      "loss": 1.0007,
+      "step": 12150
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00030369999999999996,
+      "loss": 0.989,
+      "step": 12160
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00030395,
+      "loss": 0.9978,
+      "step": 12170
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00030419999999999997,
+      "loss": 0.993,
+      "step": 12180
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00030444999999999995,
+      "loss": 0.9995,
+      "step": 12190
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003047,
+      "loss": 0.9951,
+      "step": 12200
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00030494999999999996,
+      "loss": 0.9982,
+      "step": 12210
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003052,
+      "loss": 1.0111,
+      "step": 12220
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00030544999999999997,
+      "loss": 0.9741,
+      "step": 12230
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00030569999999999995,
+      "loss": 0.9882,
+      "step": 12240
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00030595,
+      "loss": 0.9801,
+      "step": 12250
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00030619999999999996,
+      "loss": 0.9705,
+      "step": 12260
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00030645,
+      "loss": 0.9911,
+      "step": 12270
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003067,
+      "loss": 0.9778,
+      "step": 12280
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00030694999999999995,
+      "loss": 0.9778,
+      "step": 12290
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003072,
+      "loss": 0.9886,
+      "step": 12300
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00030744999999999997,
+      "loss": 0.9936,
+      "step": 12310
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003077,
+      "loss": 0.9837,
+      "step": 12320
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00030795,
+      "loss": 0.9883,
+      "step": 12330
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003082,
+      "loss": 0.9718,
+      "step": 12340
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00030845,
+      "loss": 0.9745,
+      "step": 12350
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00030869999999999997,
+      "loss": 0.9696,
+      "step": 12360
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00030895,
+      "loss": 0.9832,
+      "step": 12370
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003092,
+      "loss": 0.9874,
+      "step": 12380
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00030945,
+      "loss": 0.9961,
+      "step": 12390
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003097,
+      "loss": 0.9864,
+      "step": 12400
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003099499999999999,
+      "loss": 0.967,
+      "step": 12410
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003102,
+      "loss": 0.9945,
+      "step": 12420
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00031044999999999993,
+      "loss": 0.9778,
+      "step": 12430
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003107,
+      "loss": 0.9881,
+      "step": 12440
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00031094999999999994,
+      "loss": 0.9843,
+      "step": 12450
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00031120000000000003,
+      "loss": 0.9742,
+      "step": 12460
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00031144999999999995,
+      "loss": 0.9726,
+      "step": 12470
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00031169999999999993,
+      "loss": 0.9565,
+      "step": 12480
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00031194999999999997,
+      "loss": 0.9647,
+      "step": 12490
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00031219999999999995,
+      "loss": 0.9817,
+      "step": 12500
+    },
+    {
+      "epoch": 0.12,
+      "eval_accuracy": 0.7950684202995898,
+      "eval_loss": 1.0478515625,
+      "eval_runtime": 97.1269,
+      "eval_samples_per_second": 823.665,
+      "eval_steps_per_second": 1.616,
+      "step": 12500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00031245,
+      "loss": 0.9518,
+      "step": 12510
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00031269999999999996,
+      "loss": 0.9495,
+      "step": 12520
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00031294999999999994,
+      "loss": 0.9507,
+      "step": 12530
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00031319999999999997,
+      "loss": 0.9361,
+      "step": 12540
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00031344999999999995,
+      "loss": 0.944,
+      "step": 12550
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003137,
+      "loss": 0.9476,
+      "step": 12560
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00031394999999999996,
+      "loss": 0.9425,
+      "step": 12570
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00031419999999999994,
+      "loss": 0.9681,
+      "step": 12580
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00031444999999999997,
+      "loss": 0.9618,
+      "step": 12590
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00031469999999999995,
+      "loss": 0.9591,
+      "step": 12600
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00031495,
+      "loss": 0.9782,
+      "step": 12610
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00031519999999999996,
+      "loss": 0.9584,
+      "step": 12620
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00031545,
+      "loss": 0.9911,
+      "step": 12630
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003157,
+      "loss": 0.9893,
+      "step": 12640
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00031594999999999995,
+      "loss": 0.9754,
+      "step": 12650
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003162,
+      "loss": 0.9664,
+      "step": 12660
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00031644999999999997,
+      "loss": 0.9677,
+      "step": 12670
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003167,
+      "loss": 0.9835,
+      "step": 12680
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00031695,
+      "loss": 0.9739,
+      "step": 12690
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00031719999999999996,
+      "loss": 0.9776,
+      "step": 12700
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00031745,
+      "loss": 0.9723,
+      "step": 12710
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000317675,
+      "loss": 0.9647,
+      "step": 12720
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000317925,
+      "loss": 0.9681,
+      "step": 12730
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000318175,
+      "loss": 0.9747,
+      "step": 12740
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000318425,
+      "loss": 0.9683,
+      "step": 12750
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003186749999999999,
+      "loss": 0.9854,
+      "step": 12760
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00031892499999999996,
+      "loss": 0.9822,
+      "step": 12770
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00031917499999999994,
+      "loss": 0.9719,
+      "step": 12780
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00031942499999999997,
+      "loss": 0.9675,
+      "step": 12790
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00031967499999999995,
+      "loss": 0.9678,
+      "step": 12800
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00031992499999999993,
+      "loss": 0.9665,
+      "step": 12810
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00032017499999999996,
+      "loss": 0.9547,
+      "step": 12820
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00032042499999999994,
+      "loss": 0.9615,
+      "step": 12830
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000320675,
+      "loss": 0.9436,
+      "step": 12840
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00032092499999999995,
+      "loss": 0.985,
+      "step": 12850
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000321175,
+      "loss": 0.9761,
+      "step": 12860
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00032142499999999997,
+      "loss": 0.9892,
+      "step": 12870
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00032167499999999994,
+      "loss": 0.9831,
+      "step": 12880
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000321925,
+      "loss": 0.978,
+      "step": 12890
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00032217499999999996,
+      "loss": 0.9744,
+      "step": 12900
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000322425,
+      "loss": 0.9768,
+      "step": 12910
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00032267499999999997,
+      "loss": 0.9772,
+      "step": 12920
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00032292499999999995,
+      "loss": 0.9737,
+      "step": 12930
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000323175,
+      "loss": 0.9604,
+      "step": 12940
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00032342499999999996,
+      "loss": 0.9649,
+      "step": 12950
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000323675,
+      "loss": 0.9837,
+      "step": 12960
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00032392499999999997,
+      "loss": 0.9745,
+      "step": 12970
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00032417499999999995,
+      "loss": 0.9844,
+      "step": 12980
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000324425,
+      "loss": 0.9779,
+      "step": 12990
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00032467499999999996,
+      "loss": 0.9783,
+      "step": 13000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000324925,
+      "loss": 0.9821,
+      "step": 13010
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000325175,
+      "loss": 0.9589,
+      "step": 13020
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000325425,
+      "loss": 0.977,
+      "step": 13030
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000325675,
+      "loss": 0.9504,
+      "step": 13040
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00032592499999999997,
+      "loss": 0.9621,
+      "step": 13050
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000326175,
+      "loss": 0.9669,
+      "step": 13060
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000326425,
+      "loss": 0.9712,
+      "step": 13070
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000326675,
+      "loss": 0.9772,
+      "step": 13080
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000326925,
+      "loss": 0.969,
+      "step": 13090
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003271749999999999,
+      "loss": 0.9666,
+      "step": 13100
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000327425,
+      "loss": 0.9537,
+      "step": 13110
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00032767499999999993,
+      "loss": 0.9664,
+      "step": 13120
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000327925,
+      "loss": 0.9789,
+      "step": 13130
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00032817499999999994,
+      "loss": 0.9689,
+      "step": 13140
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003284249999999999,
+      "loss": 0.9689,
+      "step": 13150
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00032867499999999995,
+      "loss": 0.9678,
+      "step": 13160
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00032892499999999993,
+      "loss": 0.968,
+      "step": 13170
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00032917499999999996,
+      "loss": 0.952,
+      "step": 13180
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00032942499999999994,
+      "loss": 0.9559,
+      "step": 13190
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000329675,
+      "loss": 0.963,
+      "step": 13200
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00032992499999999996,
+      "loss": 0.9352,
+      "step": 13210
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00033017499999999993,
+      "loss": 0.9508,
+      "step": 13220
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00033042499999999997,
+      "loss": 0.9283,
+      "step": 13230
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00033067499999999995,
+      "loss": 0.9382,
+      "step": 13240
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000330925,
+      "loss": 0.944,
+      "step": 13250
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00033117499999999996,
+      "loss": 0.951,
+      "step": 13260
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00033142499999999994,
+      "loss": 0.9331,
+      "step": 13270
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00033167499999999997,
+      "loss": 0.9377,
+      "step": 13280
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00033192499999999995,
+      "loss": 0.934,
+      "step": 13290
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000332175,
+      "loss": 0.9317,
+      "step": 13300
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00033242499999999996,
+      "loss": 0.929,
+      "step": 13310
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000332675,
+      "loss": 0.9474,
+      "step": 13320
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000332925,
+      "loss": 0.9454,
+      "step": 13330
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00033317499999999995,
+      "loss": 0.9484,
+      "step": 13340
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000333425,
+      "loss": 0.9346,
+      "step": 13350
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00033367499999999997,
+      "loss": 0.9379,
+      "step": 13360
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000333925,
+      "loss": 0.93,
+      "step": 13370
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000334175,
+      "loss": 0.9487,
+      "step": 13380
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00033442499999999996,
+      "loss": 0.9495,
+      "step": 13390
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000334675,
+      "loss": 0.9382,
+      "step": 13400
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00033492499999999997,
+      "loss": 0.9315,
+      "step": 13410
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000335175,
+      "loss": 0.9374,
+      "step": 13420
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000335425,
+      "loss": 0.9448,
+      "step": 13430
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00033567499999999996,
+      "loss": 0.9493,
+      "step": 13440
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000335925,
+      "loss": 0.9409,
+      "step": 13450
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00033617499999999997,
+      "loss": 0.9346,
+      "step": 13460
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000336425,
+      "loss": 0.9601,
+      "step": 13470
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000336675,
+      "loss": 0.9639,
+      "step": 13480
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000336925,
+      "loss": 0.9514,
+      "step": 13490
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000337175,
+      "loss": 0.974,
+      "step": 13500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003374249999999999,
+      "loss": 0.9714,
+      "step": 13510
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000337675,
+      "loss": 0.9557,
+      "step": 13520
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00033792499999999993,
+      "loss": 0.9642,
+      "step": 13530
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000338175,
+      "loss": 0.957,
+      "step": 13540
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00033842499999999995,
+      "loss": 0.9521,
+      "step": 13550
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003386749999999999,
+      "loss": 0.9619,
+      "step": 13560
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00033892499999999996,
+      "loss": 0.9743,
+      "step": 13570
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00033917499999999994,
+      "loss": 0.9548,
+      "step": 13580
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00033942499999999997,
+      "loss": 0.9571,
+      "step": 13590
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00033967499999999995,
+      "loss": 0.9536,
+      "step": 13600
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000339925,
+      "loss": 0.9645,
+      "step": 13610
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00034017499999999996,
+      "loss": 0.9431,
+      "step": 13620
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00034042499999999994,
+      "loss": 0.9596,
+      "step": 13630
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000340675,
+      "loss": 0.9659,
+      "step": 13640
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00034092499999999995,
+      "loss": 0.9651,
+      "step": 13650
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000341175,
+      "loss": 0.9674,
+      "step": 13660
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00034142499999999996,
+      "loss": 0.9548,
+      "step": 13670
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00034167499999999994,
+      "loss": 0.9649,
+      "step": 13680
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000341925,
+      "loss": 0.9571,
+      "step": 13690
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00034217499999999996,
+      "loss": 0.9592,
+      "step": 13700
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000342425,
+      "loss": 0.9577,
+      "step": 13710
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00034265,
+      "loss": 0.9515,
+      "step": 13720
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003429,
+      "loss": 0.9422,
+      "step": 13730
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00034314999999999997,
+      "loss": 0.9448,
+      "step": 13740
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003434,
+      "loss": 0.9529,
+      "step": 13750
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00034365,
+      "loss": 0.9589,
+      "step": 13760
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003439,
+      "loss": 0.9651,
+      "step": 13770
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00034415,
+      "loss": 0.9682,
+      "step": 13780
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00034439999999999997,
+      "loss": 0.942,
+      "step": 13790
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00034465,
+      "loss": 0.9512,
+      "step": 13800
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003449,
+      "loss": 0.9686,
+      "step": 13810
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00034515,
+      "loss": 0.9516,
+      "step": 13820
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00034539999999999994,
+      "loss": 0.9463,
+      "step": 13830
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003456499999999999,
+      "loss": 0.943,
+      "step": 13840
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00034589999999999995,
+      "loss": 0.9457,
+      "step": 13850
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00034614999999999993,
+      "loss": 0.9424,
+      "step": 13860
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00034639999999999996,
+      "loss": 0.9498,
+      "step": 13870
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00034664999999999994,
+      "loss": 0.9644,
+      "step": 13880
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003469,
+      "loss": 0.9507,
+      "step": 13890
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00034714999999999995,
+      "loss": 0.9474,
+      "step": 13900
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00034739999999999993,
+      "loss": 0.9442,
+      "step": 13910
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00034764999999999997,
+      "loss": 0.9538,
+      "step": 13920
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00034789999999999995,
+      "loss": 0.9647,
+      "step": 13930
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00034815,
+      "loss": 0.9311,
+      "step": 13940
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00034839999999999996,
+      "loss": 0.9468,
+      "step": 13950
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00034864999999999994,
+      "loss": 0.9446,
+      "step": 13960
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00034889999999999997,
+      "loss": 0.9407,
+      "step": 13970
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00034914999999999995,
+      "loss": 0.928,
+      "step": 13980
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003494,
+      "loss": 0.9465,
+      "step": 13990
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00034964999999999996,
+      "loss": 0.9496,
+      "step": 14000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003499,
+      "loss": 0.9511,
+      "step": 14010
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00035015,
+      "loss": 0.9594,
+      "step": 14020
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00035039999999999995,
+      "loss": 0.9473,
+      "step": 14030
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00035065,
+      "loss": 0.9374,
+      "step": 14040
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00035089999999999996,
+      "loss": 0.9394,
+      "step": 14050
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00035115,
+      "loss": 0.9547,
+      "step": 14060
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003514,
+      "loss": 0.9557,
+      "step": 14070
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00035164999999999996,
+      "loss": 0.9297,
+      "step": 14080
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003519,
+      "loss": 0.9475,
+      "step": 14090
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00035214999999999997,
+      "loss": 0.9497,
+      "step": 14100
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003524,
+      "loss": 0.9467,
+      "step": 14110
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00035265,
+      "loss": 0.9632,
+      "step": 14120
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00035289999999999996,
+      "loss": 0.9579,
+      "step": 14130
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00035315,
+      "loss": 0.9547,
+      "step": 14140
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00035339999999999997,
+      "loss": 0.941,
+      "step": 14150
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00035365,
+      "loss": 0.9354,
+      "step": 14160
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003539,
+      "loss": 0.9401,
+      "step": 14170
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00035415,
+      "loss": 0.9526,
+      "step": 14180
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003544,
+      "loss": 0.9453,
+      "step": 14190
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003546499999999999,
+      "loss": 0.9356,
+      "step": 14200
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003549,
+      "loss": 0.9381,
+      "step": 14210
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00035514999999999993,
+      "loss": 0.944,
+      "step": 14220
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003554,
+      "loss": 0.9448,
+      "step": 14230
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00035564999999999994,
+      "loss": 0.9566,
+      "step": 14240
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003558999999999999,
+      "loss": 0.9356,
+      "step": 14250
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00035614999999999996,
+      "loss": 0.9481,
+      "step": 14260
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00035639999999999994,
+      "loss": 0.9392,
+      "step": 14270
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00035664999999999997,
+      "loss": 0.9335,
+      "step": 14280
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00035689999999999995,
+      "loss": 0.9419,
+      "step": 14290
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003571499999999999,
+      "loss": 0.9458,
+      "step": 14300
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00035739999999999996,
+      "loss": 0.949,
+      "step": 14310
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00035764999999999994,
+      "loss": 0.9428,
+      "step": 14320
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00035789999999999997,
+      "loss": 0.9429,
+      "step": 14330
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00035814999999999995,
+      "loss": 0.9408,
+      "step": 14340
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003584,
+      "loss": 0.94,
+      "step": 14350
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00035864999999999996,
+      "loss": 0.9442,
+      "step": 14360
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00035889999999999994,
+      "loss": 0.9475,
+      "step": 14370
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00035915,
+      "loss": 0.939,
+      "step": 14380
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00035939999999999995,
+      "loss": 0.9331,
+      "step": 14390
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00035965,
+      "loss": 0.9342,
+      "step": 14400
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00035989999999999997,
+      "loss": 0.9157,
+      "step": 14410
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00036014999999999995,
+      "loss": 0.9288,
+      "step": 14420
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003604,
+      "loss": 0.9442,
+      "step": 14430
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00036064999999999996,
+      "loss": 0.9333,
+      "step": 14440
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003609,
+      "loss": 0.9477,
+      "step": 14450
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00036114999999999997,
+      "loss": 0.9564,
+      "step": 14460
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003614,
+      "loss": 0.9514,
+      "step": 14470
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00036165,
+      "loss": 0.9478,
+      "step": 14480
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00036189999999999996,
+      "loss": 0.9499,
+      "step": 14490
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00036215,
+      "loss": 0.9363,
+      "step": 14500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00036239999999999997,
+      "loss": 0.9307,
+      "step": 14510
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00036265,
+      "loss": 0.933,
+      "step": 14520
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003629,
+      "loss": 0.9388,
+      "step": 14530
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00036314999999999996,
+      "loss": 0.9317,
+      "step": 14540
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003634,
+      "loss": 0.9299,
+      "step": 14550
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00036365,
+      "loss": 0.9437,
+      "step": 14560
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003639,
+      "loss": 0.9454,
+      "step": 14570
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00036415,
+      "loss": 0.9416,
+      "step": 14580
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003643999999999999,
+      "loss": 0.9183,
+      "step": 14590
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00036465,
+      "loss": 0.9327,
+      "step": 14600
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003648999999999999,
+      "loss": 0.922,
+      "step": 14610
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00036515,
+      "loss": 0.9334,
+      "step": 14620
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00036539999999999994,
+      "loss": 0.9428,
+      "step": 14630
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00036565,
+      "loss": 0.9437,
+      "step": 14640
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00036589999999999995,
+      "loss": 0.9216,
+      "step": 14650
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00036614999999999993,
+      "loss": 0.9294,
+      "step": 14660
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00036639999999999996,
+      "loss": 0.9188,
+      "step": 14670
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00036664999999999994,
+      "loss": 0.9329,
+      "step": 14680
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003669,
+      "loss": 0.9372,
+      "step": 14690
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00036714999999999995,
+      "loss": 0.9343,
+      "step": 14700
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00036739999999999993,
+      "loss": 0.9351,
+      "step": 14710
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00036764999999999996,
+      "loss": 0.9063,
+      "step": 14720
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000367875,
+      "loss": 0.9074,
+      "step": 14730
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00036812499999999996,
+      "loss": 0.9242,
+      "step": 14740
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000368375,
+      "loss": 0.9145,
+      "step": 14750
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000368625,
+      "loss": 0.9093,
+      "step": 14760
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00036887499999999995,
+      "loss": 0.9172,
+      "step": 14770
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000369125,
+      "loss": 0.9219,
+      "step": 14780
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00036937499999999997,
+      "loss": 0.9111,
+      "step": 14790
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000369625,
+      "loss": 0.9313,
+      "step": 14800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000369875,
+      "loss": 0.9135,
+      "step": 14810
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00037012499999999996,
+      "loss": 0.931,
+      "step": 14820
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000370375,
+      "loss": 0.9043,
+      "step": 14830
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00037062499999999997,
+      "loss": 0.9024,
+      "step": 14840
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000370875,
+      "loss": 0.9054,
+      "step": 14850
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000371125,
+      "loss": 0.9101,
+      "step": 14860
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000371375,
+      "loss": 0.9087,
+      "step": 14870
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000371625,
+      "loss": 0.9163,
+      "step": 14880
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003718749999999999,
+      "loss": 0.9187,
+      "step": 14890
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000372125,
+      "loss": 0.9218,
+      "step": 14900
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00037237499999999993,
+      "loss": 0.9167,
+      "step": 14910
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000372625,
+      "loss": 0.9154,
+      "step": 14920
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00037287499999999994,
+      "loss": 0.9129,
+      "step": 14930
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003731249999999999,
+      "loss": 0.9134,
+      "step": 14940
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00037337499999999995,
+      "loss": 0.914,
+      "step": 14950
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00037362499999999993,
+      "loss": 0.9167,
+      "step": 14960
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00037387499999999997,
+      "loss": 0.9252,
+      "step": 14970
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00037412499999999995,
+      "loss": 0.9329,
+      "step": 14980
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003743749999999999,
+      "loss": 0.949,
+      "step": 14990
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00037462499999999996,
+      "loss": 0.9463,
+      "step": 15000
+    },
+    {
+      "epoch": 0.15,
+      "eval_accuracy": 0.811620901381853,
+      "eval_loss": 0.94189453125,
+      "eval_runtime": 97.6826,
+      "eval_samples_per_second": 818.979,
+      "eval_steps_per_second": 1.607,
+      "step": 15000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00037487499999999994,
+      "loss": 0.9369,
+      "step": 15010
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00037512499999999997,
+      "loss": 0.9289,
+      "step": 15020
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00037537499999999995,
+      "loss": 0.9494,
+      "step": 15030
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000375625,
+      "loss": 0.9298,
+      "step": 15040
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00037587499999999996,
+      "loss": 0.9263,
+      "step": 15050
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00037612499999999994,
+      "loss": 0.9053,
+      "step": 15060
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000376375,
+      "loss": 0.8997,
+      "step": 15070
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00037662499999999995,
+      "loss": 0.8866,
+      "step": 15080
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000376875,
+      "loss": 0.9064,
+      "step": 15090
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00037712499999999996,
+      "loss": 0.9229,
+      "step": 15100
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00037737499999999994,
+      "loss": 0.928,
+      "step": 15110
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000377625,
+      "loss": 0.9462,
+      "step": 15120
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00037787499999999996,
+      "loss": 0.9331,
+      "step": 15130
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000378125,
+      "loss": 0.9253,
+      "step": 15140
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00037837499999999997,
+      "loss": 0.9153,
+      "step": 15150
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00037862499999999995,
+      "loss": 0.93,
+      "step": 15160
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000378875,
+      "loss": 0.9232,
+      "step": 15170
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00037912499999999996,
+      "loss": 0.9408,
+      "step": 15180
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000379375,
+      "loss": 0.9373,
+      "step": 15190
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00037962499999999997,
+      "loss": 0.9201,
+      "step": 15200
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000379875,
+      "loss": 0.9188,
+      "step": 15210
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000380125,
+      "loss": 0.9333,
+      "step": 15220
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038037499999999996,
+      "loss": 0.9171,
+      "step": 15230
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000380625,
+      "loss": 0.924,
+      "step": 15240
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000380875,
+      "loss": 0.938,
+      "step": 15250
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000381125,
+      "loss": 0.9111,
+      "step": 15260
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000381375,
+      "loss": 0.9052,
+      "step": 15270
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003816249999999999,
+      "loss": 0.9045,
+      "step": 15280
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000381875,
+      "loss": 0.9103,
+      "step": 15290
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003821249999999999,
+      "loss": 0.8965,
+      "step": 15300
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000382375,
+      "loss": 0.9055,
+      "step": 15310
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038262499999999994,
+      "loss": 0.885,
+      "step": 15320
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000382875,
+      "loss": 0.8948,
+      "step": 15330
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038312499999999995,
+      "loss": 0.8965,
+      "step": 15340
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003833749999999999,
+      "loss": 0.909,
+      "step": 15350
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038362499999999996,
+      "loss": 0.9016,
+      "step": 15360
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038387499999999994,
+      "loss": 0.8971,
+      "step": 15370
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038412499999999997,
+      "loss": 0.8991,
+      "step": 15380
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038437499999999995,
+      "loss": 0.9154,
+      "step": 15390
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038462499999999993,
+      "loss": 0.8933,
+      "step": 15400
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038487499999999996,
+      "loss": 0.9045,
+      "step": 15410
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038512499999999994,
+      "loss": 0.9009,
+      "step": 15420
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000385375,
+      "loss": 0.9093,
+      "step": 15430
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038562499999999995,
+      "loss": 0.8835,
+      "step": 15440
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038587499999999993,
+      "loss": 0.9153,
+      "step": 15450
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038612499999999997,
+      "loss": 0.888,
+      "step": 15460
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038637499999999995,
+      "loss": 0.8992,
+      "step": 15470
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000386625,
+      "loss": 0.9111,
+      "step": 15480
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038687499999999996,
+      "loss": 0.9072,
+      "step": 15490
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000387125,
+      "loss": 0.8989,
+      "step": 15500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038737499999999997,
+      "loss": 0.9184,
+      "step": 15510
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038762499999999995,
+      "loss": 0.9062,
+      "step": 15520
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000387875,
+      "loss": 0.8942,
+      "step": 15530
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038812499999999996,
+      "loss": 0.8903,
+      "step": 15540
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000388375,
+      "loss": 0.9002,
+      "step": 15550
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000388625,
+      "loss": 0.8949,
+      "step": 15560
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038887499999999995,
+      "loss": 0.9059,
+      "step": 15570
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000389125,
+      "loss": 0.8925,
+      "step": 15580
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038937499999999996,
+      "loss": 0.9097,
+      "step": 15590
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000389625,
+      "loss": 0.902,
+      "step": 15600
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000389875,
+      "loss": 0.9065,
+      "step": 15610
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000390125,
+      "loss": 0.9019,
+      "step": 15620
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000390375,
+      "loss": 0.9129,
+      "step": 15630
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00039062499999999997,
+      "loss": 0.9301,
+      "step": 15640
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000390875,
+      "loss": 0.9264,
+      "step": 15650
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000391125,
+      "loss": 0.9174,
+      "step": 15660
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000391375,
+      "loss": 0.9326,
+      "step": 15670
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000391625,
+      "loss": 0.9315,
+      "step": 15680
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003918749999999999,
+      "loss": 0.9348,
+      "step": 15690
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000392125,
+      "loss": 0.9141,
+      "step": 15700
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00039237499999999993,
+      "loss": 0.9214,
+      "step": 15710
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000392625,
+      "loss": 0.9206,
+      "step": 15720
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00039285,
+      "loss": 0.9257,
+      "step": 15730
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00039309999999999996,
+      "loss": 0.9082,
+      "step": 15740
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00039334999999999994,
+      "loss": 0.9235,
+      "step": 15750
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00039359999999999997,
+      "loss": 0.9195,
+      "step": 15760
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00039384999999999995,
+      "loss": 0.9134,
+      "step": 15770
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003941,
+      "loss": 0.9079,
+      "step": 15780
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00039434999999999996,
+      "loss": 0.9062,
+      "step": 15790
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00039459999999999994,
+      "loss": 0.9015,
+      "step": 15800
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00039485,
+      "loss": 0.905,
+      "step": 15810
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00039509999999999995,
+      "loss": 0.917,
+      "step": 15820
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00039535,
+      "loss": 0.9032,
+      "step": 15830
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00039559999999999997,
+      "loss": 0.9049,
+      "step": 15840
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00039584999999999995,
+      "loss": 0.9183,
+      "step": 15850
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003961,
+      "loss": 0.9155,
+      "step": 15860
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00039634999999999996,
+      "loss": 0.9321,
+      "step": 15870
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003966,
+      "loss": 0.9194,
+      "step": 15880
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00039684999999999997,
+      "loss": 0.9276,
+      "step": 15890
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003971,
+      "loss": 0.9179,
+      "step": 15900
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00039735,
+      "loss": 0.9153,
+      "step": 15910
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00039759999999999996,
+      "loss": 0.9091,
+      "step": 15920
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00039785,
+      "loss": 0.8992,
+      "step": 15930
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003981,
+      "loss": 0.9248,
+      "step": 15940
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00039835,
+      "loss": 0.8834,
+      "step": 15950
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003986,
+      "loss": 0.8917,
+      "step": 15960
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003988499999999999,
+      "loss": 0.9172,
+      "step": 15970
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003991,
+      "loss": 0.9126,
+      "step": 15980
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003993499999999999,
+      "loss": 0.9144,
+      "step": 15990
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003996,
+      "loss": 0.927,
+      "step": 16000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00039984999999999993,
+      "loss": 0.9193,
+      "step": 16010
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004000999999999999,
+      "loss": 0.9001,
+      "step": 16020
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00040034999999999995,
+      "loss": 0.9074,
+      "step": 16030
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004005999999999999,
+      "loss": 0.8888,
+      "step": 16040
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00040084999999999996,
+      "loss": 0.9051,
+      "step": 16050
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00040109999999999994,
+      "loss": 0.886,
+      "step": 16060
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00040134999999999997,
+      "loss": 0.8988,
+      "step": 16070
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00040159999999999995,
+      "loss": 0.8789,
+      "step": 16080
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00040184999999999993,
+      "loss": 0.8828,
+      "step": 16090
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00040209999999999996,
+      "loss": 0.8723,
+      "step": 16100
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00040234999999999994,
+      "loss": 0.9006,
+      "step": 16110
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004026,
+      "loss": 0.8891,
+      "step": 16120
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00040284999999999995,
+      "loss": 0.8801,
+      "step": 16130
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00040309999999999993,
+      "loss": 0.8827,
+      "step": 16140
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00040334999999999997,
+      "loss": 0.8866,
+      "step": 16150
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00040359999999999994,
+      "loss": 0.8691,
+      "step": 16160
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00040385,
+      "loss": 0.8749,
+      "step": 16170
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00040409999999999996,
+      "loss": 0.8902,
+      "step": 16180
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00040435,
+      "loss": 0.9015,
+      "step": 16190
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00040459999999999997,
+      "loss": 0.9031,
+      "step": 16200
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00040484999999999995,
+      "loss": 0.9151,
+      "step": 16210
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004051,
+      "loss": 0.9132,
+      "step": 16220
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00040534999999999996,
+      "loss": 0.9079,
+      "step": 16230
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004056,
+      "loss": 0.918,
+      "step": 16240
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00040584999999999997,
+      "loss": 0.9121,
+      "step": 16250
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00040609999999999995,
+      "loss": 0.9106,
+      "step": 16260
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00040635,
+      "loss": 0.9031,
+      "step": 16270
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00040659999999999996,
+      "loss": 0.8931,
+      "step": 16280
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00040685,
+      "loss": 0.8999,
+      "step": 16290
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004071,
+      "loss": 0.9123,
+      "step": 16300
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00040734999999999995,
+      "loss": 0.9201,
+      "step": 16310
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004076,
+      "loss": 0.9105,
+      "step": 16320
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00040784999999999997,
+      "loss": 0.9038,
+      "step": 16330
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004081,
+      "loss": 0.8916,
+      "step": 16340
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00040835,
+      "loss": 0.9051,
+      "step": 16350
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004086,
+      "loss": 0.9035,
+      "step": 16360
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00040885,
+      "loss": 0.9139,
+      "step": 16370
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004090999999999999,
+      "loss": 0.9228,
+      "step": 16380
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00040935,
+      "loss": 0.9001,
+      "step": 16390
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00040959999999999993,
+      "loss": 0.8981,
+      "step": 16400
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00040985,
+      "loss": 0.8796,
+      "step": 16410
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00041009999999999994,
+      "loss": 0.8982,
+      "step": 16420
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004103499999999999,
+      "loss": 0.9092,
+      "step": 16430
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00041059999999999995,
+      "loss": 0.9078,
+      "step": 16440
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00041084999999999993,
+      "loss": 0.9136,
+      "step": 16450
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00041109999999999996,
+      "loss": 0.8812,
+      "step": 16460
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00041134999999999994,
+      "loss": 0.8807,
+      "step": 16470
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004116,
+      "loss": 0.893,
+      "step": 16480
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00041184999999999995,
+      "loss": 0.8879,
+      "step": 16490
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00041209999999999993,
+      "loss": 0.8825,
+      "step": 16500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00041234999999999997,
+      "loss": 0.8709,
+      "step": 16510
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00041259999999999995,
+      "loss": 0.8962,
+      "step": 16520
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00041285,
+      "loss": 0.8799,
+      "step": 16530
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00041309999999999996,
+      "loss": 0.9039,
+      "step": 16540
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00041334999999999994,
+      "loss": 0.8811,
+      "step": 16550
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00041359999999999997,
+      "loss": 0.8934,
+      "step": 16560
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00041384999999999995,
+      "loss": 0.8987,
+      "step": 16570
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004141,
+      "loss": 0.8806,
+      "step": 16580
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00041434999999999996,
+      "loss": 0.8867,
+      "step": 16590
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00041459999999999994,
+      "loss": 0.8716,
+      "step": 16600
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00041485,
+      "loss": 0.8798,
+      "step": 16610
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00041509999999999995,
+      "loss": 0.8875,
+      "step": 16620
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00041535,
+      "loss": 0.8752,
+      "step": 16630
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00041559999999999996,
+      "loss": 0.8721,
+      "step": 16640
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00041585,
+      "loss": 0.8783,
+      "step": 16650
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004161,
+      "loss": 0.8941,
+      "step": 16660
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00041634999999999996,
+      "loss": 0.9065,
+      "step": 16670
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004166,
+      "loss": 0.8996,
+      "step": 16680
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00041684999999999997,
+      "loss": 0.9105,
+      "step": 16690
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004171,
+      "loss": 0.9004,
+      "step": 16700
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00041735,
+      "loss": 0.8908,
+      "step": 16710
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00041759999999999996,
+      "loss": 0.8875,
+      "step": 16720
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004178249999999999,
+      "loss": 0.8938,
+      "step": 16730
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00041807499999999996,
+      "loss": 0.8937,
+      "step": 16740
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00041832499999999994,
+      "loss": 0.9222,
+      "step": 16750
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00041857499999999997,
+      "loss": 0.905,
+      "step": 16760
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00041882499999999995,
+      "loss": 0.9112,
+      "step": 16770
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00041907499999999993,
+      "loss": 0.875,
+      "step": 16780
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00041932499999999996,
+      "loss": 0.9051,
+      "step": 16790
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00041957499999999994,
+      "loss": 0.8955,
+      "step": 16800
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00041982499999999997,
+      "loss": 0.8845,
+      "step": 16810
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00042007499999999995,
+      "loss": 0.8922,
+      "step": 16820
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00042032499999999993,
+      "loss": 0.8942,
+      "step": 16830
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00042057499999999996,
+      "loss": 0.8848,
+      "step": 16840
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00042082499999999994,
+      "loss": 0.8988,
+      "step": 16850
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000421075,
+      "loss": 0.9023,
+      "step": 16860
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00042132499999999995,
+      "loss": 0.8887,
+      "step": 16870
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000421575,
+      "loss": 0.8788,
+      "step": 16880
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00042182499999999997,
+      "loss": 0.9045,
+      "step": 16890
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00042207499999999995,
+      "loss": 0.8972,
+      "step": 16900
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000422325,
+      "loss": 0.8951,
+      "step": 16910
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00042257499999999996,
+      "loss": 0.9044,
+      "step": 16920
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000422825,
+      "loss": 0.919,
+      "step": 16930
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00042307499999999997,
+      "loss": 0.9027,
+      "step": 16940
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00042332499999999995,
+      "loss": 0.9085,
+      "step": 16950
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000423575,
+      "loss": 0.9036,
+      "step": 16960
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00042382499999999996,
+      "loss": 0.8956,
+      "step": 16970
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000424075,
+      "loss": 0.8858,
+      "step": 16980
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000424325,
+      "loss": 0.8867,
+      "step": 16990
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00042457499999999995,
+      "loss": 0.8947,
+      "step": 17000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000424825,
+      "loss": 0.8837,
+      "step": 17010
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00042507499999999996,
+      "loss": 0.8845,
+      "step": 17020
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000425325,
+      "loss": 0.9002,
+      "step": 17030
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000425575,
+      "loss": 0.903,
+      "step": 17040
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000425825,
+      "loss": 0.9118,
+      "step": 17050
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000426075,
+      "loss": 0.8913,
+      "step": 17060
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004263249999999999,
+      "loss": 0.8945,
+      "step": 17070
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000426575,
+      "loss": 0.886,
+      "step": 17080
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004268249999999999,
+      "loss": 0.8845,
+      "step": 17090
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000427075,
+      "loss": 0.882,
+      "step": 17100
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00042732499999999994,
+      "loss": 0.901,
+      "step": 17110
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004275749999999999,
+      "loss": 0.8849,
+      "step": 17120
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00042782499999999995,
+      "loss": 0.8931,
+      "step": 17130
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00042807499999999993,
+      "loss": 0.886,
+      "step": 17140
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00042832499999999996,
+      "loss": 0.8738,
+      "step": 17150
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00042857499999999994,
+      "loss": 0.8694,
+      "step": 17160
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004288249999999999,
+      "loss": 0.8888,
+      "step": 17170
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00042907499999999995,
+      "loss": 0.8771,
+      "step": 17180
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00042932499999999993,
+      "loss": 0.8846,
+      "step": 17190
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00042957499999999997,
+      "loss": 0.8768,
+      "step": 17200
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00042982499999999994,
+      "loss": 0.8796,
+      "step": 17210
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000430075,
+      "loss": 0.8849,
+      "step": 17220
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00043032499999999996,
+      "loss": 0.9074,
+      "step": 17230
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00043057499999999994,
+      "loss": 0.8949,
+      "step": 17240
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00043082499999999997,
+      "loss": 0.8993,
+      "step": 17250
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00043107499999999995,
+      "loss": 0.9037,
+      "step": 17260
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000431325,
+      "loss": 0.888,
+      "step": 17270
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00043157499999999996,
+      "loss": 0.8889,
+      "step": 17280
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00043182499999999994,
+      "loss": 0.8938,
+      "step": 17290
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00043207499999999997,
+      "loss": 0.8967,
+      "step": 17300
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00043232499999999995,
+      "loss": 0.8824,
+      "step": 17310
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000432575,
+      "loss": 0.8823,
+      "step": 17320
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00043282499999999996,
+      "loss": 0.9003,
+      "step": 17330
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000433075,
+      "loss": 0.9115,
+      "step": 17340
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000433325,
+      "loss": 0.8907,
+      "step": 17350
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00043357499999999995,
+      "loss": 0.9,
+      "step": 17360
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000433825,
+      "loss": 0.8947,
+      "step": 17370
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00043407499999999997,
+      "loss": 0.8867,
+      "step": 17380
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000434325,
+      "loss": 0.8783,
+      "step": 17390
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000434575,
+      "loss": 0.8778,
+      "step": 17400
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00043482499999999996,
+      "loss": 0.8951,
+      "step": 17410
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000435075,
+      "loss": 0.8697,
+      "step": 17420
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00043532499999999997,
+      "loss": 0.887,
+      "step": 17430
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000435575,
+      "loss": 0.8831,
+      "step": 17440
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000435825,
+      "loss": 0.8838,
+      "step": 17450
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004360749999999999,
+      "loss": 0.8913,
+      "step": 17460
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000436325,
+      "loss": 0.8844,
+      "step": 17470
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004365749999999999,
+      "loss": 0.8904,
+      "step": 17480
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000436825,
+      "loss": 0.893,
+      "step": 17490
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00043707499999999993,
+      "loss": 0.8983,
+      "step": 17500
+    },
+    {
+      "epoch": 0.17,
+      "eval_accuracy": 0.8182743179218973,
+      "eval_loss": 0.91259765625,
+      "eval_runtime": 97.2312,
+      "eval_samples_per_second": 822.781,
+      "eval_steps_per_second": 1.615,
+      "step": 17500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000437325,
+      "loss": 0.8869,
+      "step": 17510
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00043757499999999994,
+      "loss": 0.892,
+      "step": 17520
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004378249999999999,
+      "loss": 0.8886,
+      "step": 17530
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00043807499999999996,
+      "loss": 0.8904,
+      "step": 17540
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00043832499999999993,
+      "loss": 0.8751,
+      "step": 17550
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00043857499999999997,
+      "loss": 0.8874,
+      "step": 17560
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00043882499999999995,
+      "loss": 0.8813,
+      "step": 17570
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004390749999999999,
+      "loss": 0.8831,
+      "step": 17580
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00043932499999999996,
+      "loss": 0.8802,
+      "step": 17590
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00043957499999999994,
+      "loss": 0.8797,
+      "step": 17600
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00043982499999999997,
+      "loss": 0.8671,
+      "step": 17610
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00044007499999999995,
+      "loss": 0.8755,
+      "step": 17620
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000440325,
+      "loss": 0.8715,
+      "step": 17630
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00044057499999999996,
+      "loss": 0.8605,
+      "step": 17640
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00044082499999999994,
+      "loss": 0.859,
+      "step": 17650
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000441075,
+      "loss": 0.8716,
+      "step": 17660
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00044132499999999995,
+      "loss": 0.8661,
+      "step": 17670
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000441575,
+      "loss": 0.8838,
+      "step": 17680
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00044182499999999997,
+      "loss": 0.8765,
+      "step": 17690
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00044207499999999994,
+      "loss": 0.866,
+      "step": 17700
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000442325,
+      "loss": 0.8841,
+      "step": 17710
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00044257499999999996,
+      "loss": 0.8905,
+      "step": 17720
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004428,
+      "loss": 0.8865,
+      "step": 17730
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00044305,
+      "loss": 0.8793,
+      "step": 17740
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004433,
+      "loss": 0.8794,
+      "step": 17750
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004435499999999999,
+      "loss": 0.875,
+      "step": 17760
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004438,
+      "loss": 0.8675,
+      "step": 17770
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004440499999999999,
+      "loss": 0.8694,
+      "step": 17780
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004443,
+      "loss": 0.8795,
+      "step": 17790
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00044454999999999994,
+      "loss": 0.8764,
+      "step": 17800
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004447999999999999,
+      "loss": 0.8657,
+      "step": 17810
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00044504999999999995,
+      "loss": 0.8679,
+      "step": 17820
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00044529999999999993,
+      "loss": 0.8732,
+      "step": 17830
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00044554999999999996,
+      "loss": 0.8791,
+      "step": 17840
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00044579999999999994,
+      "loss": 0.8636,
+      "step": 17850
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004460499999999999,
+      "loss": 0.882,
+      "step": 17860
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00044629999999999995,
+      "loss": 0.8777,
+      "step": 17870
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00044654999999999993,
+      "loss": 0.8735,
+      "step": 17880
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00044679999999999996,
+      "loss": 0.8776,
+      "step": 17890
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00044704999999999994,
+      "loss": 0.891,
+      "step": 17900
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004473,
+      "loss": 0.9082,
+      "step": 17910
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00044754999999999996,
+      "loss": 0.9046,
+      "step": 17920
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00044779999999999993,
+      "loss": 0.8918,
+      "step": 17930
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00044804999999999997,
+      "loss": 0.8849,
+      "step": 17940
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00044829999999999995,
+      "loss": 0.8936,
+      "step": 17950
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00044855,
+      "loss": 0.8865,
+      "step": 17960
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00044879999999999996,
+      "loss": 0.8825,
+      "step": 17970
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00044904999999999994,
+      "loss": 0.8978,
+      "step": 17980
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00044929999999999997,
+      "loss": 0.8707,
+      "step": 17990
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00044954999999999995,
+      "loss": 0.8807,
+      "step": 18000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004498,
+      "loss": 0.8942,
+      "step": 18010
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00045004999999999996,
+      "loss": 0.8926,
+      "step": 18020
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00045029999999999994,
+      "loss": 0.8825,
+      "step": 18030
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00045055,
+      "loss": 0.8993,
+      "step": 18040
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00045079999999999995,
+      "loss": 0.896,
+      "step": 18050
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00045105,
+      "loss": 0.8884,
+      "step": 18060
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00045129999999999997,
+      "loss": 0.8755,
+      "step": 18070
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00045155,
+      "loss": 0.8749,
+      "step": 18080
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004518,
+      "loss": 0.9011,
+      "step": 18090
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00045204999999999996,
+      "loss": 0.8811,
+      "step": 18100
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004523,
+      "loss": 0.8883,
+      "step": 18110
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00045254999999999997,
+      "loss": 0.8898,
+      "step": 18120
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004528,
+      "loss": 0.8912,
+      "step": 18130
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00045305,
+      "loss": 0.8892,
+      "step": 18140
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004532999999999999,
+      "loss": 0.8652,
+      "step": 18150
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00045355,
+      "loss": 0.8985,
+      "step": 18160
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004537999999999999,
+      "loss": 0.8885,
+      "step": 18170
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00045405,
+      "loss": 0.8844,
+      "step": 18180
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00045429999999999993,
+      "loss": 0.8754,
+      "step": 18190
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00045455,
+      "loss": 0.8633,
+      "step": 18200
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00045479999999999994,
+      "loss": 0.8808,
+      "step": 18210
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004550499999999999,
+      "loss": 0.8859,
+      "step": 18220
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00045529999999999995,
+      "loss": 0.8836,
+      "step": 18230
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00045554999999999993,
+      "loss": 0.885,
+      "step": 18240
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00045579999999999997,
+      "loss": 0.8675,
+      "step": 18250
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00045604999999999995,
+      "loss": 0.8838,
+      "step": 18260
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004562999999999999,
+      "loss": 0.8973,
+      "step": 18270
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00045654999999999996,
+      "loss": 0.8845,
+      "step": 18280
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00045679999999999994,
+      "loss": 0.8809,
+      "step": 18290
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00045704999999999997,
+      "loss": 0.8767,
+      "step": 18300
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00045729999999999995,
+      "loss": 0.8878,
+      "step": 18310
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00045754999999999993,
+      "loss": 0.8706,
+      "step": 18320
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00045779999999999996,
+      "loss": 0.884,
+      "step": 18330
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00045804999999999994,
+      "loss": 0.8678,
+      "step": 18340
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00045829999999999997,
+      "loss": 0.8785,
+      "step": 18350
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00045854999999999995,
+      "loss": 0.8933,
+      "step": 18360
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004588,
+      "loss": 0.8826,
+      "step": 18370
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00045904999999999996,
+      "loss": 0.8625,
+      "step": 18380
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00045929999999999994,
+      "loss": 0.8767,
+      "step": 18390
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00045955,
+      "loss": 0.8611,
+      "step": 18400
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00045979999999999995,
+      "loss": 0.8732,
+      "step": 18410
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00046005,
+      "loss": 0.8704,
+      "step": 18420
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00046029999999999997,
+      "loss": 0.8753,
+      "step": 18430
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00046054999999999995,
+      "loss": 0.8651,
+      "step": 18440
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004608,
+      "loss": 0.8572,
+      "step": 18450
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00046104999999999996,
+      "loss": 0.85,
+      "step": 18460
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004613,
+      "loss": 0.8867,
+      "step": 18470
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00046154999999999997,
+      "loss": 0.8704,
+      "step": 18480
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004618,
+      "loss": 0.8707,
+      "step": 18490
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00046205,
+      "loss": 0.8768,
+      "step": 18500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00046229999999999996,
+      "loss": 0.8776,
+      "step": 18510
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00046255,
+      "loss": 0.8515,
+      "step": 18520
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004628,
+      "loss": 0.8539,
+      "step": 18530
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00046305,
+      "loss": 0.8578,
+      "step": 18540
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004633,
+      "loss": 0.8909,
+      "step": 18550
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004635499999999999,
+      "loss": 0.8641,
+      "step": 18560
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004638,
+      "loss": 0.8626,
+      "step": 18570
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004640499999999999,
+      "loss": 0.8804,
+      "step": 18580
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004643,
+      "loss": 0.8824,
+      "step": 18590
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00046454999999999993,
+      "loss": 0.88,
+      "step": 18600
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004647999999999999,
+      "loss": 0.8648,
+      "step": 18610
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00046504999999999995,
+      "loss": 0.8716,
+      "step": 18620
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004652999999999999,
+      "loss": 0.871,
+      "step": 18630
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00046554999999999996,
+      "loss": 0.884,
+      "step": 18640
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00046579999999999994,
+      "loss": 0.8656,
+      "step": 18650
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00046604999999999997,
+      "loss": 0.8872,
+      "step": 18660
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00046629999999999995,
+      "loss": 0.8771,
+      "step": 18670
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00046654999999999993,
+      "loss": 0.8846,
+      "step": 18680
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00046679999999999996,
+      "loss": 0.8677,
+      "step": 18690
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00046704999999999994,
+      "loss": 0.8832,
+      "step": 18700
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004673,
+      "loss": 0.8815,
+      "step": 18710
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00046754999999999995,
+      "loss": 0.8752,
+      "step": 18720
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00046777499999999997,
+      "loss": 0.8859,
+      "step": 18730
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00046802499999999995,
+      "loss": 0.8812,
+      "step": 18740
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000468275,
+      "loss": 0.8743,
+      "step": 18750
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00046852499999999996,
+      "loss": 0.8647,
+      "step": 18760
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000468775,
+      "loss": 0.8854,
+      "step": 18770
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000469025,
+      "loss": 0.8724,
+      "step": 18780
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00046927499999999995,
+      "loss": 0.8696,
+      "step": 18790
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000469525,
+      "loss": 0.8873,
+      "step": 18800
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00046977499999999997,
+      "loss": 0.8876,
+      "step": 18810
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000470025,
+      "loss": 0.874,
+      "step": 18820
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000470275,
+      "loss": 0.8747,
+      "step": 18830
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004705249999999999,
+      "loss": 0.8735,
+      "step": 18840
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000470775,
+      "loss": 0.8863,
+      "step": 18850
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004710249999999999,
+      "loss": 0.8821,
+      "step": 18860
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000471275,
+      "loss": 0.8747,
+      "step": 18870
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00047152499999999993,
+      "loss": 0.8709,
+      "step": 18880
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000471775,
+      "loss": 0.8837,
+      "step": 18890
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00047202499999999994,
+      "loss": 0.8673,
+      "step": 18900
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004722749999999999,
+      "loss": 0.8712,
+      "step": 18910
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00047252499999999995,
+      "loss": 0.8723,
+      "step": 18920
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00047277499999999993,
+      "loss": 0.8574,
+      "step": 18930
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00047302499999999996,
+      "loss": 0.8789,
+      "step": 18940
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00047327499999999994,
+      "loss": 0.8872,
+      "step": 18950
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004735249999999999,
+      "loss": 0.8802,
+      "step": 18960
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00047377499999999996,
+      "loss": 0.8921,
+      "step": 18970
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00047402499999999993,
+      "loss": 0.8751,
+      "step": 18980
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00047427499999999997,
+      "loss": 0.8665,
+      "step": 18990
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00047452499999999995,
+      "loss": 0.8721,
+      "step": 19000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004747749999999999,
+      "loss": 0.8839,
+      "step": 19010
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00047502499999999996,
+      "loss": 0.8676,
+      "step": 19020
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00047527499999999994,
+      "loss": 0.8704,
+      "step": 19030
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00047552499999999997,
+      "loss": 0.8782,
+      "step": 19040
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00047577499999999995,
+      "loss": 0.8763,
+      "step": 19050
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000476025,
+      "loss": 0.8463,
+      "step": 19060
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00047627499999999996,
+      "loss": 0.8617,
+      "step": 19070
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00047652499999999994,
+      "loss": 0.8501,
+      "step": 19080
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000476775,
+      "loss": 0.8741,
+      "step": 19090
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00047702499999999995,
+      "loss": 0.876,
+      "step": 19100
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000477275,
+      "loss": 0.8645,
+      "step": 19110
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00047752499999999997,
+      "loss": 0.8666,
+      "step": 19120
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00047777499999999994,
+      "loss": 0.869,
+      "step": 19130
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000478025,
+      "loss": 0.8542,
+      "step": 19140
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00047827499999999996,
+      "loss": 0.8722,
+      "step": 19150
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000478525,
+      "loss": 0.8885,
+      "step": 19160
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00047877499999999997,
+      "loss": 0.876,
+      "step": 19170
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00047902499999999995,
+      "loss": 0.8765,
+      "step": 19180
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000479275,
+      "loss": 0.8642,
+      "step": 19190
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00047952499999999996,
+      "loss": 0.8772,
+      "step": 19200
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000479775,
+      "loss": 0.8533,
+      "step": 19210
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00048002499999999997,
+      "loss": 0.8791,
+      "step": 19220
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000480275,
+      "loss": 0.8693,
+      "step": 19230
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000480525,
+      "loss": 0.871,
+      "step": 19240
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004807749999999999,
+      "loss": 0.8712,
+      "step": 19250
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000481025,
+      "loss": 0.8774,
+      "step": 19260
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004812749999999999,
+      "loss": 0.8788,
+      "step": 19270
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000481525,
+      "loss": 0.873,
+      "step": 19280
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00048177499999999993,
+      "loss": 0.8738,
+      "step": 19290
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004820249999999999,
+      "loss": 0.8689,
+      "step": 19300
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00048227499999999995,
+      "loss": 0.8711,
+      "step": 19310
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004825249999999999,
+      "loss": 0.8706,
+      "step": 19320
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00048277499999999996,
+      "loss": 0.8562,
+      "step": 19330
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00048302499999999994,
+      "loss": 0.8691,
+      "step": 19340
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00048327499999999997,
+      "loss": 0.8735,
+      "step": 19350
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00048352499999999995,
+      "loss": 0.8635,
+      "step": 19360
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00048377499999999993,
+      "loss": 0.855,
+      "step": 19370
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00048402499999999996,
+      "loss": 0.8673,
+      "step": 19380
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00048427499999999994,
+      "loss": 0.8831,
+      "step": 19390
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000484525,
+      "loss": 0.8582,
+      "step": 19400
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00048477499999999995,
+      "loss": 0.8718,
+      "step": 19410
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00048502499999999993,
+      "loss": 0.8744,
+      "step": 19420
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00048527499999999996,
+      "loss": 0.8592,
+      "step": 19430
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00048552499999999994,
+      "loss": 0.8467,
+      "step": 19440
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000485775,
+      "loss": 0.8335,
+      "step": 19450
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00048602499999999996,
+      "loss": 0.8515,
+      "step": 19460
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00048627499999999993,
+      "loss": 0.8433,
+      "step": 19470
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00048652499999999997,
+      "loss": 0.85,
+      "step": 19480
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00048677499999999995,
+      "loss": 0.8501,
+      "step": 19490
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000487025,
+      "loss": 0.8527,
+      "step": 19500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00048727499999999996,
+      "loss": 0.8431,
+      "step": 19510
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000487525,
+      "loss": 0.8531,
+      "step": 19520
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00048777499999999997,
+      "loss": 0.8515,
+      "step": 19530
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00048802499999999995,
+      "loss": 0.8508,
+      "step": 19540
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000488275,
+      "loss": 0.846,
+      "step": 19550
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000488525,
+      "loss": 0.8413,
+      "step": 19560
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000488775,
+      "loss": 0.8532,
+      "step": 19570
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004890249999999999,
+      "loss": 0.8424,
+      "step": 19580
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000489275,
+      "loss": 0.8488,
+      "step": 19590
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000489525,
+      "loss": 0.8475,
+      "step": 19600
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004897749999999999,
+      "loss": 0.8432,
+      "step": 19610
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004900249999999999,
+      "loss": 0.8494,
+      "step": 19620
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000490275,
+      "loss": 0.849,
+      "step": 19630
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000490525,
+      "loss": 0.8531,
+      "step": 19640
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004907749999999999,
+      "loss": 0.8527,
+      "step": 19650
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000491025,
+      "loss": 0.8762,
+      "step": 19660
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000491275,
+      "loss": 0.867,
+      "step": 19670
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004915249999999999,
+      "loss": 0.8551,
+      "step": 19680
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000491775,
+      "loss": 0.854,
+      "step": 19690
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000492025,
+      "loss": 0.868,
+      "step": 19700
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004922749999999999,
+      "loss": 0.8572,
+      "step": 19710
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000492525,
+      "loss": 0.863,
+      "step": 19720
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000492775,
+      "loss": 0.8684,
+      "step": 19730
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000493,
+      "loss": 0.8623,
+      "step": 19740
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00049325,
+      "loss": 0.8543,
+      "step": 19750
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004935,
+      "loss": 0.8525,
+      "step": 19760
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004937499999999999,
+      "loss": 0.8636,
+      "step": 19770
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000494,
+      "loss": 0.8801,
+      "step": 19780
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004942499999999999,
+      "loss": 0.8639,
+      "step": 19790
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004944999999999999,
+      "loss": 0.8695,
+      "step": 19800
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00049475,
+      "loss": 0.8659,
+      "step": 19810
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004949999999999999,
+      "loss": 0.864,
+      "step": 19820
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004952499999999999,
+      "loss": 0.8424,
+      "step": 19830
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004955,
+      "loss": 0.8659,
+      "step": 19840
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00049575,
+      "loss": 0.8535,
+      "step": 19850
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004959999999999999,
+      "loss": 0.871,
+      "step": 19860
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00049625,
+      "loss": 0.8599,
+      "step": 19870
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004965,
+      "loss": 0.849,
+      "step": 19880
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004967499999999999,
+      "loss": 0.8578,
+      "step": 19890
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004969999999999999,
+      "loss": 0.8619,
+      "step": 19900
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00049725,
+      "loss": 0.8659,
+      "step": 19910
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004975,
+      "loss": 0.866,
+      "step": 19920
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004977499999999999,
+      "loss": 0.8616,
+      "step": 19930
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000498,
+      "loss": 0.8478,
+      "step": 19940
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00049825,
+      "loss": 0.8596,
+      "step": 19950
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004984999999999999,
+      "loss": 0.862,
+      "step": 19960
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00049875,
+      "loss": 0.8678,
+      "step": 19970
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000499,
+      "loss": 0.879,
+      "step": 19980
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004992499999999999,
+      "loss": 0.8744,
+      "step": 19990
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004994999999999999,
+      "loss": 0.8751,
+      "step": 20000
+    },
+    {
+      "epoch": 0.2,
+      "eval_accuracy": 0.8198587930184829,
+      "eval_loss": 0.89892578125,
+      "eval_runtime": 96.5681,
+      "eval_samples_per_second": 828.431,
+      "eval_steps_per_second": 1.626,
+      "step": 20000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00049975,
+      "loss": 0.8638,
+      "step": 20010
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005,
+      "loss": 0.8699,
+      "step": 20020
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005002499999999999,
+      "loss": 0.8509,
+      "step": 20030
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005005,
+      "loss": 0.853,
+      "step": 20040
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00050075,
+      "loss": 0.8575,
+      "step": 20050
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005009999999999999,
+      "loss": 0.8483,
+      "step": 20060
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00050125,
+      "loss": 0.8648,
+      "step": 20070
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005015,
+      "loss": 0.8515,
+      "step": 20080
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00050175,
+      "loss": 0.8699,
+      "step": 20090
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000502,
+      "loss": 0.8394,
+      "step": 20100
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00050225,
+      "loss": 0.8705,
+      "step": 20110
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005025,
+      "loss": 0.8578,
+      "step": 20120
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005027499999999999,
+      "loss": 0.8525,
+      "step": 20130
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000503,
+      "loss": 0.8574,
+      "step": 20140
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005032499999999999,
+      "loss": 0.8574,
+      "step": 20150
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005034999999999999,
+      "loss": 0.8503,
+      "step": 20160
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00050375,
+      "loss": 0.859,
+      "step": 20170
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005039999999999999,
+      "loss": 0.8582,
+      "step": 20180
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005042499999999999,
+      "loss": 0.8535,
+      "step": 20190
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005045,
+      "loss": 0.8333,
+      "step": 20200
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00050475,
+      "loss": 0.8377,
+      "step": 20210
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005049999999999999,
+      "loss": 0.8509,
+      "step": 20220
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00050525,
+      "loss": 0.8661,
+      "step": 20230
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005055,
+      "loss": 0.8696,
+      "step": 20240
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005057499999999999,
+      "loss": 0.8537,
+      "step": 20250
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005059999999999999,
+      "loss": 0.8625,
+      "step": 20260
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00050625,
+      "loss": 0.859,
+      "step": 20270
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005064999999999999,
+      "loss": 0.8617,
+      "step": 20280
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005067499999999999,
+      "loss": 0.8424,
+      "step": 20290
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000507,
+      "loss": 0.8521,
+      "step": 20300
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00050725,
+      "loss": 0.8474,
+      "step": 20310
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005074999999999999,
+      "loss": 0.8544,
+      "step": 20320
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00050775,
+      "loss": 0.8454,
+      "step": 20330
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000508,
+      "loss": 0.8555,
+      "step": 20340
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005082499999999999,
+      "loss": 0.8385,
+      "step": 20350
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005085,
+      "loss": 0.8507,
+      "step": 20360
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00050875,
+      "loss": 0.8535,
+      "step": 20370
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000509,
+      "loss": 0.8636,
+      "step": 20380
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005092499999999999,
+      "loss": 0.8672,
+      "step": 20390
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005095,
+      "loss": 0.8736,
+      "step": 20400
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00050975,
+      "loss": 0.8626,
+      "step": 20410
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005099999999999999,
+      "loss": 0.8581,
+      "step": 20420
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00051025,
+      "loss": 0.8369,
+      "step": 20430
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005105,
+      "loss": 0.8421,
+      "step": 20440
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005107499999999999,
+      "loss": 0.8572,
+      "step": 20450
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000511,
+      "loss": 0.8611,
+      "step": 20460
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00051125,
+      "loss": 0.849,
+      "step": 20470
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005115,
+      "loss": 0.8553,
+      "step": 20480
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005117499999999999,
+      "loss": 0.8542,
+      "step": 20490
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000512,
+      "loss": 0.851,
+      "step": 20500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00051225,
+      "loss": 0.8633,
+      "step": 20510
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005124999999999999,
+      "loss": 0.8568,
+      "step": 20520
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00051275,
+      "loss": 0.8648,
+      "step": 20530
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005129999999999999,
+      "loss": 0.8609,
+      "step": 20540
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00051325,
+      "loss": 0.872,
+      "step": 20550
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005135,
+      "loss": 0.8543,
+      "step": 20560
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005137499999999999,
+      "loss": 0.8456,
+      "step": 20570
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005139999999999999,
+      "loss": 0.8625,
+      "step": 20580
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00051425,
+      "loss": 0.863,
+      "step": 20590
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005145,
+      "loss": 0.8661,
+      "step": 20600
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005147499999999999,
+      "loss": 0.853,
+      "step": 20610
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005149999999999999,
+      "loss": 0.8355,
+      "step": 20620
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00051525,
+      "loss": 0.8516,
+      "step": 20630
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005154999999999999,
+      "loss": 0.8582,
+      "step": 20640
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005157499999999999,
+      "loss": 0.8616,
+      "step": 20650
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000516,
+      "loss": 0.872,
+      "step": 20660
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00051625,
+      "loss": 0.8662,
+      "step": 20670
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005164999999999999,
+      "loss": 0.8629,
+      "step": 20680
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00051675,
+      "loss": 0.8667,
+      "step": 20690
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000517,
+      "loss": 0.8714,
+      "step": 20700
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005172499999999999,
+      "loss": 0.8544,
+      "step": 20710
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005175,
+      "loss": 0.858,
+      "step": 20720
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00051775,
+      "loss": 0.8394,
+      "step": 20730
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000517975,
+      "loss": 0.8487,
+      "step": 20740
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000518225,
+      "loss": 0.8566,
+      "step": 20750
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000518475,
+      "loss": 0.8522,
+      "step": 20760
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005187249999999999,
+      "loss": 0.8654,
+      "step": 20770
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000518975,
+      "loss": 0.8526,
+      "step": 20780
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000519225,
+      "loss": 0.8596,
+      "step": 20790
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005194749999999999,
+      "loss": 0.8721,
+      "step": 20800
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000519725,
+      "loss": 0.8496,
+      "step": 20810
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000519975,
+      "loss": 0.8524,
+      "step": 20820
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000520225,
+      "loss": 0.8407,
+      "step": 20830
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000520475,
+      "loss": 0.842,
+      "step": 20840
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005207249999999999,
+      "loss": 0.8428,
+      "step": 20850
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000520975,
+      "loss": 0.8262,
+      "step": 20860
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005212249999999999,
+      "loss": 0.8374,
+      "step": 20870
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000521475,
+      "loss": 0.8292,
+      "step": 20880
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005217249999999999,
+      "loss": 0.8493,
+      "step": 20890
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005219749999999999,
+      "loss": 0.8546,
+      "step": 20900
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000522225,
+      "loss": 0.8577,
+      "step": 20910
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005224749999999999,
+      "loss": 0.8526,
+      "step": 20920
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005227249999999999,
+      "loss": 0.8755,
+      "step": 20930
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000522975,
+      "loss": 0.8588,
+      "step": 20940
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000523225,
+      "loss": 0.8551,
+      "step": 20950
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005234749999999999,
+      "loss": 0.8439,
+      "step": 20960
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000523725,
+      "loss": 0.8404,
+      "step": 20970
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000523975,
+      "loss": 0.8425,
+      "step": 20980
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005242249999999999,
+      "loss": 0.8488,
+      "step": 20990
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005244749999999999,
+      "loss": 0.8544,
+      "step": 21000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000524725,
+      "loss": 0.8515,
+      "step": 21010
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005249749999999999,
+      "loss": 0.8592,
+      "step": 21020
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005252249999999999,
+      "loss": 0.8461,
+      "step": 21030
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000525475,
+      "loss": 0.8671,
+      "step": 21040
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000525725,
+      "loss": 0.848,
+      "step": 21050
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005259749999999999,
+      "loss": 0.8458,
+      "step": 21060
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000526225,
+      "loss": 0.8492,
+      "step": 21070
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000526475,
+      "loss": 0.8559,
+      "step": 21080
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005267249999999999,
+      "loss": 0.8533,
+      "step": 21090
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000526975,
+      "loss": 0.8399,
+      "step": 21100
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000527225,
+      "loss": 0.839,
+      "step": 21110
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000527475,
+      "loss": 0.8347,
+      "step": 21120
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005277249999999999,
+      "loss": 0.8422,
+      "step": 21130
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000527975,
+      "loss": 0.8471,
+      "step": 21140
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000528225,
+      "loss": 0.8427,
+      "step": 21150
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005284749999999999,
+      "loss": 0.8431,
+      "step": 21160
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000528725,
+      "loss": 0.8579,
+      "step": 21170
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000528975,
+      "loss": 0.8445,
+      "step": 21180
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005292249999999999,
+      "loss": 0.8808,
+      "step": 21190
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000529475,
+      "loss": 0.8514,
+      "step": 21200
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005297249999999999,
+      "loss": 0.8644,
+      "step": 21210
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000529975,
+      "loss": 0.8643,
+      "step": 21220
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005302249999999999,
+      "loss": 0.8652,
+      "step": 21230
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000530475,
+      "loss": 0.8588,
+      "step": 21240
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005307249999999999,
+      "loss": 0.8664,
+      "step": 21250
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005309749999999999,
+      "loss": 0.8559,
+      "step": 21260
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000531225,
+      "loss": 0.8528,
+      "step": 21270
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005314749999999999,
+      "loss": 0.8701,
+      "step": 21280
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005317249999999999,
+      "loss": 0.8587,
+      "step": 21290
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000531975,
+      "loss": 0.8486,
+      "step": 21300
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005322249999999999,
+      "loss": 0.834,
+      "step": 21310
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005324749999999999,
+      "loss": 0.8394,
+      "step": 21320
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000532725,
+      "loss": 0.8362,
+      "step": 21330
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000532975,
+      "loss": 0.8399,
+      "step": 21340
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005332249999999999,
+      "loss": 0.8561,
+      "step": 21350
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005334749999999999,
+      "loss": 0.8599,
+      "step": 21360
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000533725,
+      "loss": 0.8555,
+      "step": 21370
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005339749999999999,
+      "loss": 0.8606,
+      "step": 21380
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005342249999999999,
+      "loss": 0.8653,
+      "step": 21390
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000534475,
+      "loss": 0.8586,
+      "step": 21400
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000534725,
+      "loss": 0.8409,
+      "step": 21410
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005349749999999999,
+      "loss": 0.8412,
+      "step": 21420
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000535225,
+      "loss": 0.8464,
+      "step": 21430
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000535475,
+      "loss": 0.8684,
+      "step": 21440
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005357249999999999,
+      "loss": 0.8533,
+      "step": 21450
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000535975,
+      "loss": 0.8652,
+      "step": 21460
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000536225,
+      "loss": 0.8394,
+      "step": 21470
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005364749999999999,
+      "loss": 0.8395,
+      "step": 21480
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005367249999999999,
+      "loss": 0.8405,
+      "step": 21490
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000536975,
+      "loss": 0.861,
+      "step": 21500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000537225,
+      "loss": 0.8605,
+      "step": 21510
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005374749999999999,
+      "loss": 0.8449,
+      "step": 21520
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000537725,
+      "loss": 0.8417,
+      "step": 21530
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000537975,
+      "loss": 0.842,
+      "step": 21540
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005382249999999999,
+      "loss": 0.8467,
+      "step": 21550
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000538475,
+      "loss": 0.8455,
+      "step": 21560
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000538725,
+      "loss": 0.8414,
+      "step": 21570
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000538975,
+      "loss": 0.8568,
+      "step": 21580
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000539225,
+      "loss": 0.85,
+      "step": 21590
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005394749999999999,
+      "loss": 0.8696,
+      "step": 21600
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000539725,
+      "loss": 0.8536,
+      "step": 21610
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005399749999999999,
+      "loss": 0.8483,
+      "step": 21620
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000540225,
+      "loss": 0.8525,
+      "step": 21630
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005404749999999999,
+      "loss": 0.8526,
+      "step": 21640
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005407249999999999,
+      "loss": 0.8366,
+      "step": 21650
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000540975,
+      "loss": 0.8595,
+      "step": 21660
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005412249999999999,
+      "loss": 0.84,
+      "step": 21670
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005414749999999999,
+      "loss": 0.8594,
+      "step": 21680
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000541725,
+      "loss": 0.8491,
+      "step": 21690
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000541975,
+      "loss": 0.8567,
+      "step": 21700
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005422249999999999,
+      "loss": 0.8567,
+      "step": 21710
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005424749999999999,
+      "loss": 0.8429,
+      "step": 21720
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000542725,
+      "loss": 0.8601,
+      "step": 21730
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005429499999999999,
+      "loss": 0.8466,
+      "step": 21740
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005432,
+      "loss": 0.8704,
+      "step": 21750
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00054345,
+      "loss": 0.851,
+      "step": 21760
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005436999999999999,
+      "loss": 0.8732,
+      "step": 21770
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00054395,
+      "loss": 0.8686,
+      "step": 21780
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005442,
+      "loss": 0.852,
+      "step": 21790
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005444499999999999,
+      "loss": 0.8505,
+      "step": 21800
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005447,
+      "loss": 0.8207,
+      "step": 21810
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00054495,
+      "loss": 0.8405,
+      "step": 21820
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005451999999999999,
+      "loss": 0.8289,
+      "step": 21830
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00054545,
+      "loss": 0.8277,
+      "step": 21840
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005457,
+      "loss": 0.848,
+      "step": 21850
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00054595,
+      "loss": 0.8432,
+      "step": 21860
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005461999999999999,
+      "loss": 0.8451,
+      "step": 21870
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00054645,
+      "loss": 0.8518,
+      "step": 21880
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005467,
+      "loss": 0.8537,
+      "step": 21890
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005469499999999999,
+      "loss": 0.8512,
+      "step": 21900
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005472,
+      "loss": 0.8544,
+      "step": 21910
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00054745,
+      "loss": 0.8408,
+      "step": 21920
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005477,
+      "loss": 0.8433,
+      "step": 21930
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00054795,
+      "loss": 0.8379,
+      "step": 21940
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005481999999999999,
+      "loss": 0.8502,
+      "step": 21950
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005484499999999999,
+      "loss": 0.8576,
+      "step": 21960
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005487,
+      "loss": 0.8571,
+      "step": 21970
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00054895,
+      "loss": 0.8501,
+      "step": 21980
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005491999999999999,
+      "loss": 0.8382,
+      "step": 21990
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005494499999999999,
+      "loss": 0.8386,
+      "step": 22000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005497,
+      "loss": 0.8449,
+      "step": 22010
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005499499999999999,
+      "loss": 0.855,
+      "step": 22020
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005501999999999999,
+      "loss": 0.8483,
+      "step": 22030
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00055045,
+      "loss": 0.8802,
+      "step": 22040
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005506999999999999,
+      "loss": 0.8487,
+      "step": 22050
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005509499999999999,
+      "loss": 0.8665,
+      "step": 22060
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005512,
+      "loss": 0.8622,
+      "step": 22070
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00055145,
+      "loss": 0.8407,
+      "step": 22080
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005516999999999999,
+      "loss": 0.8573,
+      "step": 22090
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005519499999999999,
+      "loss": 0.8576,
+      "step": 22100
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005522,
+      "loss": 0.8431,
+      "step": 22110
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005524499999999999,
+      "loss": 0.8333,
+      "step": 22120
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005526999999999999,
+      "loss": 0.8526,
+      "step": 22130
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00055295,
+      "loss": 0.8583,
+      "step": 22140
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005532,
+      "loss": 0.8526,
+      "step": 22150
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005534499999999999,
+      "loss": 0.8519,
+      "step": 22160
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005537,
+      "loss": 0.8788,
+      "step": 22170
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00055395,
+      "loss": 0.8524,
+      "step": 22180
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005541999999999999,
+      "loss": 0.8541,
+      "step": 22190
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00055445,
+      "loss": 0.8464,
+      "step": 22200
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005547,
+      "loss": 0.8489,
+      "step": 22210
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00055495,
+      "loss": 0.84,
+      "step": 22220
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005551999999999999,
+      "loss": 0.8484,
+      "step": 22230
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00055545,
+      "loss": 0.8545,
+      "step": 22240
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005557,
+      "loss": 0.8498,
+      "step": 22250
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005559499999999999,
+      "loss": 0.8379,
+      "step": 22260
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005562,
+      "loss": 0.8475,
+      "step": 22270
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00055645,
+      "loss": 0.8605,
+      "step": 22280
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005566999999999999,
+      "loss": 0.8661,
+      "step": 22290
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00055695,
+      "loss": 0.8457,
+      "step": 22300
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005571999999999999,
+      "loss": 0.8488,
+      "step": 22310
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00055745,
+      "loss": 0.8541,
+      "step": 22320
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005577,
+      "loss": 0.8565,
+      "step": 22330
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005579499999999999,
+      "loss": 0.8472,
+      "step": 22340
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005581999999999999,
+      "loss": 0.8344,
+      "step": 22350
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005584499999999999,
+      "loss": 0.8216,
+      "step": 22360
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005587,
+      "loss": 0.8391,
+      "step": 22370
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005589499999999999,
+      "loss": 0.8371,
+      "step": 22380
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005591999999999999,
+      "loss": 0.85,
+      "step": 22390
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00055945,
+      "loss": 0.8419,
+      "step": 22400
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005596999999999999,
+      "loss": 0.8556,
+      "step": 22410
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005599499999999999,
+      "loss": 0.8554,
+      "step": 22420
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005602,
+      "loss": 0.8672,
+      "step": 22430
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00056045,
+      "loss": 0.8626,
+      "step": 22440
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005606999999999999,
+      "loss": 0.8509,
+      "step": 22450
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00056095,
+      "loss": 0.8573,
+      "step": 22460
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005612,
+      "loss": 0.8384,
+      "step": 22470
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005614499999999999,
+      "loss": 0.8403,
+      "step": 22480
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005616999999999999,
+      "loss": 0.8321,
+      "step": 22490
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00056195,
+      "loss": 0.8331,
+      "step": 22500
+    },
+    {
+      "epoch": 0.23,
+      "eval_accuracy": 0.8251687860262823,
+      "eval_loss": 0.86767578125,
+      "eval_runtime": 96.699,
+      "eval_samples_per_second": 827.309,
+      "eval_steps_per_second": 1.624,
+      "step": 22500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005622,
+      "loss": 0.8412,
+      "step": 22510
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005624499999999999,
+      "loss": 0.8565,
+      "step": 22520
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005627,
+      "loss": 0.8348,
+      "step": 22530
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00056295,
+      "loss": 0.8422,
+      "step": 22540
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005631999999999999,
+      "loss": 0.8546,
+      "step": 22550
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00056345,
+      "loss": 0.8493,
+      "step": 22560
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005637,
+      "loss": 0.8426,
+      "step": 22570
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005639499999999999,
+      "loss": 0.8362,
+      "step": 22580
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005641999999999999,
+      "loss": 0.8515,
+      "step": 22590
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00056445,
+      "loss": 0.8461,
+      "step": 22600
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005647,
+      "loss": 0.8419,
+      "step": 22610
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005649499999999999,
+      "loss": 0.8442,
+      "step": 22620
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005652,
+      "loss": 0.8306,
+      "step": 22630
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00056545,
+      "loss": 0.8336,
+      "step": 22640
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005656999999999999,
+      "loss": 0.8601,
+      "step": 22650
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00056595,
+      "loss": 0.8297,
+      "step": 22660
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005662,
+      "loss": 0.8361,
+      "step": 22670
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00056645,
+      "loss": 0.8433,
+      "step": 22680
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005667,
+      "loss": 0.85,
+      "step": 22690
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005669499999999999,
+      "loss": 0.8516,
+      "step": 22700
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005672,
+      "loss": 0.8673,
+      "step": 22710
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005674499999999999,
+      "loss": 0.858,
+      "step": 22720
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005677,
+      "loss": 0.8526,
+      "step": 22730
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005679249999999999,
+      "loss": 0.8419,
+      "step": 22740
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000568175,
+      "loss": 0.8424,
+      "step": 22750
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005684249999999999,
+      "loss": 0.8331,
+      "step": 22760
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005686749999999999,
+      "loss": 0.8234,
+      "step": 22770
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000568925,
+      "loss": 0.8343,
+      "step": 22780
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000569175,
+      "loss": 0.8336,
+      "step": 22790
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005694249999999999,
+      "loss": 0.8394,
+      "step": 22800
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000569675,
+      "loss": 0.8365,
+      "step": 22810
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000569925,
+      "loss": 0.8417,
+      "step": 22820
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005701749999999999,
+      "loss": 0.8479,
+      "step": 22830
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000570425,
+      "loss": 0.8533,
+      "step": 22840
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000570675,
+      "loss": 0.8259,
+      "step": 22850
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005709249999999999,
+      "loss": 0.8621,
+      "step": 22860
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005711749999999999,
+      "loss": 0.8413,
+      "step": 22870
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000571425,
+      "loss": 0.8336,
+      "step": 22880
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000571675,
+      "loss": 0.8262,
+      "step": 22890
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005719249999999999,
+      "loss": 0.8366,
+      "step": 22900
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000572175,
+      "loss": 0.8295,
+      "step": 22910
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000572425,
+      "loss": 0.8373,
+      "step": 22920
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005726749999999999,
+      "loss": 0.8466,
+      "step": 22930
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000572925,
+      "loss": 0.8341,
+      "step": 22940
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000573175,
+      "loss": 0.8429,
+      "step": 22950
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000573425,
+      "loss": 0.8465,
+      "step": 22960
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005736749999999999,
+      "loss": 0.8499,
+      "step": 22970
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000573925,
+      "loss": 0.8308,
+      "step": 22980
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000574175,
+      "loss": 0.8424,
+      "step": 22990
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005744249999999999,
+      "loss": 0.8312,
+      "step": 23000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000574675,
+      "loss": 0.8453,
+      "step": 23010
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005749249999999999,
+      "loss": 0.8476,
+      "step": 23020
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005751749999999999,
+      "loss": 0.8329,
+      "step": 23030
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000575425,
+      "loss": 0.8365,
+      "step": 23040
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005756749999999999,
+      "loss": 0.8487,
+      "step": 23050
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005759249999999999,
+      "loss": 0.8407,
+      "step": 23060
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000576175,
+      "loss": 0.8449,
+      "step": 23070
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000576425,
+      "loss": 0.8644,
+      "step": 23080
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005766749999999999,
+      "loss": 0.8564,
+      "step": 23090
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005769249999999999,
+      "loss": 0.8385,
+      "step": 23100
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000577175,
+      "loss": 0.8285,
+      "step": 23110
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005774249999999999,
+      "loss": 0.8194,
+      "step": 23120
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005776749999999999,
+      "loss": 0.814,
+      "step": 23130
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000577925,
+      "loss": 0.8261,
+      "step": 23140
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005781749999999999,
+      "loss": 0.8141,
+      "step": 23150
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005784249999999999,
+      "loss": 0.8089,
+      "step": 23160
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000578675,
+      "loss": 0.8003,
+      "step": 23170
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000578925,
+      "loss": 0.8148,
+      "step": 23180
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005791749999999999,
+      "loss": 0.814,
+      "step": 23190
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000579425,
+      "loss": 0.8206,
+      "step": 23200
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000579675,
+      "loss": 0.7938,
+      "step": 23210
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005799249999999999,
+      "loss": 0.8371,
+      "step": 23220
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005801749999999999,
+      "loss": 0.8515,
+      "step": 23230
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000580425,
+      "loss": 0.853,
+      "step": 23240
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000580675,
+      "loss": 0.8462,
+      "step": 23250
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005809249999999999,
+      "loss": 0.8354,
+      "step": 23260
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000581175,
+      "loss": 0.8352,
+      "step": 23270
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000581425,
+      "loss": 0.8431,
+      "step": 23280
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005816749999999999,
+      "loss": 0.8469,
+      "step": 23290
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000581925,
+      "loss": 0.8279,
+      "step": 23300
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000582175,
+      "loss": 0.8322,
+      "step": 23310
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005824249999999999,
+      "loss": 0.8279,
+      "step": 23320
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000582675,
+      "loss": 0.8376,
+      "step": 23330
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000582925,
+      "loss": 0.8371,
+      "step": 23340
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000583175,
+      "loss": 0.8421,
+      "step": 23350
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005834249999999999,
+      "loss": 0.8337,
+      "step": 23360
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000583675,
+      "loss": 0.8473,
+      "step": 23370
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000583925,
+      "loss": 0.8464,
+      "step": 23380
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005841749999999999,
+      "loss": 0.8401,
+      "step": 23390
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000584425,
+      "loss": 0.849,
+      "step": 23400
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005846749999999999,
+      "loss": 0.8351,
+      "step": 23410
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000584925,
+      "loss": 0.8496,
+      "step": 23420
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000585175,
+      "loss": 0.8413,
+      "step": 23430
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005854249999999999,
+      "loss": 0.8396,
+      "step": 23440
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005856749999999999,
+      "loss": 0.8265,
+      "step": 23450
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005859249999999999,
+      "loss": 0.8421,
+      "step": 23460
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000586175,
+      "loss": 0.8418,
+      "step": 23470
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005864249999999999,
+      "loss": 0.8253,
+      "step": 23480
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005866749999999999,
+      "loss": 0.7979,
+      "step": 23490
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000586925,
+      "loss": 0.823,
+      "step": 23500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005871749999999999,
+      "loss": 0.839,
+      "step": 23510
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005874249999999999,
+      "loss": 0.8571,
+      "step": 23520
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000587675,
+      "loss": 0.8338,
+      "step": 23530
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000587925,
+      "loss": 0.8443,
+      "step": 23540
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005881749999999999,
+      "loss": 0.8499,
+      "step": 23550
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000588425,
+      "loss": 0.8371,
+      "step": 23560
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000588675,
+      "loss": 0.833,
+      "step": 23570
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005889249999999999,
+      "loss": 0.8266,
+      "step": 23580
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005891749999999999,
+      "loss": 0.8274,
+      "step": 23590
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000589425,
+      "loss": 0.8249,
+      "step": 23600
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005896749999999999,
+      "loss": 0.8405,
+      "step": 23610
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005899249999999999,
+      "loss": 0.8365,
+      "step": 23620
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000590175,
+      "loss": 0.8367,
+      "step": 23630
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000590425,
+      "loss": 0.8378,
+      "step": 23640
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005906749999999999,
+      "loss": 0.8411,
+      "step": 23650
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000590925,
+      "loss": 0.8432,
+      "step": 23660
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000591175,
+      "loss": 0.8525,
+      "step": 23670
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005914249999999999,
+      "loss": 0.8394,
+      "step": 23680
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000591675,
+      "loss": 0.8471,
+      "step": 23690
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000591925,
+      "loss": 0.8306,
+      "step": 23700
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000592175,
+      "loss": 0.8147,
+      "step": 23710
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005924249999999999,
+      "loss": 0.8235,
+      "step": 23720
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000592675,
+      "loss": 0.8362,
+      "step": 23730
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005928999999999999,
+      "loss": 0.8339,
+      "step": 23740
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00059315,
+      "loss": 0.839,
+      "step": 23750
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005933999999999999,
+      "loss": 0.8156,
+      "step": 23760
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005936499999999999,
+      "loss": 0.8529,
+      "step": 23770
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005939,
+      "loss": 0.8476,
+      "step": 23780
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005941499999999999,
+      "loss": 0.8443,
+      "step": 23790
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005943999999999999,
+      "loss": 0.8345,
+      "step": 23800
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00059465,
+      "loss": 0.8323,
+      "step": 23810
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005949,
+      "loss": 0.835,
+      "step": 23820
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005951499999999999,
+      "loss": 0.8255,
+      "step": 23830
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005953999999999999,
+      "loss": 0.8313,
+      "step": 23840
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00059565,
+      "loss": 0.8261,
+      "step": 23850
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005958999999999999,
+      "loss": 0.8401,
+      "step": 23860
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005961499999999999,
+      "loss": 0.8233,
+      "step": 23870
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005964,
+      "loss": 0.8306,
+      "step": 23880
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005966499999999999,
+      "loss": 0.843,
+      "step": 23890
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005968999999999999,
+      "loss": 0.8355,
+      "step": 23900
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00059715,
+      "loss": 0.8333,
+      "step": 23910
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005974,
+      "loss": 0.8079,
+      "step": 23920
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005976499999999999,
+      "loss": 0.8175,
+      "step": 23930
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005979,
+      "loss": 0.8355,
+      "step": 23940
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00059815,
+      "loss": 0.8416,
+      "step": 23950
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005983999999999999,
+      "loss": 0.8322,
+      "step": 23960
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005986499999999999,
+      "loss": 0.839,
+      "step": 23970
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005989,
+      "loss": 0.8322,
+      "step": 23980
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00059915,
+      "loss": 0.8334,
+      "step": 23990
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005993999999999999,
+      "loss": 0.8202,
+      "step": 24000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00059965,
+      "loss": 0.8244,
+      "step": 24010
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005999,
+      "loss": 0.8304,
+      "step": 24020
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005999526315789473,
+      "loss": 0.8353,
+      "step": 24030
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005998736842105262,
+      "loss": 0.8389,
+      "step": 24040
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005997947368421052,
+      "loss": 0.8423,
+      "step": 24050
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005997157894736841,
+      "loss": 0.8378,
+      "step": 24060
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005996368421052631,
+      "loss": 0.8273,
+      "step": 24070
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000599557894736842,
+      "loss": 0.8392,
+      "step": 24080
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000599478947368421,
+      "loss": 0.8372,
+      "step": 24090
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005993999999999999,
+      "loss": 0.8344,
+      "step": 24100
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005993210526315789,
+      "loss": 0.8248,
+      "step": 24110
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005992421052631578,
+      "loss": 0.8391,
+      "step": 24120
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005991631578947368,
+      "loss": 0.8254,
+      "step": 24130
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005990842105263157,
+      "loss": 0.8272,
+      "step": 24140
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005990052631578947,
+      "loss": 0.8237,
+      "step": 24150
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005989263157894736,
+      "loss": 0.8403,
+      "step": 24160
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005988473684210526,
+      "loss": 0.8436,
+      "step": 24170
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005987684210526315,
+      "loss": 0.8277,
+      "step": 24180
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005986894736842105,
+      "loss": 0.8273,
+      "step": 24190
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005986105263157894,
+      "loss": 0.8324,
+      "step": 24200
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005985315789473684,
+      "loss": 0.8262,
+      "step": 24210
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005984526315789473,
+      "loss": 0.8249,
+      "step": 24220
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005983736842105263,
+      "loss": 0.8301,
+      "step": 24230
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005982947368421052,
+      "loss": 0.8368,
+      "step": 24240
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005982157894736842,
+      "loss": 0.8256,
+      "step": 24250
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005981368421052631,
+      "loss": 0.8332,
+      "step": 24260
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000598057894736842,
+      "loss": 0.8273,
+      "step": 24270
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000597978947368421,
+      "loss": 0.823,
+      "step": 24280
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005979,
+      "loss": 0.8095,
+      "step": 24290
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005978210526315789,
+      "loss": 0.8352,
+      "step": 24300
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005977421052631578,
+      "loss": 0.8351,
+      "step": 24310
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005976631578947368,
+      "loss": 0.8267,
+      "step": 24320
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005975842105263157,
+      "loss": 0.8318,
+      "step": 24330
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005975052631578947,
+      "loss": 0.8272,
+      "step": 24340
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005974263157894736,
+      "loss": 0.8129,
+      "step": 24350
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005973473684210526,
+      "loss": 0.8201,
+      "step": 24360
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005972684210526315,
+      "loss": 0.8211,
+      "step": 24370
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005971894736842105,
+      "loss": 0.8217,
+      "step": 24380
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005971105263157894,
+      "loss": 0.8114,
+      "step": 24390
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005970315789473684,
+      "loss": 0.8189,
+      "step": 24400
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005969526315789473,
+      "loss": 0.8383,
+      "step": 24410
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005968736842105263,
+      "loss": 0.8187,
+      "step": 24420
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005967947368421052,
+      "loss": 0.8175,
+      "step": 24430
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005967157894736842,
+      "loss": 0.8105,
+      "step": 24440
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005966368421052631,
+      "loss": 0.8205,
+      "step": 24450
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005965578947368421,
+      "loss": 0.8291,
+      "step": 24460
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000596478947368421,
+      "loss": 0.8141,
+      "step": 24470
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005964,
+      "loss": 0.8309,
+      "step": 24480
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005963210526315789,
+      "loss": 0.8155,
+      "step": 24490
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005962421052631579,
+      "loss": 0.8303,
+      "step": 24500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005961631578947368,
+      "loss": 0.8319,
+      "step": 24510
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005960842105263158,
+      "loss": 0.8229,
+      "step": 24520
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005960052631578947,
+      "loss": 0.8236,
+      "step": 24530
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005959263157894737,
+      "loss": 0.8219,
+      "step": 24540
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005958473684210526,
+      "loss": 0.8101,
+      "step": 24550
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005957684210526314,
+      "loss": 0.8133,
+      "step": 24560
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005956894736842105,
+      "loss": 0.8224,
+      "step": 24570
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005956105263157895,
+      "loss": 0.8266,
+      "step": 24580
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005955315789473684,
+      "loss": 0.8204,
+      "step": 24590
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005954526315789473,
+      "loss": 0.8312,
+      "step": 24600
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005953736842105263,
+      "loss": 0.8187,
+      "step": 24610
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005952947368421052,
+      "loss": 0.8262,
+      "step": 24620
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005952157894736841,
+      "loss": 0.8141,
+      "step": 24630
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005951368421052631,
+      "loss": 0.8192,
+      "step": 24640
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000595057894736842,
+      "loss": 0.8178,
+      "step": 24650
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000594978947368421,
+      "loss": 0.8326,
+      "step": 24660
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005949,
+      "loss": 0.8171,
+      "step": 24670
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005948210526315789,
+      "loss": 0.8305,
+      "step": 24680
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005947421052631579,
+      "loss": 0.8214,
+      "step": 24690
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005946631578947367,
+      "loss": 0.8161,
+      "step": 24700
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005945842105263158,
+      "loss": 0.8281,
+      "step": 24710
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005945052631578946,
+      "loss": 0.8173,
+      "step": 24720
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005944263157894737,
+      "loss": 0.819,
+      "step": 24730
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005943473684210525,
+      "loss": 0.8153,
+      "step": 24740
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005942763157894737,
+      "loss": 0.8132,
+      "step": 24750
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005941973684210525,
+      "loss": 0.8105,
+      "step": 24760
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005941184210526316,
+      "loss": 0.8286,
+      "step": 24770
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005940394736842104,
+      "loss": 0.8345,
+      "step": 24780
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005939605263157894,
+      "loss": 0.8198,
+      "step": 24790
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005938815789473683,
+      "loss": 0.8296,
+      "step": 24800
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005938026315789473,
+      "loss": 0.8112,
+      "step": 24810
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005937236842105263,
+      "loss": 0.8217,
+      "step": 24820
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005936447368421052,
+      "loss": 0.8239,
+      "step": 24830
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005935657894736842,
+      "loss": 0.8473,
+      "step": 24840
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005934868421052631,
+      "loss": 0.8215,
+      "step": 24850
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000593407894736842,
+      "loss": 0.798,
+      "step": 24860
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000593328947368421,
+      "loss": 0.8289,
+      "step": 24870
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005932499999999999,
+      "loss": 0.8461,
+      "step": 24880
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005931710526315789,
+      "loss": 0.8384,
+      "step": 24890
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005930921052631578,
+      "loss": 0.8234,
+      "step": 24900
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005930131578947369,
+      "loss": 0.8142,
+      "step": 24910
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005929342105263157,
+      "loss": 0.8214,
+      "step": 24920
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005928552631578947,
+      "loss": 0.8247,
+      "step": 24930
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005927763157894736,
+      "loss": 0.8213,
+      "step": 24940
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005926973684210526,
+      "loss": 0.8267,
+      "step": 24950
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005926184210526315,
+      "loss": 0.8101,
+      "step": 24960
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005925394736842105,
+      "loss": 0.8124,
+      "step": 24970
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005924605263157894,
+      "loss": 0.8121,
+      "step": 24980
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005923815789473684,
+      "loss": 0.8292,
+      "step": 24990
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005923026315789473,
+      "loss": 0.8129,
+      "step": 25000
+    },
+    {
+      "epoch": 0.25,
+      "eval_accuracy": 0.8254310925043975,
+      "eval_loss": 0.86767578125,
+      "eval_runtime": 96.7476,
+      "eval_samples_per_second": 826.894,
+      "eval_steps_per_second": 1.623,
+      "step": 25000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005922236842105262,
+      "loss": 0.8243,
+      "step": 25010
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005921447368421052,
+      "loss": 0.8351,
+      "step": 25020
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005920657894736841,
+      "loss": 0.8311,
+      "step": 25030
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005919868421052631,
+      "loss": 0.8154,
+      "step": 25040
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000591907894736842,
+      "loss": 0.8111,
+      "step": 25050
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000591828947368421,
+      "loss": 0.8188,
+      "step": 25060
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005917499999999999,
+      "loss": 0.8169,
+      "step": 25070
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005916710526315789,
+      "loss": 0.8221,
+      "step": 25080
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005915921052631578,
+      "loss": 0.8105,
+      "step": 25090
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005915131578947368,
+      "loss": 0.8269,
+      "step": 25100
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005914342105263157,
+      "loss": 0.8161,
+      "step": 25110
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005913552631578947,
+      "loss": 0.815,
+      "step": 25120
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005912763157894736,
+      "loss": 0.8187,
+      "step": 25130
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005911973684210526,
+      "loss": 0.8193,
+      "step": 25140
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005911184210526315,
+      "loss": 0.8331,
+      "step": 25150
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005910394736842105,
+      "loss": 0.818,
+      "step": 25160
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005909605263157894,
+      "loss": 0.8096,
+      "step": 25170
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005908815789473684,
+      "loss": 0.7797,
+      "step": 25180
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005908026315789473,
+      "loss": 0.8,
+      "step": 25190
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005907236842105263,
+      "loss": 0.7881,
+      "step": 25200
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005906447368421052,
+      "loss": 0.7861,
+      "step": 25210
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005905657894736842,
+      "loss": 0.793,
+      "step": 25220
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005904868421052631,
+      "loss": 0.8112,
+      "step": 25230
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005904078947368421,
+      "loss": 0.8275,
+      "step": 25240
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000590328947368421,
+      "loss": 0.8055,
+      "step": 25250
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00059025,
+      "loss": 0.8217,
+      "step": 25260
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005901710526315789,
+      "loss": 0.8035,
+      "step": 25270
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005900921052631579,
+      "loss": 0.8117,
+      "step": 25280
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005900131578947368,
+      "loss": 0.8194,
+      "step": 25290
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005899342105263157,
+      "loss": 0.8228,
+      "step": 25300
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005898552631578947,
+      "loss": 0.7978,
+      "step": 25310
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005897763157894736,
+      "loss": 0.8268,
+      "step": 25320
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005896973684210526,
+      "loss": 0.807,
+      "step": 25330
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005896184210526315,
+      "loss": 0.81,
+      "step": 25340
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005895394736842105,
+      "loss": 0.8109,
+      "step": 25350
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005894605263157894,
+      "loss": 0.8241,
+      "step": 25360
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005893815789473684,
+      "loss": 0.8234,
+      "step": 25370
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005893026315789473,
+      "loss": 0.8096,
+      "step": 25380
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005892236842105263,
+      "loss": 0.8234,
+      "step": 25390
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005891447368421052,
+      "loss": 0.8193,
+      "step": 25400
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005890657894736842,
+      "loss": 0.8166,
+      "step": 25410
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005889868421052631,
+      "loss": 0.821,
+      "step": 25420
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005889078947368421,
+      "loss": 0.8112,
+      "step": 25430
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000588828947368421,
+      "loss": 0.8002,
+      "step": 25440
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00058875,
+      "loss": 0.8111,
+      "step": 25450
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005886710526315789,
+      "loss": 0.8138,
+      "step": 25460
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005885921052631579,
+      "loss": 0.8053,
+      "step": 25470
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005885131578947368,
+      "loss": 0.795,
+      "step": 25480
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005884342105263158,
+      "loss": 0.8374,
+      "step": 25490
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005883552631578947,
+      "loss": 0.8137,
+      "step": 25500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005882763157894737,
+      "loss": 0.8247,
+      "step": 25510
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005881973684210526,
+      "loss": 0.8148,
+      "step": 25520
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005881184210526316,
+      "loss": 0.8125,
+      "step": 25530
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005880394736842104,
+      "loss": 0.8205,
+      "step": 25540
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005879605263157895,
+      "loss": 0.8077,
+      "step": 25550
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005878815789473683,
+      "loss": 0.8192,
+      "step": 25560
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005878026315789474,
+      "loss": 0.8213,
+      "step": 25570
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005877236842105263,
+      "loss": 0.8145,
+      "step": 25580
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005876447368421052,
+      "loss": 0.8032,
+      "step": 25590
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005875657894736842,
+      "loss": 0.8245,
+      "step": 25600
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000587486842105263,
+      "loss": 0.8164,
+      "step": 25610
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005874078947368421,
+      "loss": 0.7993,
+      "step": 25620
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005873289473684209,
+      "loss": 0.8149,
+      "step": 25630
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00058725,
+      "loss": 0.8116,
+      "step": 25640
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005871710526315788,
+      "loss": 0.8087,
+      "step": 25650
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005870921052631579,
+      "loss": 0.8107,
+      "step": 25660
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005870131578947368,
+      "loss": 0.807,
+      "step": 25670
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005869342105263158,
+      "loss": 0.8366,
+      "step": 25680
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005868552631578947,
+      "loss": 0.8207,
+      "step": 25690
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005867763157894736,
+      "loss": 0.8152,
+      "step": 25700
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005866973684210526,
+      "loss": 0.8138,
+      "step": 25710
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005866184210526315,
+      "loss": 0.819,
+      "step": 25720
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005865394736842105,
+      "loss": 0.8066,
+      "step": 25730
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005864605263157894,
+      "loss": 0.8157,
+      "step": 25740
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005863894736842105,
+      "loss": 0.8029,
+      "step": 25750
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005863105263157894,
+      "loss": 0.8109,
+      "step": 25760
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005862315789473683,
+      "loss": 0.825,
+      "step": 25770
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005861526315789473,
+      "loss": 0.8045,
+      "step": 25780
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005860736842105262,
+      "loss": 0.8127,
+      "step": 25790
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005859947368421052,
+      "loss": 0.8105,
+      "step": 25800
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005859157894736841,
+      "loss": 0.8017,
+      "step": 25810
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005858368421052632,
+      "loss": 0.8192,
+      "step": 25820
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000585757894736842,
+      "loss": 0.8236,
+      "step": 25830
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000585678947368421,
+      "loss": 0.8142,
+      "step": 25840
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005855999999999999,
+      "loss": 0.8061,
+      "step": 25850
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005855210526315789,
+      "loss": 0.8054,
+      "step": 25860
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005854421052631578,
+      "loss": 0.8058,
+      "step": 25870
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005853631578947368,
+      "loss": 0.8033,
+      "step": 25880
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005852842105263157,
+      "loss": 0.8201,
+      "step": 25890
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005852052631578947,
+      "loss": 0.8065,
+      "step": 25900
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005851263157894736,
+      "loss": 0.8104,
+      "step": 25910
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005850473684210526,
+      "loss": 0.7849,
+      "step": 25920
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005849684210526315,
+      "loss": 0.7978,
+      "step": 25930
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005848894736842105,
+      "loss": 0.812,
+      "step": 25940
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005848105263157894,
+      "loss": 0.8091,
+      "step": 25950
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005847315789473684,
+      "loss": 0.8145,
+      "step": 25960
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005846526315789473,
+      "loss": 0.8113,
+      "step": 25970
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005845736842105263,
+      "loss": 0.8041,
+      "step": 25980
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005844947368421052,
+      "loss": 0.8074,
+      "step": 25990
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005844157894736841,
+      "loss": 0.805,
+      "step": 26000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005843368421052631,
+      "loss": 0.8021,
+      "step": 26010
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000584257894736842,
+      "loss": 0.8067,
+      "step": 26020
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000584178947368421,
+      "loss": 0.8017,
+      "step": 26030
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005840999999999999,
+      "loss": 0.8088,
+      "step": 26040
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005840210526315789,
+      "loss": 0.8108,
+      "step": 26050
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005839421052631578,
+      "loss": 0.7941,
+      "step": 26060
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005838631578947368,
+      "loss": 0.8195,
+      "step": 26070
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005837842105263157,
+      "loss": 0.8182,
+      "step": 26080
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005837052631578947,
+      "loss": 0.8138,
+      "step": 26090
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005836263157894736,
+      "loss": 0.8126,
+      "step": 26100
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005835473684210526,
+      "loss": 0.817,
+      "step": 26110
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005834684210526315,
+      "loss": 0.81,
+      "step": 26120
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005833894736842105,
+      "loss": 0.7905,
+      "step": 26130
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005833105263157894,
+      "loss": 0.8001,
+      "step": 26140
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005832315789473684,
+      "loss": 0.8134,
+      "step": 26150
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005831526315789473,
+      "loss": 0.8116,
+      "step": 26160
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005830736842105263,
+      "loss": 0.7971,
+      "step": 26170
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005829947368421052,
+      "loss": 0.795,
+      "step": 26180
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005829157894736842,
+      "loss": 0.8128,
+      "step": 26190
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005828368421052631,
+      "loss": 0.8144,
+      "step": 26200
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005827578947368421,
+      "loss": 0.8022,
+      "step": 26210
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000582678947368421,
+      "loss": 0.8091,
+      "step": 26220
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005826,
+      "loss": 0.7954,
+      "step": 26230
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005825210526315789,
+      "loss": 0.797,
+      "step": 26240
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005824421052631579,
+      "loss": 0.8183,
+      "step": 26250
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005823631578947368,
+      "loss": 0.8078,
+      "step": 26260
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005822842105263158,
+      "loss": 0.8063,
+      "step": 26270
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005822052631578947,
+      "loss": 0.7925,
+      "step": 26280
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005821263157894736,
+      "loss": 0.7894,
+      "step": 26290
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005820473684210526,
+      "loss": 0.7882,
+      "step": 26300
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005819684210526315,
+      "loss": 0.7932,
+      "step": 26310
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005818894736842105,
+      "loss": 0.7961,
+      "step": 26320
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005818105263157894,
+      "loss": 0.7923,
+      "step": 26330
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005817315789473684,
+      "loss": 0.7974,
+      "step": 26340
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005816526315789473,
+      "loss": 0.798,
+      "step": 26350
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005815736842105263,
+      "loss": 0.792,
+      "step": 26360
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005814947368421052,
+      "loss": 0.8103,
+      "step": 26370
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005814157894736842,
+      "loss": 0.8009,
+      "step": 26380
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005813368421052631,
+      "loss": 0.8062,
+      "step": 26390
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005812578947368421,
+      "loss": 0.8107,
+      "step": 26400
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000581178947368421,
+      "loss": 0.814,
+      "step": 26410
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005811,
+      "loss": 0.8003,
+      "step": 26420
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005810210526315789,
+      "loss": 0.7893,
+      "step": 26430
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005809421052631579,
+      "loss": 0.7695,
+      "step": 26440
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005808631578947367,
+      "loss": 0.7856,
+      "step": 26450
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005807842105263158,
+      "loss": 0.7692,
+      "step": 26460
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005807052631578947,
+      "loss": 0.7626,
+      "step": 26470
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005806263157894737,
+      "loss": 0.7926,
+      "step": 26480
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005805473684210526,
+      "loss": 0.8006,
+      "step": 26490
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005804684210526316,
+      "loss": 0.802,
+      "step": 26500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005803894736842105,
+      "loss": 0.8092,
+      "step": 26510
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005803105263157894,
+      "loss": 0.8018,
+      "step": 26520
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005802315789473684,
+      "loss": 0.7875,
+      "step": 26530
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005801526315789472,
+      "loss": 0.8163,
+      "step": 26540
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005800736842105263,
+      "loss": 0.7923,
+      "step": 26550
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005799947368421053,
+      "loss": 0.8019,
+      "step": 26560
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005799157894736842,
+      "loss": 0.8018,
+      "step": 26570
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005798368421052631,
+      "loss": 0.7946,
+      "step": 26580
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005797578947368421,
+      "loss": 0.7982,
+      "step": 26590
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000579678947368421,
+      "loss": 0.7962,
+      "step": 26600
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005795999999999999,
+      "loss": 0.797,
+      "step": 26610
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005795210526315789,
+      "loss": 0.7993,
+      "step": 26620
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005794421052631578,
+      "loss": 0.7906,
+      "step": 26630
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005793631578947368,
+      "loss": 0.8111,
+      "step": 26640
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005792842105263157,
+      "loss": 0.8141,
+      "step": 26650
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005792052631578947,
+      "loss": 0.8079,
+      "step": 26660
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005791263157894737,
+      "loss": 0.7938,
+      "step": 26670
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005790473684210525,
+      "loss": 0.8108,
+      "step": 26680
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005789684210526316,
+      "loss": 0.8085,
+      "step": 26690
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005788894736842104,
+      "loss": 0.7985,
+      "step": 26700
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005788105263157895,
+      "loss": 0.7647,
+      "step": 26710
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005787315789473683,
+      "loss": 0.8048,
+      "step": 26720
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005786526315789474,
+      "loss": 0.7926,
+      "step": 26730
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005785736842105262,
+      "loss": 0.8004,
+      "step": 26740
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005785026315789473,
+      "loss": 0.7965,
+      "step": 26750
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005784236842105262,
+      "loss": 0.8007,
+      "step": 26760
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005783447368421052,
+      "loss": 0.7922,
+      "step": 26770
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005782657894736841,
+      "loss": 0.8026,
+      "step": 26780
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005781868421052631,
+      "loss": 0.8007,
+      "step": 26790
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000578107894736842,
+      "loss": 0.7939,
+      "step": 26800
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000578028947368421,
+      "loss": 0.795,
+      "step": 26810
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005779499999999999,
+      "loss": 0.7852,
+      "step": 26820
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005778710526315789,
+      "loss": 0.7874,
+      "step": 26830
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005777921052631578,
+      "loss": 0.7885,
+      "step": 26840
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005777131578947368,
+      "loss": 0.7963,
+      "step": 26850
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005776342105263157,
+      "loss": 0.795,
+      "step": 26860
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005775552631578947,
+      "loss": 0.8115,
+      "step": 26870
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005774763157894736,
+      "loss": 0.7849,
+      "step": 26880
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005773973684210526,
+      "loss": 0.7789,
+      "step": 26890
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005773184210526315,
+      "loss": 0.8073,
+      "step": 26900
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005772394736842105,
+      "loss": 0.7947,
+      "step": 26910
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005771605263157894,
+      "loss": 0.7839,
+      "step": 26920
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005770815789473684,
+      "loss": 0.8056,
+      "step": 26930
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005770026315789473,
+      "loss": 0.7972,
+      "step": 26940
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005769236842105263,
+      "loss": 0.8008,
+      "step": 26950
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005768447368421052,
+      "loss": 0.7911,
+      "step": 26960
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005767657894736842,
+      "loss": 0.8002,
+      "step": 26970
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005766868421052631,
+      "loss": 0.7812,
+      "step": 26980
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000576607894736842,
+      "loss": 0.7864,
+      "step": 26990
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000576528947368421,
+      "loss": 0.7809,
+      "step": 27000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005764499999999999,
+      "loss": 0.8083,
+      "step": 27010
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005763710526315789,
+      "loss": 0.7935,
+      "step": 27020
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005762921052631578,
+      "loss": 0.79,
+      "step": 27030
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005762131578947368,
+      "loss": 0.7868,
+      "step": 27040
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005761342105263157,
+      "loss": 0.7978,
+      "step": 27050
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005760552631578947,
+      "loss": 0.7975,
+      "step": 27060
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005759763157894736,
+      "loss": 0.7971,
+      "step": 27070
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005758973684210526,
+      "loss": 0.7861,
+      "step": 27080
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005758184210526315,
+      "loss": 0.7807,
+      "step": 27090
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005757394736842105,
+      "loss": 0.7933,
+      "step": 27100
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005756605263157894,
+      "loss": 0.8048,
+      "step": 27110
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005755815789473684,
+      "loss": 0.7791,
+      "step": 27120
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005755026315789473,
+      "loss": 0.7752,
+      "step": 27130
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005754236842105263,
+      "loss": 0.7753,
+      "step": 27140
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005753447368421052,
+      "loss": 0.7925,
+      "step": 27150
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005752657894736842,
+      "loss": 0.7894,
+      "step": 27160
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005751868421052631,
+      "loss": 0.7974,
+      "step": 27170
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005751078947368421,
+      "loss": 0.7849,
+      "step": 27180
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000575028947368421,
+      "loss": 0.8053,
+      "step": 27190
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00057495,
+      "loss": 0.8107,
+      "step": 27200
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005748710526315789,
+      "loss": 0.7912,
+      "step": 27210
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005747921052631579,
+      "loss": 0.7863,
+      "step": 27220
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005747131578947368,
+      "loss": 0.7913,
+      "step": 27230
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005746342105263158,
+      "loss": 0.7951,
+      "step": 27240
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005745552631578947,
+      "loss": 0.7872,
+      "step": 27250
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005744763157894737,
+      "loss": 0.7988,
+      "step": 27260
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005743973684210526,
+      "loss": 0.8092,
+      "step": 27270
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005743184210526315,
+      "loss": 0.7866,
+      "step": 27280
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005742394736842105,
+      "loss": 0.7957,
+      "step": 27290
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005741605263157894,
+      "loss": 0.7757,
+      "step": 27300
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005740815789473684,
+      "loss": 0.792,
+      "step": 27310
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005740026315789473,
+      "loss": 0.7942,
+      "step": 27320
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005739236842105263,
+      "loss": 0.7846,
+      "step": 27330
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005738447368421052,
+      "loss": 0.7727,
+      "step": 27340
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005737657894736842,
+      "loss": 0.7893,
+      "step": 27350
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000573686842105263,
+      "loss": 0.807,
+      "step": 27360
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005736078947368421,
+      "loss": 0.8077,
+      "step": 27370
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000573528947368421,
+      "loss": 0.7967,
+      "step": 27380
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00057345,
+      "loss": 0.7939,
+      "step": 27390
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005733710526315789,
+      "loss": 0.7984,
+      "step": 27400
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005732921052631579,
+      "loss": 0.7908,
+      "step": 27410
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005732131578947368,
+      "loss": 0.7929,
+      "step": 27420
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005731342105263157,
+      "loss": 0.7869,
+      "step": 27430
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005730552631578947,
+      "loss": 0.7889,
+      "step": 27440
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005729763157894736,
+      "loss": 0.7958,
+      "step": 27450
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005728973684210526,
+      "loss": 0.7759,
+      "step": 27460
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005728184210526316,
+      "loss": 0.7919,
+      "step": 27470
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005727394736842105,
+      "loss": 0.7901,
+      "step": 27480
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005726605263157895,
+      "loss": 0.7976,
+      "step": 27490
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005725815789473684,
+      "loss": 0.7826,
+      "step": 27500
+    },
+    {
+      "epoch": 0.28,
+      "eval_accuracy": 0.832012190127962,
+      "eval_loss": 0.82568359375,
+      "eval_runtime": 96.9614,
+      "eval_samples_per_second": 825.071,
+      "eval_steps_per_second": 1.619,
+      "step": 27500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005725026315789474,
+      "loss": 0.7844,
+      "step": 27510
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005724236842105262,
+      "loss": 0.7746,
+      "step": 27520
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005723447368421053,
+      "loss": 0.7749,
+      "step": 27530
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005722657894736841,
+      "loss": 0.7559,
+      "step": 27540
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005721868421052632,
+      "loss": 0.7876,
+      "step": 27550
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005721078947368421,
+      "loss": 0.7792,
+      "step": 27560
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000572028947368421,
+      "loss": 0.7734,
+      "step": 27570
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00057195,
+      "loss": 0.7701,
+      "step": 27580
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005718710526315788,
+      "loss": 0.7835,
+      "step": 27590
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005717921052631579,
+      "loss": 0.7688,
+      "step": 27600
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005717131578947367,
+      "loss": 0.8105,
+      "step": 27610
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005716342105263158,
+      "loss": 0.8011,
+      "step": 27620
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005715552631578946,
+      "loss": 0.8044,
+      "step": 27630
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005714763157894737,
+      "loss": 0.8107,
+      "step": 27640
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005713973684210525,
+      "loss": 0.8012,
+      "step": 27650
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005713184210526315,
+      "loss": 0.796,
+      "step": 27660
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005712394736842105,
+      "loss": 0.7865,
+      "step": 27670
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005711605263157894,
+      "loss": 0.7908,
+      "step": 27680
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005710815789473684,
+      "loss": 0.8021,
+      "step": 27690
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005710026315789473,
+      "loss": 0.7944,
+      "step": 27700
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005709236842105263,
+      "loss": 0.7863,
+      "step": 27710
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005708447368421052,
+      "loss": 0.7902,
+      "step": 27720
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005707657894736841,
+      "loss": 0.7787,
+      "step": 27730
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005706868421052631,
+      "loss": 0.7952,
+      "step": 27740
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005706157894736841,
+      "loss": 0.7887,
+      "step": 27750
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005705368421052631,
+      "loss": 0.7949,
+      "step": 27760
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000570457894736842,
+      "loss": 0.7929,
+      "step": 27770
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000570378947368421,
+      "loss": 0.802,
+      "step": 27780
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005702999999999999,
+      "loss": 0.7668,
+      "step": 27790
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005702210526315789,
+      "loss": 0.7895,
+      "step": 27800
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005701421052631578,
+      "loss": 0.7788,
+      "step": 27810
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005700631578947368,
+      "loss": 0.7903,
+      "step": 27820
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005699842105263157,
+      "loss": 0.7973,
+      "step": 27830
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005699052631578947,
+      "loss": 0.8102,
+      "step": 27840
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005698263157894736,
+      "loss": 0.7954,
+      "step": 27850
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005697473684210526,
+      "loss": 0.7891,
+      "step": 27860
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005696684210526315,
+      "loss": 0.8007,
+      "step": 27870
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005695894736842105,
+      "loss": 0.8003,
+      "step": 27880
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005695105263157894,
+      "loss": 0.7952,
+      "step": 27890
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005694315789473684,
+      "loss": 0.807,
+      "step": 27900
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005693526315789473,
+      "loss": 0.7858,
+      "step": 27910
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005692736842105263,
+      "loss": 0.7877,
+      "step": 27920
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005691947368421052,
+      "loss": 0.7896,
+      "step": 27930
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005691157894736842,
+      "loss": 0.7885,
+      "step": 27940
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005690368421052631,
+      "loss": 0.7758,
+      "step": 27950
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000568957894736842,
+      "loss": 0.8041,
+      "step": 27960
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000568878947368421,
+      "loss": 0.7975,
+      "step": 27970
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005688,
+      "loss": 0.789,
+      "step": 27980
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005687210526315789,
+      "loss": 0.7913,
+      "step": 27990
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005686421052631578,
+      "loss": 0.8035,
+      "step": 28000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005685631578947368,
+      "loss": 0.7875,
+      "step": 28010
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005684842105263157,
+      "loss": 0.7854,
+      "step": 28020
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005684052631578947,
+      "loss": 0.7866,
+      "step": 28030
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005683263157894736,
+      "loss": 0.7918,
+      "step": 28040
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005682473684210526,
+      "loss": 0.7825,
+      "step": 28050
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005681684210526315,
+      "loss": 0.7777,
+      "step": 28060
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005680894736842105,
+      "loss": 0.7947,
+      "step": 28070
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005680105263157894,
+      "loss": 0.7992,
+      "step": 28080
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005679315789473684,
+      "loss": 0.7978,
+      "step": 28090
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005678526315789473,
+      "loss": 0.8152,
+      "step": 28100
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005677736842105263,
+      "loss": 0.7904,
+      "step": 28110
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005676947368421052,
+      "loss": 0.8009,
+      "step": 28120
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005676157894736842,
+      "loss": 0.7671,
+      "step": 28130
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005675368421052631,
+      "loss": 0.7616,
+      "step": 28140
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005674578947368421,
+      "loss": 0.7662,
+      "step": 28150
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000567378947368421,
+      "loss": 0.7741,
+      "step": 28160
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005673,
+      "loss": 0.7594,
+      "step": 28170
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005672210526315789,
+      "loss": 0.7638,
+      "step": 28180
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005671421052631579,
+      "loss": 0.7669,
+      "step": 28190
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005670631578947368,
+      "loss": 0.7689,
+      "step": 28200
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005669842105263158,
+      "loss": 0.7642,
+      "step": 28210
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005669052631578947,
+      "loss": 0.7709,
+      "step": 28220
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005668263157894737,
+      "loss": 0.7563,
+      "step": 28230
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005667473684210526,
+      "loss": 0.7715,
+      "step": 28240
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005666684210526316,
+      "loss": 0.7512,
+      "step": 28250
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005665894736842105,
+      "loss": 0.7723,
+      "step": 28260
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005665105263157893,
+      "loss": 0.7626,
+      "step": 28270
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005664315789473684,
+      "loss": 0.7751,
+      "step": 28280
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005663526315789473,
+      "loss": 0.7474,
+      "step": 28290
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005662736842105263,
+      "loss": 0.7722,
+      "step": 28300
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005661947368421052,
+      "loss": 0.7533,
+      "step": 28310
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005661157894736842,
+      "loss": 0.7673,
+      "step": 28320
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005660368421052631,
+      "loss": 0.7517,
+      "step": 28330
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000565957894736842,
+      "loss": 0.7431,
+      "step": 28340
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000565878947368421,
+      "loss": 0.7501,
+      "step": 28350
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005657999999999999,
+      "loss": 0.7459,
+      "step": 28360
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005657210526315789,
+      "loss": 0.7557,
+      "step": 28370
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005656421052631579,
+      "loss": 0.7627,
+      "step": 28380
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005655631578947368,
+      "loss": 0.7589,
+      "step": 28390
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005654842105263158,
+      "loss": 0.7543,
+      "step": 28400
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005654052631578947,
+      "loss": 0.7585,
+      "step": 28410
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005653263157894737,
+      "loss": 0.7612,
+      "step": 28420
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005652473684210525,
+      "loss": 0.7489,
+      "step": 28430
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005651684210526316,
+      "loss": 0.7545,
+      "step": 28440
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005650894736842104,
+      "loss": 0.7444,
+      "step": 28450
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005650105263157895,
+      "loss": 0.751,
+      "step": 28460
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005649315789473684,
+      "loss": 0.7676,
+      "step": 28470
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005648526315789474,
+      "loss": 0.7867,
+      "step": 28480
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005647736842105263,
+      "loss": 0.7744,
+      "step": 28490
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005646947368421052,
+      "loss": 0.7803,
+      "step": 28500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005646157894736842,
+      "loss": 0.7884,
+      "step": 28510
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000564536842105263,
+      "loss": 0.7509,
+      "step": 28520
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005644578947368421,
+      "loss": 0.7463,
+      "step": 28530
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005643789473684209,
+      "loss": 0.7696,
+      "step": 28540
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005643,
+      "loss": 0.7679,
+      "step": 28550
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000564221052631579,
+      "loss": 0.7665,
+      "step": 28560
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005641421052631578,
+      "loss": 0.7845,
+      "step": 28570
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005640631578947368,
+      "loss": 0.7717,
+      "step": 28580
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005639842105263157,
+      "loss": 0.7749,
+      "step": 28590
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005639052631578947,
+      "loss": 0.7643,
+      "step": 28600
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005638263157894736,
+      "loss": 0.7721,
+      "step": 28610
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005637473684210526,
+      "loss": 0.7806,
+      "step": 28620
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005636684210526315,
+      "loss": 0.7703,
+      "step": 28630
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005635894736842104,
+      "loss": 0.775,
+      "step": 28640
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005635105263157895,
+      "loss": 0.7549,
+      "step": 28650
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005634315789473683,
+      "loss": 0.7824,
+      "step": 28660
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005633526315789474,
+      "loss": 0.777,
+      "step": 28670
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005632736842105262,
+      "loss": 0.7699,
+      "step": 28680
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005631947368421053,
+      "loss": 0.7834,
+      "step": 28690
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005631157894736841,
+      "loss": 0.7771,
+      "step": 28700
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005630368421052631,
+      "loss": 0.776,
+      "step": 28710
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000562957894736842,
+      "loss": 0.7718,
+      "step": 28720
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000562878947368421,
+      "loss": 0.7746,
+      "step": 28730
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005627999999999999,
+      "loss": 0.7664,
+      "step": 28740
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000562728947368421,
+      "loss": 0.7783,
+      "step": 28750
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005626499999999999,
+      "loss": 0.7738,
+      "step": 28760
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005625710526315789,
+      "loss": 0.7636,
+      "step": 28770
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005624921052631578,
+      "loss": 0.7785,
+      "step": 28780
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005624131578947368,
+      "loss": 0.7855,
+      "step": 28790
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005623342105263157,
+      "loss": 0.7823,
+      "step": 28800
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005622552631578947,
+      "loss": 0.7698,
+      "step": 28810
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005621763157894736,
+      "loss": 0.7731,
+      "step": 28820
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005620973684210526,
+      "loss": 0.7695,
+      "step": 28830
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005620184210526315,
+      "loss": 0.7962,
+      "step": 28840
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005619394736842105,
+      "loss": 0.7916,
+      "step": 28850
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005618605263157894,
+      "loss": 0.7721,
+      "step": 28860
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005617815789473684,
+      "loss": 0.7727,
+      "step": 28870
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005617026315789473,
+      "loss": 0.7596,
+      "step": 28880
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005616236842105263,
+      "loss": 0.779,
+      "step": 28890
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005615447368421052,
+      "loss": 0.7699,
+      "step": 28900
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005614657894736842,
+      "loss": 0.7563,
+      "step": 28910
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005613868421052631,
+      "loss": 0.7691,
+      "step": 28920
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005613078947368421,
+      "loss": 0.762,
+      "step": 28930
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000561228947368421,
+      "loss": 0.758,
+      "step": 28940
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00056115,
+      "loss": 0.7634,
+      "step": 28950
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005610710526315789,
+      "loss": 0.7748,
+      "step": 28960
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005609921052631578,
+      "loss": 0.7597,
+      "step": 28970
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005609131578947368,
+      "loss": 0.7689,
+      "step": 28980
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005608342105263157,
+      "loss": 0.7684,
+      "step": 28990
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005607552631578947,
+      "loss": 0.7795,
+      "step": 29000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005606763157894736,
+      "loss": 0.7696,
+      "step": 29010
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005605973684210526,
+      "loss": 0.7731,
+      "step": 29020
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005605184210526315,
+      "loss": 0.7873,
+      "step": 29030
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005604394736842105,
+      "loss": 0.7711,
+      "step": 29040
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005603605263157894,
+      "loss": 0.7722,
+      "step": 29050
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005602815789473684,
+      "loss": 0.7827,
+      "step": 29060
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005602026315789473,
+      "loss": 0.7647,
+      "step": 29070
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005601236842105263,
+      "loss": 0.7693,
+      "step": 29080
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005600447368421052,
+      "loss": 0.7714,
+      "step": 29090
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005599657894736842,
+      "loss": 0.777,
+      "step": 29100
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005598868421052631,
+      "loss": 0.7623,
+      "step": 29110
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005598078947368421,
+      "loss": 0.7806,
+      "step": 29120
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000559728947368421,
+      "loss": 0.7622,
+      "step": 29130
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00055965,
+      "loss": 0.7731,
+      "step": 29140
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005595710526315789,
+      "loss": 0.7647,
+      "step": 29150
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005594921052631579,
+      "loss": 0.787,
+      "step": 29160
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005594131578947368,
+      "loss": 0.7716,
+      "step": 29170
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005593342105263158,
+      "loss": 0.7641,
+      "step": 29180
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005592552631578947,
+      "loss": 0.7624,
+      "step": 29190
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005591763157894737,
+      "loss": 0.7725,
+      "step": 29200
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005590973684210526,
+      "loss": 0.7749,
+      "step": 29210
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005590184210526316,
+      "loss": 0.7672,
+      "step": 29220
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005589394736842105,
+      "loss": 0.7671,
+      "step": 29230
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005588605263157895,
+      "loss": 0.7642,
+      "step": 29240
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005587815789473684,
+      "loss": 0.7693,
+      "step": 29250
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005587026315789473,
+      "loss": 0.7706,
+      "step": 29260
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005586236842105262,
+      "loss": 0.7733,
+      "step": 29270
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005585447368421052,
+      "loss": 0.7648,
+      "step": 29280
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005584657894736842,
+      "loss": 0.7538,
+      "step": 29290
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005583868421052631,
+      "loss": 0.7698,
+      "step": 29300
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005583078947368421,
+      "loss": 0.7615,
+      "step": 29310
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000558228947368421,
+      "loss": 0.7724,
+      "step": 29320
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00055815,
+      "loss": 0.745,
+      "step": 29330
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005580710526315788,
+      "loss": 0.7809,
+      "step": 29340
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005579921052631579,
+      "loss": 0.7832,
+      "step": 29350
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005579131578947367,
+      "loss": 0.7819,
+      "step": 29360
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005578342105263158,
+      "loss": 0.7809,
+      "step": 29370
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005577552631578947,
+      "loss": 0.7701,
+      "step": 29380
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005576763157894737,
+      "loss": 0.7612,
+      "step": 29390
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005575973684210526,
+      "loss": 0.7629,
+      "step": 29400
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005575184210526315,
+      "loss": 0.7665,
+      "step": 29410
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005574394736842105,
+      "loss": 0.7607,
+      "step": 29420
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005573605263157894,
+      "loss": 0.7558,
+      "step": 29430
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005572815789473684,
+      "loss": 0.7546,
+      "step": 29440
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005572026315789473,
+      "loss": 0.7674,
+      "step": 29450
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005571236842105263,
+      "loss": 0.7651,
+      "step": 29460
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005570447368421053,
+      "loss": 0.7576,
+      "step": 29470
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005569657894736841,
+      "loss": 0.7815,
+      "step": 29480
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005568868421052632,
+      "loss": 0.7652,
+      "step": 29490
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000556807894736842,
+      "loss": 0.7596,
+      "step": 29500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005567289473684211,
+      "loss": 0.7656,
+      "step": 29510
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005566499999999999,
+      "loss": 0.7722,
+      "step": 29520
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000556571052631579,
+      "loss": 0.7622,
+      "step": 29530
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005564921052631578,
+      "loss": 0.783,
+      "step": 29540
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005564131578947367,
+      "loss": 0.756,
+      "step": 29550
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005563342105263158,
+      "loss": 0.7565,
+      "step": 29560
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005562552631578946,
+      "loss": 0.7852,
+      "step": 29570
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005561763157894737,
+      "loss": 0.764,
+      "step": 29580
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005560973684210525,
+      "loss": 0.7621,
+      "step": 29590
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005560184210526316,
+      "loss": 0.773,
+      "step": 29600
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005559394736842104,
+      "loss": 0.7667,
+      "step": 29610
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005558605263157895,
+      "loss": 0.7601,
+      "step": 29620
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005557815789473683,
+      "loss": 0.7701,
+      "step": 29630
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005557026315789473,
+      "loss": 0.7738,
+      "step": 29640
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005556236842105263,
+      "loss": 0.7708,
+      "step": 29650
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005555447368421052,
+      "loss": 0.7586,
+      "step": 29660
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005554657894736842,
+      "loss": 0.7645,
+      "step": 29670
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005553868421052631,
+      "loss": 0.7619,
+      "step": 29680
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005553078947368421,
+      "loss": 0.7617,
+      "step": 29690
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000555228947368421,
+      "loss": 0.7491,
+      "step": 29700
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005551499999999999,
+      "loss": 0.767,
+      "step": 29710
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005550710526315789,
+      "loss": 0.7461,
+      "step": 29720
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005549921052631578,
+      "loss": 0.7609,
+      "step": 29730
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005549131578947369,
+      "loss": 0.748,
+      "step": 29740
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005548342105263157,
+      "loss": 0.7701,
+      "step": 29750
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005547631578947368,
+      "loss": 0.7538,
+      "step": 29760
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005546842105263157,
+      "loss": 0.7573,
+      "step": 29770
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005546052631578947,
+      "loss": 0.7668,
+      "step": 29780
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005545263157894736,
+      "loss": 0.7655,
+      "step": 29790
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005544473684210526,
+      "loss": 0.7486,
+      "step": 29800
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005543684210526315,
+      "loss": 0.7578,
+      "step": 29810
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005542894736842105,
+      "loss": 0.7572,
+      "step": 29820
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005542105263157894,
+      "loss": 0.7479,
+      "step": 29830
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005541315789473684,
+      "loss": 0.7673,
+      "step": 29840
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005540526315789473,
+      "loss": 0.7573,
+      "step": 29850
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005539736842105263,
+      "loss": 0.7462,
+      "step": 29860
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005538947368421052,
+      "loss": 0.7546,
+      "step": 29870
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005538157894736842,
+      "loss": 0.7494,
+      "step": 29880
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005537368421052631,
+      "loss": 0.7548,
+      "step": 29890
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005536578947368421,
+      "loss": 0.7572,
+      "step": 29900
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000553578947368421,
+      "loss": 0.7564,
+      "step": 29910
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005535,
+      "loss": 0.7634,
+      "step": 29920
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005534210526315789,
+      "loss": 0.7591,
+      "step": 29930
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005533421052631579,
+      "loss": 0.7549,
+      "step": 29940
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005532631578947368,
+      "loss": 0.7569,
+      "step": 29950
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005531842105263157,
+      "loss": 0.7466,
+      "step": 29960
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005531052631578947,
+      "loss": 0.7431,
+      "step": 29970
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005530263157894736,
+      "loss": 0.7572,
+      "step": 29980
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005529473684210526,
+      "loss": 0.7676,
+      "step": 29990
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005528684210526315,
+      "loss": 0.7634,
+      "step": 30000
+    },
+    {
+      "epoch": 0.3,
+      "eval_accuracy": 0.8433683834141076,
+      "eval_loss": 0.76123046875,
+      "eval_runtime": 96.8175,
+      "eval_samples_per_second": 826.297,
+      "eval_steps_per_second": 1.622,
+      "step": 30000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005527894736842105,
+      "loss": 0.74,
+      "step": 30010
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005527105263157894,
+      "loss": 0.762,
+      "step": 30020
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005526315789473684,
+      "loss": 0.7723,
+      "step": 30030
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005525526315789473,
+      "loss": 0.7572,
+      "step": 30040
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005524736842105263,
+      "loss": 0.7527,
+      "step": 30050
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005523947368421052,
+      "loss": 0.7593,
+      "step": 30060
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005523157894736842,
+      "loss": 0.7476,
+      "step": 30070
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005522368421052631,
+      "loss": 0.7554,
+      "step": 30080
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005521578947368421,
+      "loss": 0.7492,
+      "step": 30090
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000552078947368421,
+      "loss": 0.7597,
+      "step": 30100
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000552,
+      "loss": 0.7607,
+      "step": 30110
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005519210526315789,
+      "loss": 0.7455,
+      "step": 30120
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005518421052631579,
+      "loss": 0.7545,
+      "step": 30130
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005517631578947368,
+      "loss": 0.7592,
+      "step": 30140
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005516842105263158,
+      "loss": 0.7563,
+      "step": 30150
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005516052631578947,
+      "loss": 0.7479,
+      "step": 30160
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005515263157894737,
+      "loss": 0.7569,
+      "step": 30170
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005514473684210526,
+      "loss": 0.7524,
+      "step": 30180
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005513684210526316,
+      "loss": 0.7568,
+      "step": 30190
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005512894736842105,
+      "loss": 0.7635,
+      "step": 30200
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005512105263157895,
+      "loss": 0.7526,
+      "step": 30210
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005511315789473684,
+      "loss": 0.7605,
+      "step": 30220
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005510526315789474,
+      "loss": 0.7517,
+      "step": 30230
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005509736842105263,
+      "loss": 0.7414,
+      "step": 30240
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005508947368421051,
+      "loss": 0.7578,
+      "step": 30250
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005508157894736842,
+      "loss": 0.7417,
+      "step": 30260
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000550736842105263,
+      "loss": 0.7437,
+      "step": 30270
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005506578947368421,
+      "loss": 0.7469,
+      "step": 30280
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000550578947368421,
+      "loss": 0.7589,
+      "step": 30290
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005505,
+      "loss": 0.7563,
+      "step": 30300
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005504210526315789,
+      "loss": 0.7491,
+      "step": 30310
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005503421052631578,
+      "loss": 0.7615,
+      "step": 30320
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005502631578947368,
+      "loss": 0.7601,
+      "step": 30330
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005501842105263157,
+      "loss": 0.7411,
+      "step": 30340
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005501052631578947,
+      "loss": 0.7549,
+      "step": 30350
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005500263157894736,
+      "loss": 0.7653,
+      "step": 30360
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005499473684210526,
+      "loss": 0.7534,
+      "step": 30370
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005498684210526316,
+      "loss": 0.754,
+      "step": 30380
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005497894736842104,
+      "loss": 0.7536,
+      "step": 30390
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005497105263157895,
+      "loss": 0.7557,
+      "step": 30400
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005496315789473683,
+      "loss": 0.7542,
+      "step": 30410
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005495526315789474,
+      "loss": 0.761,
+      "step": 30420
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005494736842105262,
+      "loss": 0.7696,
+      "step": 30430
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005493947368421053,
+      "loss": 0.752,
+      "step": 30440
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005493157894736841,
+      "loss": 0.7548,
+      "step": 30450
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005492368421052631,
+      "loss": 0.7614,
+      "step": 30460
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005491578947368421,
+      "loss": 0.7621,
+      "step": 30470
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000549078947368421,
+      "loss": 0.7664,
+      "step": 30480
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000549,
+      "loss": 0.7525,
+      "step": 30490
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005489210526315788,
+      "loss": 0.7453,
+      "step": 30500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005488421052631579,
+      "loss": 0.7522,
+      "step": 30510
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005487631578947367,
+      "loss": 0.7564,
+      "step": 30520
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005486842105263158,
+      "loss": 0.772,
+      "step": 30530
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005486052631578946,
+      "loss": 0.7477,
+      "step": 30540
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005485263157894736,
+      "loss": 0.7515,
+      "step": 30550
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005484473684210526,
+      "loss": 0.7445,
+      "step": 30560
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005483684210526315,
+      "loss": 0.7702,
+      "step": 30570
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005482894736842105,
+      "loss": 0.758,
+      "step": 30580
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005482105263157894,
+      "loss": 0.7574,
+      "step": 30590
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005481315789473684,
+      "loss": 0.7539,
+      "step": 30600
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005480526315789473,
+      "loss": 0.7551,
+      "step": 30610
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005479736842105262,
+      "loss": 0.7581,
+      "step": 30620
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005478947368421052,
+      "loss": 0.7581,
+      "step": 30630
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005478157894736841,
+      "loss": 0.7569,
+      "step": 30640
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005477368421052632,
+      "loss": 0.7576,
+      "step": 30650
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000547657894736842,
+      "loss": 0.7551,
+      "step": 30660
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005475789473684211,
+      "loss": 0.7675,
+      "step": 30670
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005474999999999999,
+      "loss": 0.7667,
+      "step": 30680
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005474210526315789,
+      "loss": 0.7623,
+      "step": 30690
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005473421052631578,
+      "loss": 0.7515,
+      "step": 30700
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005472631578947368,
+      "loss": 0.7656,
+      "step": 30710
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005471842105263157,
+      "loss": 0.7542,
+      "step": 30720
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005471052631578947,
+      "loss": 0.7649,
+      "step": 30730
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005470263157894737,
+      "loss": 0.7544,
+      "step": 30740
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005469473684210526,
+      "loss": 0.7633,
+      "step": 30750
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005468763157894736,
+      "loss": 0.7596,
+      "step": 30760
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005467973684210526,
+      "loss": 0.7561,
+      "step": 30770
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005467184210526315,
+      "loss": 0.7612,
+      "step": 30780
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005466394736842105,
+      "loss": 0.7541,
+      "step": 30790
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005465605263157894,
+      "loss": 0.7617,
+      "step": 30800
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005464815789473684,
+      "loss": 0.749,
+      "step": 30810
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005464026315789473,
+      "loss": 0.7634,
+      "step": 30820
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005463236842105263,
+      "loss": 0.7516,
+      "step": 30830
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005462447368421052,
+      "loss": 0.7469,
+      "step": 30840
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005461657894736842,
+      "loss": 0.7483,
+      "step": 30850
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005460868421052631,
+      "loss": 0.7579,
+      "step": 30860
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005460078947368421,
+      "loss": 0.7577,
+      "step": 30870
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000545928947368421,
+      "loss": 0.7598,
+      "step": 30880
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00054585,
+      "loss": 0.7574,
+      "step": 30890
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005457710526315789,
+      "loss": 0.7502,
+      "step": 30900
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005456921052631579,
+      "loss": 0.7667,
+      "step": 30910
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005456131578947368,
+      "loss": 0.7648,
+      "step": 30920
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005455342105263158,
+      "loss": 0.7654,
+      "step": 30930
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005454552631578947,
+      "loss": 0.7654,
+      "step": 30940
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005453763157894736,
+      "loss": 0.7554,
+      "step": 30950
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005452973684210526,
+      "loss": 0.7673,
+      "step": 30960
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005452184210526315,
+      "loss": 0.7704,
+      "step": 30970
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005451394736842105,
+      "loss": 0.7606,
+      "step": 30980
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005450605263157894,
+      "loss": 0.7612,
+      "step": 30990
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005449815789473684,
+      "loss": 0.7536,
+      "step": 31000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005449026315789473,
+      "loss": 0.7804,
+      "step": 31010
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005448236842105263,
+      "loss": 0.7685,
+      "step": 31020
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005447447368421052,
+      "loss": 0.7722,
+      "step": 31030
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005446657894736842,
+      "loss": 0.7534,
+      "step": 31040
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005445868421052631,
+      "loss": 0.7505,
+      "step": 31050
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005445078947368421,
+      "loss": 0.7667,
+      "step": 31060
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000544428947368421,
+      "loss": 0.7583,
+      "step": 31070
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00054435,
+      "loss": 0.7462,
+      "step": 31080
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005442710526315789,
+      "loss": 0.7504,
+      "step": 31090
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005441921052631579,
+      "loss": 0.7445,
+      "step": 31100
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005441131578947368,
+      "loss": 0.757,
+      "step": 31110
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005440342105263158,
+      "loss": 0.7521,
+      "step": 31120
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005439552631578947,
+      "loss": 0.7544,
+      "step": 31130
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005438763157894737,
+      "loss": 0.776,
+      "step": 31140
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005437973684210526,
+      "loss": 0.7558,
+      "step": 31150
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005437184210526315,
+      "loss": 0.7536,
+      "step": 31160
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005436394736842105,
+      "loss": 0.7672,
+      "step": 31170
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005435605263157895,
+      "loss": 0.7524,
+      "step": 31180
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005434815789473684,
+      "loss": 0.7593,
+      "step": 31190
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005434026315789474,
+      "loss": 0.7449,
+      "step": 31200
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005433236842105263,
+      "loss": 0.7637,
+      "step": 31210
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005432447368421053,
+      "loss": 0.749,
+      "step": 31220
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005431657894736841,
+      "loss": 0.7534,
+      "step": 31230
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005430868421052631,
+      "loss": 0.7537,
+      "step": 31240
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000543007894736842,
+      "loss": 0.7719,
+      "step": 31250
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000542928947368421,
+      "loss": 0.7491,
+      "step": 31260
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00054285,
+      "loss": 0.7525,
+      "step": 31270
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005427710526315789,
+      "loss": 0.7667,
+      "step": 31280
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005426921052631579,
+      "loss": 0.7592,
+      "step": 31290
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005426131578947367,
+      "loss": 0.7334,
+      "step": 31300
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005425342105263158,
+      "loss": 0.7551,
+      "step": 31310
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005424552631578946,
+      "loss": 0.7591,
+      "step": 31320
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005423763157894737,
+      "loss": 0.7398,
+      "step": 31330
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005422973684210525,
+      "loss": 0.7528,
+      "step": 31340
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005422184210526316,
+      "loss": 0.7515,
+      "step": 31350
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005421394736842104,
+      "loss": 0.7617,
+      "step": 31360
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005420605263157894,
+      "loss": 0.7424,
+      "step": 31370
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005419815789473684,
+      "loss": 0.7467,
+      "step": 31380
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005419026315789473,
+      "loss": 0.752,
+      "step": 31390
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005418236842105263,
+      "loss": 0.7533,
+      "step": 31400
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005417447368421052,
+      "loss": 0.7586,
+      "step": 31410
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005416657894736842,
+      "loss": 0.7534,
+      "step": 31420
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005415868421052631,
+      "loss": 0.7456,
+      "step": 31430
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005415078947368421,
+      "loss": 0.7484,
+      "step": 31440
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000541428947368421,
+      "loss": 0.7525,
+      "step": 31450
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005413499999999999,
+      "loss": 0.75,
+      "step": 31460
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000541271052631579,
+      "loss": 0.7527,
+      "step": 31470
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005411921052631578,
+      "loss": 0.7389,
+      "step": 31480
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005411131578947369,
+      "loss": 0.7251,
+      "step": 31490
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005410342105263157,
+      "loss": 0.7385,
+      "step": 31500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005409552631578948,
+      "loss": 0.7557,
+      "step": 31510
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005408763157894736,
+      "loss": 0.7681,
+      "step": 31520
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005407973684210525,
+      "loss": 0.752,
+      "step": 31530
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005407184210526315,
+      "loss": 0.7445,
+      "step": 31540
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005406394736842104,
+      "loss": 0.7544,
+      "step": 31550
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005405605263157895,
+      "loss": 0.7519,
+      "step": 31560
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005404815789473683,
+      "loss": 0.7554,
+      "step": 31570
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005404026315789474,
+      "loss": 0.7552,
+      "step": 31580
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005403236842105262,
+      "loss": 0.7531,
+      "step": 31590
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005402447368421052,
+      "loss": 0.7452,
+      "step": 31600
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005401657894736841,
+      "loss": 0.7409,
+      "step": 31610
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005400868421052631,
+      "loss": 0.7496,
+      "step": 31620
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000540007894736842,
+      "loss": 0.7526,
+      "step": 31630
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000539928947368421,
+      "loss": 0.7584,
+      "step": 31640
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00053985,
+      "loss": 0.7541,
+      "step": 31650
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005397710526315789,
+      "loss": 0.7505,
+      "step": 31660
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005396921052631578,
+      "loss": 0.7549,
+      "step": 31670
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005396131578947368,
+      "loss": 0.7518,
+      "step": 31680
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005395342105263157,
+      "loss": 0.7668,
+      "step": 31690
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005394552631578947,
+      "loss": 0.7565,
+      "step": 31700
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005393763157894736,
+      "loss": 0.7698,
+      "step": 31710
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005392973684210526,
+      "loss": 0.7571,
+      "step": 31720
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005392184210526315,
+      "loss": 0.7514,
+      "step": 31730
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005391394736842105,
+      "loss": 0.7492,
+      "step": 31740
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005390605263157894,
+      "loss": 0.7502,
+      "step": 31750
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005389894736842105,
+      "loss": 0.7524,
+      "step": 31760
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005389105263157894,
+      "loss": 0.7314,
+      "step": 31770
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005388315789473684,
+      "loss": 0.7403,
+      "step": 31780
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005387526315789473,
+      "loss": 0.7702,
+      "step": 31790
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005386736842105263,
+      "loss": 0.7652,
+      "step": 31800
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005385947368421052,
+      "loss": 0.7406,
+      "step": 31810
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005385157894736842,
+      "loss": 0.7541,
+      "step": 31820
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005384368421052631,
+      "loss": 0.7522,
+      "step": 31830
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005383578947368421,
+      "loss": 0.7502,
+      "step": 31840
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000538278947368421,
+      "loss": 0.7463,
+      "step": 31850
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005382,
+      "loss": 0.7564,
+      "step": 31860
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005381210526315789,
+      "loss": 0.7583,
+      "step": 31870
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005380421052631579,
+      "loss": 0.7582,
+      "step": 31880
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005379631578947368,
+      "loss": 0.7337,
+      "step": 31890
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005378842105263158,
+      "loss": 0.7439,
+      "step": 31900
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005378052631578947,
+      "loss": 0.7558,
+      "step": 31910
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005377263157894737,
+      "loss": 0.7434,
+      "step": 31920
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005376473684210526,
+      "loss": 0.7456,
+      "step": 31930
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005375684210526315,
+      "loss": 0.7559,
+      "step": 31940
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005374894736842105,
+      "loss": 0.7459,
+      "step": 31950
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005374105263157894,
+      "loss": 0.7421,
+      "step": 31960
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005373315789473684,
+      "loss": 0.7134,
+      "step": 31970
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005372526315789473,
+      "loss": 0.737,
+      "step": 31980
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005371736842105263,
+      "loss": 0.7193,
+      "step": 31990
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005370947368421052,
+      "loss": 0.7333,
+      "step": 32000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005370157894736842,
+      "loss": 0.7228,
+      "step": 32010
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005369368421052631,
+      "loss": 0.7304,
+      "step": 32020
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005368578947368421,
+      "loss": 0.7186,
+      "step": 32030
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000536778947368421,
+      "loss": 0.7269,
+      "step": 32040
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005367,
+      "loss": 0.7097,
+      "step": 32050
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005366210526315789,
+      "loss": 0.73,
+      "step": 32060
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005365421052631578,
+      "loss": 0.7304,
+      "step": 32070
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005364631578947368,
+      "loss": 0.7259,
+      "step": 32080
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005363842105263158,
+      "loss": 0.7218,
+      "step": 32090
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005363052631578947,
+      "loss": 0.7385,
+      "step": 32100
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005362263157894737,
+      "loss": 0.7354,
+      "step": 32110
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005361473684210526,
+      "loss": 0.7346,
+      "step": 32120
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005360684210526316,
+      "loss": 0.73,
+      "step": 32130
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005359894736842104,
+      "loss": 0.728,
+      "step": 32140
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005359105263157895,
+      "loss": 0.7234,
+      "step": 32150
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005358315789473683,
+      "loss": 0.7322,
+      "step": 32160
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005357526315789474,
+      "loss": 0.726,
+      "step": 32170
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005356736842105263,
+      "loss": 0.7364,
+      "step": 32180
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005355947368421053,
+      "loss": 0.7312,
+      "step": 32190
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005355157894736842,
+      "loss": 0.7171,
+      "step": 32200
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000535436842105263,
+      "loss": 0.72,
+      "step": 32210
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005353578947368421,
+      "loss": 0.7491,
+      "step": 32220
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005352789473684209,
+      "loss": 0.7367,
+      "step": 32230
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005352,
+      "loss": 0.742,
+      "step": 32240
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005351210526315788,
+      "loss": 0.7544,
+      "step": 32250
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005350421052631579,
+      "loss": 0.7528,
+      "step": 32260
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005349631578947368,
+      "loss": 0.7569,
+      "step": 32270
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005348842105263157,
+      "loss": 0.7343,
+      "step": 32280
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005348052631578947,
+      "loss": 0.7507,
+      "step": 32290
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005347263157894736,
+      "loss": 0.7515,
+      "step": 32300
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005346473684210526,
+      "loss": 0.7291,
+      "step": 32310
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005345684210526315,
+      "loss": 0.7468,
+      "step": 32320
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005344894736842105,
+      "loss": 0.7384,
+      "step": 32330
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005344105263157894,
+      "loss": 0.7359,
+      "step": 32340
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005343315789473684,
+      "loss": 0.7395,
+      "step": 32350
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005342526315789474,
+      "loss": 0.7398,
+      "step": 32360
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005341736842105262,
+      "loss": 0.7363,
+      "step": 32370
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005340947368421053,
+      "loss": 0.7574,
+      "step": 32380
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005340157894736841,
+      "loss": 0.7408,
+      "step": 32390
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005339368421052632,
+      "loss": 0.7352,
+      "step": 32400
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000533857894736842,
+      "loss": 0.7276,
+      "step": 32410
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005337789473684211,
+      "loss": 0.7376,
+      "step": 32420
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005336999999999999,
+      "loss": 0.7443,
+      "step": 32430
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005336210526315789,
+      "loss": 0.7413,
+      "step": 32440
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005335421052631578,
+      "loss": 0.7386,
+      "step": 32450
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005334631578947368,
+      "loss": 0.732,
+      "step": 32460
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005333842105263158,
+      "loss": 0.7392,
+      "step": 32470
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005333052631578946,
+      "loss": 0.7499,
+      "step": 32480
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005332263157894737,
+      "loss": 0.7445,
+      "step": 32490
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005331473684210525,
+      "loss": 0.7359,
+      "step": 32500
+    },
+    {
+      "epoch": 0.33,
+      "eval_accuracy": 0.8473949658152091,
+      "eval_loss": 0.73388671875,
+      "eval_runtime": 97.0409,
+      "eval_samples_per_second": 824.395,
+      "eval_steps_per_second": 1.618,
+      "step": 32500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005330684210526315,
+      "loss": 0.7473,
+      "step": 32510
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005329894736842104,
+      "loss": 0.7474,
+      "step": 32520
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005329105263157894,
+      "loss": 0.7436,
+      "step": 32530
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005328315789473683,
+      "loss": 0.7425,
+      "step": 32540
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005327526315789473,
+      "loss": 0.7468,
+      "step": 32550
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005326736842105263,
+      "loss": 0.7484,
+      "step": 32560
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005325947368421052,
+      "loss": 0.7538,
+      "step": 32570
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005325157894736841,
+      "loss": 0.7434,
+      "step": 32580
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005324368421052631,
+      "loss": 0.7416,
+      "step": 32590
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000532357894736842,
+      "loss": 0.7532,
+      "step": 32600
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000532278947368421,
+      "loss": 0.7541,
+      "step": 32610
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005321999999999999,
+      "loss": 0.7343,
+      "step": 32620
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005321210526315789,
+      "loss": 0.7493,
+      "step": 32630
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005320421052631578,
+      "loss": 0.7483,
+      "step": 32640
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005319631578947368,
+      "loss": 0.7399,
+      "step": 32650
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005318842105263157,
+      "loss": 0.7524,
+      "step": 32660
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005318052631578947,
+      "loss": 0.7564,
+      "step": 32670
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005317263157894736,
+      "loss": 0.7478,
+      "step": 32680
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005316473684210526,
+      "loss": 0.7527,
+      "step": 32690
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005315684210526315,
+      "loss": 0.738,
+      "step": 32700
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005314894736842105,
+      "loss": 0.7476,
+      "step": 32710
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005314105263157894,
+      "loss": 0.7487,
+      "step": 32720
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005313315789473684,
+      "loss": 0.7353,
+      "step": 32730
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005312526315789473,
+      "loss": 0.7482,
+      "step": 32740
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005311736842105263,
+      "loss": 0.751,
+      "step": 32750
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005311026315789473,
+      "loss": 0.7522,
+      "step": 32760
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005310236842105263,
+      "loss": 0.7365,
+      "step": 32770
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005309447368421052,
+      "loss": 0.7514,
+      "step": 32780
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005308657894736842,
+      "loss": 0.7545,
+      "step": 32790
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005307868421052631,
+      "loss": 0.7517,
+      "step": 32800
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005307078947368421,
+      "loss": 0.746,
+      "step": 32810
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000530628947368421,
+      "loss": 0.743,
+      "step": 32820
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00053055,
+      "loss": 0.7451,
+      "step": 32830
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005304710526315789,
+      "loss": 0.7388,
+      "step": 32840
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005303921052631579,
+      "loss": 0.7551,
+      "step": 32850
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005303131578947368,
+      "loss": 0.7417,
+      "step": 32860
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005302342105263158,
+      "loss": 0.75,
+      "step": 32870
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005301552631578947,
+      "loss": 0.7354,
+      "step": 32880
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005300763157894737,
+      "loss": 0.7264,
+      "step": 32890
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005299973684210526,
+      "loss": 0.7326,
+      "step": 32900
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005299184210526316,
+      "loss": 0.7509,
+      "step": 32910
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005298394736842105,
+      "loss": 0.7294,
+      "step": 32920
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005297605263157894,
+      "loss": 0.7486,
+      "step": 32930
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005296815789473684,
+      "loss": 0.7352,
+      "step": 32940
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005296026315789473,
+      "loss": 0.7414,
+      "step": 32950
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005295236842105263,
+      "loss": 0.7458,
+      "step": 32960
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005294447368421052,
+      "loss": 0.7415,
+      "step": 32970
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005293657894736841,
+      "loss": 0.7358,
+      "step": 32980
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005292868421052631,
+      "loss": 0.7365,
+      "step": 32990
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005292078947368421,
+      "loss": 0.7359,
+      "step": 33000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000529128947368421,
+      "loss": 0.7468,
+      "step": 33010
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00052905,
+      "loss": 0.742,
+      "step": 33020
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005289710526315789,
+      "loss": 0.7374,
+      "step": 33030
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005288921052631579,
+      "loss": 0.7457,
+      "step": 33040
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005288131578947367,
+      "loss": 0.7384,
+      "step": 33050
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005287342105263158,
+      "loss": 0.7441,
+      "step": 33060
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005286552631578946,
+      "loss": 0.747,
+      "step": 33070
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005285763157894737,
+      "loss": 0.7361,
+      "step": 33080
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005284973684210526,
+      "loss": 0.7278,
+      "step": 33090
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005284184210526316,
+      "loss": 0.7421,
+      "step": 33100
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005283394736842105,
+      "loss": 0.7444,
+      "step": 33110
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005282605263157894,
+      "loss": 0.7428,
+      "step": 33120
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005281815789473684,
+      "loss": 0.7315,
+      "step": 33130
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005281026315789473,
+      "loss": 0.7349,
+      "step": 33140
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005280236842105263,
+      "loss": 0.7473,
+      "step": 33150
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005279447368421052,
+      "loss": 0.7376,
+      "step": 33160
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005278657894736842,
+      "loss": 0.7405,
+      "step": 33170
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005277868421052632,
+      "loss": 0.7406,
+      "step": 33180
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005277078947368421,
+      "loss": 0.7355,
+      "step": 33190
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000527628947368421,
+      "loss": 0.7402,
+      "step": 33200
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005275499999999999,
+      "loss": 0.7363,
+      "step": 33210
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000527471052631579,
+      "loss": 0.7324,
+      "step": 33220
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005273921052631578,
+      "loss": 0.7352,
+      "step": 33230
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005273131578947368,
+      "loss": 0.7458,
+      "step": 33240
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005272342105263157,
+      "loss": 0.743,
+      "step": 33250
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005271552631578947,
+      "loss": 0.7454,
+      "step": 33260
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005270763157894737,
+      "loss": 0.7403,
+      "step": 33270
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005269973684210525,
+      "loss": 0.7582,
+      "step": 33280
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005269184210526316,
+      "loss": 0.7467,
+      "step": 33290
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005268394736842104,
+      "loss": 0.7244,
+      "step": 33300
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005267605263157895,
+      "loss": 0.7226,
+      "step": 33310
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005266815789473683,
+      "loss": 0.7257,
+      "step": 33320
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005266026315789474,
+      "loss": 0.7287,
+      "step": 33330
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005265236842105262,
+      "loss": 0.716,
+      "step": 33340
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005264447368421052,
+      "loss": 0.7227,
+      "step": 33350
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005263657894736842,
+      "loss": 0.746,
+      "step": 33360
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005262868421052631,
+      "loss": 0.7369,
+      "step": 33370
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005262078947368421,
+      "loss": 0.7387,
+      "step": 33380
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000526128947368421,
+      "loss": 0.7452,
+      "step": 33390
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00052605,
+      "loss": 0.7576,
+      "step": 33400
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005259710526315789,
+      "loss": 0.7297,
+      "step": 33410
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005258921052631578,
+      "loss": 0.7438,
+      "step": 33420
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005258131578947368,
+      "loss": 0.7385,
+      "step": 33430
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005257342105263157,
+      "loss": 0.737,
+      "step": 33440
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005256552631578948,
+      "loss": 0.7463,
+      "step": 33450
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005255763157894736,
+      "loss": 0.7375,
+      "step": 33460
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005254973684210527,
+      "loss": 0.7349,
+      "step": 33470
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005254184210526315,
+      "loss": 0.7489,
+      "step": 33480
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005253394736842104,
+      "loss": 0.717,
+      "step": 33490
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005252605263157894,
+      "loss": 0.7439,
+      "step": 33500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005251815789473683,
+      "loss": 0.7317,
+      "step": 33510
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005251026315789473,
+      "loss": 0.7183,
+      "step": 33520
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005250236842105262,
+      "loss": 0.7291,
+      "step": 33530
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005249447368421052,
+      "loss": 0.7308,
+      "step": 33540
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005248657894736841,
+      "loss": 0.732,
+      "step": 33550
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005247868421052632,
+      "loss": 0.7272,
+      "step": 33560
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000524707894736842,
+      "loss": 0.714,
+      "step": 33570
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000524628947368421,
+      "loss": 0.7142,
+      "step": 33580
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005245499999999999,
+      "loss": 0.7135,
+      "step": 33590
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005244710526315789,
+      "loss": 0.7322,
+      "step": 33600
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005243921052631578,
+      "loss": 0.7229,
+      "step": 33610
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005243131578947368,
+      "loss": 0.7302,
+      "step": 33620
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005242342105263157,
+      "loss": 0.7106,
+      "step": 33630
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005241552631578947,
+      "loss": 0.7233,
+      "step": 33640
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005240763157894736,
+      "loss": 0.7127,
+      "step": 33650
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005239973684210526,
+      "loss": 0.7349,
+      "step": 33660
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005239184210526315,
+      "loss": 0.7202,
+      "step": 33670
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005238394736842105,
+      "loss": 0.7272,
+      "step": 33680
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005237605263157894,
+      "loss": 0.7234,
+      "step": 33690
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005236815789473684,
+      "loss": 0.7197,
+      "step": 33700
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005236026315789473,
+      "loss": 0.7289,
+      "step": 33710
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005235236842105263,
+      "loss": 0.7314,
+      "step": 33720
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005234447368421052,
+      "loss": 0.7327,
+      "step": 33730
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005233657894736842,
+      "loss": 0.7303,
+      "step": 33740
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005232868421052631,
+      "loss": 0.7394,
+      "step": 33750
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005232157894736842,
+      "loss": 0.738,
+      "step": 33760
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005231368421052631,
+      "loss": 0.7384,
+      "step": 33770
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005230578947368421,
+      "loss": 0.7299,
+      "step": 33780
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000522978947368421,
+      "loss": 0.7335,
+      "step": 33790
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005229,
+      "loss": 0.7289,
+      "step": 33800
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005228210526315789,
+      "loss": 0.7436,
+      "step": 33810
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005227421052631579,
+      "loss": 0.7481,
+      "step": 33820
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005226631578947368,
+      "loss": 0.7415,
+      "step": 33830
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005225842105263158,
+      "loss": 0.7187,
+      "step": 33840
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005225052631578947,
+      "loss": 0.7431,
+      "step": 33850
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005224263157894737,
+      "loss": 0.7364,
+      "step": 33860
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005223473684210526,
+      "loss": 0.7437,
+      "step": 33870
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005222684210526316,
+      "loss": 0.7448,
+      "step": 33880
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005221894736842105,
+      "loss": 0.7319,
+      "step": 33890
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005221105263157895,
+      "loss": 0.7297,
+      "step": 33900
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005220315789473684,
+      "loss": 0.7319,
+      "step": 33910
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005219526315789474,
+      "loss": 0.7375,
+      "step": 33920
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005218736842105263,
+      "loss": 0.7316,
+      "step": 33930
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005217947368421052,
+      "loss": 0.7477,
+      "step": 33940
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005217157894736842,
+      "loss": 0.7337,
+      "step": 33950
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000521636842105263,
+      "loss": 0.7414,
+      "step": 33960
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005215578947368421,
+      "loss": 0.7522,
+      "step": 33970
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005214789473684209,
+      "loss": 0.7463,
+      "step": 33980
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005214,
+      "loss": 0.7378,
+      "step": 33990
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005213210526315789,
+      "loss": 0.7408,
+      "step": 34000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005212421052631579,
+      "loss": 0.7547,
+      "step": 34010
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005211631578947368,
+      "loss": 0.742,
+      "step": 34020
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005210842105263157,
+      "loss": 0.7354,
+      "step": 34030
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005210052631578947,
+      "loss": 0.7217,
+      "step": 34040
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005209263157894736,
+      "loss": 0.7308,
+      "step": 34050
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005208473684210526,
+      "loss": 0.7299,
+      "step": 34060
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005207684210526315,
+      "loss": 0.7335,
+      "step": 34070
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005206894736842105,
+      "loss": 0.7334,
+      "step": 34080
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005206105263157895,
+      "loss": 0.7333,
+      "step": 34090
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005205315789473684,
+      "loss": 0.7233,
+      "step": 34100
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005204526315789474,
+      "loss": 0.7271,
+      "step": 34110
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005203736842105262,
+      "loss": 0.7459,
+      "step": 34120
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005202947368421053,
+      "loss": 0.7334,
+      "step": 34130
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005202157894736841,
+      "loss": 0.7251,
+      "step": 34140
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005201368421052632,
+      "loss": 0.7363,
+      "step": 34150
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000520057894736842,
+      "loss": 0.7462,
+      "step": 34160
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005199789473684211,
+      "loss": 0.7539,
+      "step": 34170
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005199,
+      "loss": 0.7338,
+      "step": 34180
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005198210526315788,
+      "loss": 0.7268,
+      "step": 34190
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005197421052631579,
+      "loss": 0.7311,
+      "step": 34200
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005196631578947367,
+      "loss": 0.7339,
+      "step": 34210
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005195842105263158,
+      "loss": 0.7382,
+      "step": 34220
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005195052631578946,
+      "loss": 0.7358,
+      "step": 34230
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005194263157894737,
+      "loss": 0.7483,
+      "step": 34240
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005193473684210525,
+      "loss": 0.7288,
+      "step": 34250
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005192684210526315,
+      "loss": 0.7244,
+      "step": 34260
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005191894736842105,
+      "loss": 0.7238,
+      "step": 34270
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005191105263157894,
+      "loss": 0.7276,
+      "step": 34280
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005190315789473684,
+      "loss": 0.7254,
+      "step": 34290
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005189526315789473,
+      "loss": 0.7365,
+      "step": 34300
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005188736842105263,
+      "loss": 0.7291,
+      "step": 34310
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005187947368421052,
+      "loss": 0.7361,
+      "step": 34320
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005187157894736841,
+      "loss": 0.7284,
+      "step": 34330
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005186368421052631,
+      "loss": 0.735,
+      "step": 34340
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000518557894736842,
+      "loss": 0.7269,
+      "step": 34350
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005184789473684211,
+      "loss": 0.7336,
+      "step": 34360
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005183999999999999,
+      "loss": 0.7274,
+      "step": 34370
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000518321052631579,
+      "loss": 0.7402,
+      "step": 34380
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005182421052631578,
+      "loss": 0.7573,
+      "step": 34390
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005181631578947368,
+      "loss": 0.7416,
+      "step": 34400
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005180842105263157,
+      "loss": 0.7312,
+      "step": 34410
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005180052631578947,
+      "loss": 0.7386,
+      "step": 34420
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005179263157894736,
+      "loss": 0.7324,
+      "step": 34430
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005178473684210526,
+      "loss": 0.7301,
+      "step": 34440
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005177684210526316,
+      "loss": 0.7395,
+      "step": 34450
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005176894736842105,
+      "loss": 0.7325,
+      "step": 34460
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005176105263157895,
+      "loss": 0.7335,
+      "step": 34470
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005175315789473683,
+      "loss": 0.7278,
+      "step": 34480
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005174526315789473,
+      "loss": 0.7283,
+      "step": 34490
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005173736842105262,
+      "loss": 0.7152,
+      "step": 34500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005172947368421052,
+      "loss": 0.7211,
+      "step": 34510
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005172157894736841,
+      "loss": 0.7264,
+      "step": 34520
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005171368421052631,
+      "loss": 0.7357,
+      "step": 34530
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005170578947368421,
+      "loss": 0.724,
+      "step": 34540
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000516978947368421,
+      "loss": 0.7043,
+      "step": 34550
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005168999999999999,
+      "loss": 0.7032,
+      "step": 34560
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005168210526315789,
+      "loss": 0.7127,
+      "step": 34570
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005167421052631578,
+      "loss": 0.7052,
+      "step": 34580
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005166631578947368,
+      "loss": 0.7144,
+      "step": 34590
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005165842105263157,
+      "loss": 0.7111,
+      "step": 34600
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005165052631578947,
+      "loss": 0.7303,
+      "step": 34610
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005164263157894736,
+      "loss": 0.7421,
+      "step": 34620
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005163473684210526,
+      "loss": 0.7404,
+      "step": 34630
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005162684210526315,
+      "loss": 0.7299,
+      "step": 34640
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005161894736842105,
+      "loss": 0.7335,
+      "step": 34650
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005161105263157894,
+      "loss": 0.7416,
+      "step": 34660
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005160315789473684,
+      "loss": 0.7242,
+      "step": 34670
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005159526315789473,
+      "loss": 0.7375,
+      "step": 34680
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005158736842105263,
+      "loss": 0.7385,
+      "step": 34690
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005157947368421052,
+      "loss": 0.7338,
+      "step": 34700
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005157157894736842,
+      "loss": 0.7183,
+      "step": 34710
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005156368421052631,
+      "loss": 0.7195,
+      "step": 34720
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005155578947368421,
+      "loss": 0.729,
+      "step": 34730
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000515478947368421,
+      "loss": 0.7371,
+      "step": 34740
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005154,
+      "loss": 0.7364,
+      "step": 34750
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005153210526315789,
+      "loss": 0.7323,
+      "step": 34760
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00051525,
+      "loss": 0.7454,
+      "step": 34770
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005151710526315789,
+      "loss": 0.7339,
+      "step": 34780
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005150921052631579,
+      "loss": 0.7392,
+      "step": 34790
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005150131578947368,
+      "loss": 0.7508,
+      "step": 34800
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005149342105263158,
+      "loss": 0.7465,
+      "step": 34810
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005148552631578947,
+      "loss": 0.7408,
+      "step": 34820
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005147763157894737,
+      "loss": 0.7322,
+      "step": 34830
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005146973684210526,
+      "loss": 0.7364,
+      "step": 34840
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005146184210526316,
+      "loss": 0.7434,
+      "step": 34850
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005145394736842105,
+      "loss": 0.7299,
+      "step": 34860
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005144605263157893,
+      "loss": 0.7336,
+      "step": 34870
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005143815789473684,
+      "loss": 0.748,
+      "step": 34880
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005143026315789474,
+      "loss": 0.7367,
+      "step": 34890
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005142236842105263,
+      "loss": 0.7221,
+      "step": 34900
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005141447368421053,
+      "loss": 0.7281,
+      "step": 34910
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005140657894736842,
+      "loss": 0.7425,
+      "step": 34920
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005139868421052631,
+      "loss": 0.7206,
+      "step": 34930
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000513907894736842,
+      "loss": 0.7306,
+      "step": 34940
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000513828947368421,
+      "loss": 0.7212,
+      "step": 34950
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005137499999999999,
+      "loss": 0.7291,
+      "step": 34960
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005136710526315789,
+      "loss": 0.7288,
+      "step": 34970
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005135921052631579,
+      "loss": 0.7239,
+      "step": 34980
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005135131578947368,
+      "loss": 0.713,
+      "step": 34990
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005134342105263158,
+      "loss": 0.7263,
+      "step": 35000
+    },
+    {
+      "epoch": 0.35,
+      "eval_accuracy": 0.8494970666829337,
+      "eval_loss": 0.72412109375,
+      "eval_runtime": 96.4579,
+      "eval_samples_per_second": 829.378,
+      "eval_steps_per_second": 1.628,
+      "step": 35000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005133552631578947,
+      "loss": 0.7198,
+      "step": 35010
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005132763157894737,
+      "loss": 0.7359,
+      "step": 35020
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005132052631578946,
+      "loss": 0.7295,
+      "step": 35030
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005131263157894736,
+      "loss": 0.7355,
+      "step": 35040
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005130473684210525,
+      "loss": 0.7266,
+      "step": 35050
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005129684210526316,
+      "loss": 0.7277,
+      "step": 35060
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005128894736842104,
+      "loss": 0.7344,
+      "step": 35070
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005128105263157895,
+      "loss": 0.716,
+      "step": 35080
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005127315789473683,
+      "loss": 0.7256,
+      "step": 35090
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005126526315789473,
+      "loss": 0.7139,
+      "step": 35100
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005125736842105262,
+      "loss": 0.7295,
+      "step": 35110
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005124947368421052,
+      "loss": 0.7233,
+      "step": 35120
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005124157894736841,
+      "loss": 0.7247,
+      "step": 35130
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005123368421052631,
+      "loss": 0.7187,
+      "step": 35140
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005122578947368421,
+      "loss": 0.73,
+      "step": 35150
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000512178947368421,
+      "loss": 0.7323,
+      "step": 35160
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005120999999999999,
+      "loss": 0.725,
+      "step": 35170
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005120210526315789,
+      "loss": 0.7245,
+      "step": 35180
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005119421052631578,
+      "loss": 0.7183,
+      "step": 35190
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005118631578947368,
+      "loss": 0.7255,
+      "step": 35200
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005117842105263157,
+      "loss": 0.7156,
+      "step": 35210
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005117052631578947,
+      "loss": 0.7228,
+      "step": 35220
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005116263157894736,
+      "loss": 0.7315,
+      "step": 35230
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005115473684210526,
+      "loss": 0.7284,
+      "step": 35240
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005114684210526315,
+      "loss": 0.7303,
+      "step": 35250
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005113894736842105,
+      "loss": 0.7175,
+      "step": 35260
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005113105263157894,
+      "loss": 0.7213,
+      "step": 35270
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005112315789473684,
+      "loss": 0.7116,
+      "step": 35280
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005111526315789473,
+      "loss": 0.7201,
+      "step": 35290
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005110736842105263,
+      "loss": 0.7214,
+      "step": 35300
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005109947368421052,
+      "loss": 0.7244,
+      "step": 35310
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005109157894736841,
+      "loss": 0.7074,
+      "step": 35320
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005108368421052631,
+      "loss": 0.7161,
+      "step": 35330
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000510757894736842,
+      "loss": 0.7144,
+      "step": 35340
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000510678947368421,
+      "loss": 0.7091,
+      "step": 35350
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005105999999999999,
+      "loss": 0.7014,
+      "step": 35360
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005105210526315789,
+      "loss": 0.7294,
+      "step": 35370
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005104421052631578,
+      "loss": 0.7478,
+      "step": 35380
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005103631578947368,
+      "loss": 0.7266,
+      "step": 35390
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005102842105263157,
+      "loss": 0.7205,
+      "step": 35400
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005102052631578947,
+      "loss": 0.7177,
+      "step": 35410
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005101263157894736,
+      "loss": 0.7212,
+      "step": 35420
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005100473684210526,
+      "loss": 0.7242,
+      "step": 35430
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005099684210526315,
+      "loss": 0.723,
+      "step": 35440
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005098894736842105,
+      "loss": 0.7216,
+      "step": 35450
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005098105263157894,
+      "loss": 0.7143,
+      "step": 35460
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005097315789473684,
+      "loss": 0.7294,
+      "step": 35470
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005096526315789473,
+      "loss": 0.7084,
+      "step": 35480
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005095736842105263,
+      "loss": 0.7262,
+      "step": 35490
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005094947368421052,
+      "loss": 0.7139,
+      "step": 35500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005094157894736842,
+      "loss": 0.7152,
+      "step": 35510
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005093368421052631,
+      "loss": 0.7184,
+      "step": 35520
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005092578947368421,
+      "loss": 0.7286,
+      "step": 35530
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000509178947368421,
+      "loss": 0.7231,
+      "step": 35540
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005091,
+      "loss": 0.7326,
+      "step": 35550
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005090210526315789,
+      "loss": 0.7314,
+      "step": 35560
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005089421052631579,
+      "loss": 0.7175,
+      "step": 35570
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005088631578947368,
+      "loss": 0.7295,
+      "step": 35580
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005087842105263158,
+      "loss": 0.7197,
+      "step": 35590
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005087052631578947,
+      "loss": 0.7279,
+      "step": 35600
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005086263157894737,
+      "loss": 0.7315,
+      "step": 35610
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005085473684210526,
+      "loss": 0.7144,
+      "step": 35620
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005084684210526315,
+      "loss": 0.7167,
+      "step": 35630
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005083894736842105,
+      "loss": 0.7193,
+      "step": 35640
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005083105263157894,
+      "loss": 0.728,
+      "step": 35650
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005082315789473684,
+      "loss": 0.7328,
+      "step": 35660
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005081526315789473,
+      "loss": 0.7275,
+      "step": 35670
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005080736842105263,
+      "loss": 0.7131,
+      "step": 35680
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005079947368421052,
+      "loss": 0.7281,
+      "step": 35690
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005079157894736842,
+      "loss": 0.739,
+      "step": 35700
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005078368421052631,
+      "loss": 0.7187,
+      "step": 35710
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005077578947368421,
+      "loss": 0.7069,
+      "step": 35720
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000507678947368421,
+      "loss": 0.709,
+      "step": 35730
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005076,
+      "loss": 0.7044,
+      "step": 35740
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005075210526315789,
+      "loss": 0.7058,
+      "step": 35750
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005074421052631579,
+      "loss": 0.7104,
+      "step": 35760
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005073631578947368,
+      "loss": 0.7264,
+      "step": 35770
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005072842105263157,
+      "loss": 0.7214,
+      "step": 35780
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005072052631578947,
+      "loss": 0.7252,
+      "step": 35790
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005071263157894737,
+      "loss": 0.7139,
+      "step": 35800
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005070473684210526,
+      "loss": 0.7245,
+      "step": 35810
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005069684210526316,
+      "loss": 0.7257,
+      "step": 35820
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005068894736842105,
+      "loss": 0.732,
+      "step": 35830
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005068105263157895,
+      "loss": 0.7318,
+      "step": 35840
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005067315789473683,
+      "loss": 0.7307,
+      "step": 35850
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005066526315789474,
+      "loss": 0.7198,
+      "step": 35860
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005065736842105262,
+      "loss": 0.7259,
+      "step": 35870
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005064947368421053,
+      "loss": 0.7207,
+      "step": 35880
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005064157894736842,
+      "loss": 0.7214,
+      "step": 35890
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005063368421052632,
+      "loss": 0.716,
+      "step": 35900
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005062578947368421,
+      "loss": 0.7169,
+      "step": 35910
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000506178947368421,
+      "loss": 0.7143,
+      "step": 35920
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005061,
+      "loss": 0.7174,
+      "step": 35930
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005060210526315788,
+      "loss": 0.7246,
+      "step": 35940
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005059421052631579,
+      "loss": 0.7279,
+      "step": 35950
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005058631578947367,
+      "loss": 0.7095,
+      "step": 35960
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005057842105263158,
+      "loss": 0.7167,
+      "step": 35970
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005057052631578947,
+      "loss": 0.7285,
+      "step": 35980
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005056263157894737,
+      "loss": 0.7328,
+      "step": 35990
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005055473684210526,
+      "loss": 0.7097,
+      "step": 36000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005054684210526315,
+      "loss": 0.7217,
+      "step": 36010
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005053894736842105,
+      "loss": 0.7156,
+      "step": 36020
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005053105263157894,
+      "loss": 0.7147,
+      "step": 36030
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005052315789473684,
+      "loss": 0.7142,
+      "step": 36040
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005051526315789473,
+      "loss": 0.7222,
+      "step": 36050
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005050736842105263,
+      "loss": 0.7305,
+      "step": 36060
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005049947368421053,
+      "loss": 0.7179,
+      "step": 36070
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005049157894736841,
+      "loss": 0.7175,
+      "step": 36080
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005048368421052632,
+      "loss": 0.7293,
+      "step": 36090
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000504757894736842,
+      "loss": 0.718,
+      "step": 36100
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005046789473684211,
+      "loss": 0.7118,
+      "step": 36110
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005045999999999999,
+      "loss": 0.7249,
+      "step": 36120
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000504521052631579,
+      "loss": 0.718,
+      "step": 36130
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005044421052631578,
+      "loss": 0.7196,
+      "step": 36140
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005043631578947368,
+      "loss": 0.722,
+      "step": 36150
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005042842105263157,
+      "loss": 0.7193,
+      "step": 36160
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005042052631578946,
+      "loss": 0.7274,
+      "step": 36170
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005041263157894737,
+      "loss": 0.7426,
+      "step": 36180
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005040473684210525,
+      "loss": 0.7385,
+      "step": 36190
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005039684210526316,
+      "loss": 0.7281,
+      "step": 36200
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005038894736842104,
+      "loss": 0.7254,
+      "step": 36210
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005038105263157894,
+      "loss": 0.7201,
+      "step": 36220
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005037315789473683,
+      "loss": 0.7271,
+      "step": 36230
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005036526315789473,
+      "loss": 0.712,
+      "step": 36240
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005035736842105262,
+      "loss": 0.7131,
+      "step": 36250
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005034947368421052,
+      "loss": 0.7257,
+      "step": 36260
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005034157894736842,
+      "loss": 0.7171,
+      "step": 36270
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005033368421052631,
+      "loss": 0.7261,
+      "step": 36280
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005032578947368421,
+      "loss": 0.7113,
+      "step": 36290
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000503178947368421,
+      "loss": 0.7236,
+      "step": 36300
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005030999999999999,
+      "loss": 0.7233,
+      "step": 36310
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005030210526315789,
+      "loss": 0.7248,
+      "step": 36320
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005029421052631578,
+      "loss": 0.7149,
+      "step": 36330
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005028631578947368,
+      "loss": 0.7202,
+      "step": 36340
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005027842105263157,
+      "loss": 0.7303,
+      "step": 36350
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005027052631578948,
+      "loss": 0.7318,
+      "step": 36360
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005026263157894736,
+      "loss": 0.7295,
+      "step": 36370
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005025473684210526,
+      "loss": 0.7153,
+      "step": 36380
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005024684210526315,
+      "loss": 0.7242,
+      "step": 36390
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005023894736842105,
+      "loss": 0.7082,
+      "step": 36400
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005023105263157894,
+      "loss": 0.7298,
+      "step": 36410
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005022315789473684,
+      "loss": 0.7359,
+      "step": 36420
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005021526315789473,
+      "loss": 0.7232,
+      "step": 36430
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005020736842105263,
+      "loss": 0.7138,
+      "step": 36440
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005019947368421052,
+      "loss": 0.7274,
+      "step": 36450
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005019157894736841,
+      "loss": 0.7177,
+      "step": 36460
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005018368421052631,
+      "loss": 0.7288,
+      "step": 36470
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000501757894736842,
+      "loss": 0.7133,
+      "step": 36480
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000501678947368421,
+      "loss": 0.7078,
+      "step": 36490
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005015999999999999,
+      "loss": 0.7188,
+      "step": 36500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0005015210526315789,
+      "loss": 0.7166,
+      "step": 36510
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0005014421052631578,
+      "loss": 0.7206,
+      "step": 36520
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0005013631578947368,
+      "loss": 0.7089,
+      "step": 36530
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0005012842105263157,
+      "loss": 0.7096,
+      "step": 36540
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0005012052631578947,
+      "loss": 0.7141,
+      "step": 36550
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0005011263157894736,
+      "loss": 0.7171,
+      "step": 36560
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0005010473684210526,
+      "loss": 0.7153,
+      "step": 36570
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0005009684210526315,
+      "loss": 0.7071,
+      "step": 36580
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0005008894736842105,
+      "loss": 0.7147,
+      "step": 36590
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0005008105263157894,
+      "loss": 0.7072,
+      "step": 36600
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0005007315789473684,
+      "loss": 0.7211,
+      "step": 36610
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0005006526315789473,
+      "loss": 0.7277,
+      "step": 36620
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0005005736842105263,
+      "loss": 0.721,
+      "step": 36630
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0005004947368421052,
+      "loss": 0.713,
+      "step": 36640
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0005004157894736842,
+      "loss": 0.7116,
+      "step": 36650
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0005003368421052631,
+      "loss": 0.7118,
+      "step": 36660
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0005002578947368421,
+      "loss": 0.7117,
+      "step": 36670
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000500178947368421,
+      "loss": 0.7096,
+      "step": 36680
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0005001,
+      "loss": 0.7168,
+      "step": 36690
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0005000210526315789,
+      "loss": 0.7151,
+      "step": 36700
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004999421052631579,
+      "loss": 0.699,
+      "step": 36710
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004998631578947368,
+      "loss": 0.7075,
+      "step": 36720
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004997842105263158,
+      "loss": 0.7075,
+      "step": 36730
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004997052631578947,
+      "loss": 0.7169,
+      "step": 36740
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004996263157894736,
+      "loss": 0.7234,
+      "step": 36750
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004995473684210526,
+      "loss": 0.7185,
+      "step": 36760
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004994684210526315,
+      "loss": 0.7203,
+      "step": 36770
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004993894736842105,
+      "loss": 0.7247,
+      "step": 36780
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004993105263157894,
+      "loss": 0.7275,
+      "step": 36790
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004992315789473684,
+      "loss": 0.7052,
+      "step": 36800
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004991526315789473,
+      "loss": 0.7142,
+      "step": 36810
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004990736842105263,
+      "loss": 0.7164,
+      "step": 36820
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004989947368421052,
+      "loss": 0.7231,
+      "step": 36830
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004989157894736842,
+      "loss": 0.7199,
+      "step": 36840
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004988368421052631,
+      "loss": 0.7085,
+      "step": 36850
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004987578947368421,
+      "loss": 0.7109,
+      "step": 36860
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000498678947368421,
+      "loss": 0.7202,
+      "step": 36870
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004986,
+      "loss": 0.7232,
+      "step": 36880
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004985210526315789,
+      "loss": 0.7137,
+      "step": 36890
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004984421052631579,
+      "loss": 0.721,
+      "step": 36900
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004983631578947368,
+      "loss": 0.7191,
+      "step": 36910
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004982842105263158,
+      "loss": 0.7214,
+      "step": 36920
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004982052631578947,
+      "loss": 0.7221,
+      "step": 36930
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004981263157894737,
+      "loss": 0.7255,
+      "step": 36940
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004980473684210526,
+      "loss": 0.7214,
+      "step": 36950
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004979684210526316,
+      "loss": 0.7181,
+      "step": 36960
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004978894736842105,
+      "loss": 0.7225,
+      "step": 36970
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004978105263157895,
+      "loss": 0.7284,
+      "step": 36980
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004977315789473683,
+      "loss": 0.7201,
+      "step": 36990
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004976526315789474,
+      "loss": 0.727,
+      "step": 37000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004975736842105263,
+      "loss": 0.7146,
+      "step": 37010
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004974947368421053,
+      "loss": 0.7201,
+      "step": 37020
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004974236842105262,
+      "loss": 0.7216,
+      "step": 37030
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004973447368421053,
+      "loss": 0.7165,
+      "step": 37040
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004972657894736841,
+      "loss": 0.7176,
+      "step": 37050
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004971868421052631,
+      "loss": 0.7259,
+      "step": 37060
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004971078947368421,
+      "loss": 0.7075,
+      "step": 37070
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000497028947368421,
+      "loss": 0.7061,
+      "step": 37080
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00049695,
+      "loss": 0.7284,
+      "step": 37090
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004968710526315789,
+      "loss": 0.7247,
+      "step": 37100
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004967921052631579,
+      "loss": 0.7131,
+      "step": 37110
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004967131578947368,
+      "loss": 0.7106,
+      "step": 37120
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004966342105263158,
+      "loss": 0.7251,
+      "step": 37130
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004965552631578947,
+      "loss": 0.7151,
+      "step": 37140
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004964763157894736,
+      "loss": 0.7147,
+      "step": 37150
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004963973684210525,
+      "loss": 0.7136,
+      "step": 37160
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004963184210526315,
+      "loss": 0.6948,
+      "step": 37170
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004962394736842106,
+      "loss": 0.7092,
+      "step": 37180
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004961605263157894,
+      "loss": 0.7282,
+      "step": 37190
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004960815789473684,
+      "loss": 0.7155,
+      "step": 37200
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004960026315789473,
+      "loss": 0.7101,
+      "step": 37210
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004959236842105262,
+      "loss": 0.7093,
+      "step": 37220
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004958447368421052,
+      "loss": 0.695,
+      "step": 37230
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004957657894736841,
+      "loss": 0.7034,
+      "step": 37240
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004956868421052631,
+      "loss": 0.6962,
+      "step": 37250
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000495607894736842,
+      "loss": 0.697,
+      "step": 37260
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004955289473684211,
+      "loss": 0.6937,
+      "step": 37270
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004954499999999999,
+      "loss": 0.7066,
+      "step": 37280
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004953710526315789,
+      "loss": 0.6947,
+      "step": 37290
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004952921052631578,
+      "loss": 0.6934,
+      "step": 37300
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004952131578947368,
+      "loss": 0.7065,
+      "step": 37310
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004951342105263157,
+      "loss": 0.6958,
+      "step": 37320
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004950552631578947,
+      "loss": 0.6939,
+      "step": 37330
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004949763157894736,
+      "loss": 0.7098,
+      "step": 37340
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004948973684210526,
+      "loss": 0.6964,
+      "step": 37350
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004948184210526315,
+      "loss": 0.7114,
+      "step": 37360
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004947394736842105,
+      "loss": 0.699,
+      "step": 37370
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004946605263157894,
+      "loss": 0.7035,
+      "step": 37380
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004945815789473684,
+      "loss": 0.7026,
+      "step": 37390
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004945026315789473,
+      "loss": 0.6915,
+      "step": 37400
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004944236842105263,
+      "loss": 0.7018,
+      "step": 37410
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004943447368421052,
+      "loss": 0.6961,
+      "step": 37420
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004942657894736842,
+      "loss": 0.7012,
+      "step": 37430
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004941868421052631,
+      "loss": 0.6963,
+      "step": 37440
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000494107894736842,
+      "loss": 0.6939,
+      "step": 37450
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000494028947368421,
+      "loss": 0.7096,
+      "step": 37460
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004939499999999999,
+      "loss": 0.7065,
+      "step": 37470
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004938710526315789,
+      "loss": 0.7188,
+      "step": 37480
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004937921052631578,
+      "loss": 0.7202,
+      "step": 37490
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004937131578947368,
+      "loss": 0.719,
+      "step": 37500
+    },
+    {
+      "epoch": 0.38,
+      "eval_accuracy": 0.8522274619295023,
+      "eval_loss": 0.70361328125,
+      "eval_runtime": 97.0969,
+      "eval_samples_per_second": 823.919,
+      "eval_steps_per_second": 1.617,
+      "step": 37500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004936342105263157,
+      "loss": 0.7156,
+      "step": 37510
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004935552631578947,
+      "loss": 0.7196,
+      "step": 37520
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004934763157894736,
+      "loss": 0.715,
+      "step": 37530
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004933973684210526,
+      "loss": 0.7078,
+      "step": 37540
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004933184210526315,
+      "loss": 0.7027,
+      "step": 37550
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004932394736842105,
+      "loss": 0.7141,
+      "step": 37560
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004931605263157894,
+      "loss": 0.7137,
+      "step": 37570
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004930815789473684,
+      "loss": 0.7156,
+      "step": 37580
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004930026315789473,
+      "loss": 0.7046,
+      "step": 37590
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004929236842105263,
+      "loss": 0.7017,
+      "step": 37600
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004928447368421052,
+      "loss": 0.7096,
+      "step": 37610
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004927657894736842,
+      "loss": 0.7229,
+      "step": 37620
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004926868421052631,
+      "loss": 0.7186,
+      "step": 37630
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004926078947368421,
+      "loss": 0.7103,
+      "step": 37640
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000492528947368421,
+      "loss": 0.7069,
+      "step": 37650
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00049245,
+      "loss": 0.7328,
+      "step": 37660
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004923710526315789,
+      "loss": 0.7067,
+      "step": 37670
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004922921052631579,
+      "loss": 0.7214,
+      "step": 37680
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004922131578947368,
+      "loss": 0.7246,
+      "step": 37690
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004921342105263158,
+      "loss": 0.72,
+      "step": 37700
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004920552631578947,
+      "loss": 0.7203,
+      "step": 37710
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004919763157894737,
+      "loss": 0.7188,
+      "step": 37720
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004918973684210526,
+      "loss": 0.7234,
+      "step": 37730
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004918184210526315,
+      "loss": 0.7165,
+      "step": 37740
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004917394736842105,
+      "loss": 0.7195,
+      "step": 37750
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004916605263157894,
+      "loss": 0.7046,
+      "step": 37760
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004915815789473684,
+      "loss": 0.7145,
+      "step": 37770
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004915026315789473,
+      "loss": 0.7107,
+      "step": 37780
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004914236842105263,
+      "loss": 0.7143,
+      "step": 37790
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004913447368421052,
+      "loss": 0.7283,
+      "step": 37800
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004912657894736842,
+      "loss": 0.7108,
+      "step": 37810
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004911868421052631,
+      "loss": 0.699,
+      "step": 37820
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004911078947368421,
+      "loss": 0.7078,
+      "step": 37830
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000491028947368421,
+      "loss": 0.7112,
+      "step": 37840
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00049095,
+      "loss": 0.713,
+      "step": 37850
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004908710526315789,
+      "loss": 0.7074,
+      "step": 37860
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004907921052631579,
+      "loss": 0.7112,
+      "step": 37870
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004907131578947368,
+      "loss": 0.7033,
+      "step": 37880
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004906342105263158,
+      "loss": 0.7322,
+      "step": 37890
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004905552631578946,
+      "loss": 0.7089,
+      "step": 37900
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004904763157894737,
+      "loss": 0.7081,
+      "step": 37910
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004903973684210526,
+      "loss": 0.7105,
+      "step": 37920
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004903184210526316,
+      "loss": 0.7295,
+      "step": 37930
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004902394736842105,
+      "loss": 0.6959,
+      "step": 37940
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004901605263157895,
+      "loss": 0.7188,
+      "step": 37950
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004900815789473684,
+      "loss": 0.7233,
+      "step": 37960
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004900026315789473,
+      "loss": 0.7068,
+      "step": 37970
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004899236842105263,
+      "loss": 0.7028,
+      "step": 37980
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004898447368421051,
+      "loss": 0.7088,
+      "step": 37990
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004897657894736842,
+      "loss": 0.7146,
+      "step": 38000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004896868421052632,
+      "loss": 0.7026,
+      "step": 38010
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004896078947368421,
+      "loss": 0.6985,
+      "step": 38020
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000489528947368421,
+      "loss": 0.7207,
+      "step": 38030
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004894578947368421,
+      "loss": 0.7109,
+      "step": 38040
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000489378947368421,
+      "loss": 0.7072,
+      "step": 38050
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004892999999999999,
+      "loss": 0.7085,
+      "step": 38060
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000489221052631579,
+      "loss": 0.6987,
+      "step": 38070
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004891421052631578,
+      "loss": 0.6996,
+      "step": 38080
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004890631578947369,
+      "loss": 0.6952,
+      "step": 38090
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004889842105263157,
+      "loss": 0.7141,
+      "step": 38100
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004889052631578948,
+      "loss": 0.6997,
+      "step": 38110
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004888263157894736,
+      "loss": 0.6953,
+      "step": 38120
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004887473684210526,
+      "loss": 0.7028,
+      "step": 38130
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004886684210526315,
+      "loss": 0.7287,
+      "step": 38140
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004885894736842104,
+      "loss": 0.7033,
+      "step": 38150
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004885105263157895,
+      "loss": 0.6955,
+      "step": 38160
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004884315789473683,
+      "loss": 0.7003,
+      "step": 38170
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004883526315789474,
+      "loss": 0.6893,
+      "step": 38180
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00048827368421052624,
+      "loss": 0.7037,
+      "step": 38190
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00048819473684210524,
+      "loss": 0.723,
+      "step": 38200
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00048811578947368414,
+      "loss": 0.6862,
+      "step": 38210
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00048803684210526314,
+      "loss": 0.7082,
+      "step": 38220
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00048795789473684203,
+      "loss": 0.7009,
+      "step": 38230
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00048787894736842104,
+      "loss": 0.7058,
+      "step": 38240
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00048779999999999993,
+      "loss": 0.7091,
+      "step": 38250
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004877210526315789,
+      "loss": 0.7238,
+      "step": 38260
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004876421052631579,
+      "loss": 0.696,
+      "step": 38270
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004875631578947368,
+      "loss": 0.7026,
+      "step": 38280
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004874842105263158,
+      "loss": 0.7143,
+      "step": 38290
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00048740526315789467,
+      "loss": 0.7052,
+      "step": 38300
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004873263157894737,
+      "loss": 0.7143,
+      "step": 38310
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00048724736842105257,
+      "loss": 0.7069,
+      "step": 38320
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004871684210526315,
+      "loss": 0.71,
+      "step": 38330
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00048708947368421047,
+      "loss": 0.695,
+      "step": 38340
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004870105263157894,
+      "loss": 0.7147,
+      "step": 38350
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004869315789473684,
+      "loss": 0.6984,
+      "step": 38360
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004868526315789473,
+      "loss": 0.6981,
+      "step": 38370
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004867736842105263,
+      "loss": 0.708,
+      "step": 38380
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004866947368421052,
+      "loss": 0.6915,
+      "step": 38390
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00048661578947368416,
+      "loss": 0.6967,
+      "step": 38400
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004865368421052631,
+      "loss": 0.7076,
+      "step": 38410
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00048645789473684205,
+      "loss": 0.7114,
+      "step": 38420
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000486378947368421,
+      "loss": 0.7053,
+      "step": 38430
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00048629999999999995,
+      "loss": 0.6946,
+      "step": 38440
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00048622105263157895,
+      "loss": 0.6925,
+      "step": 38450
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00048614210526315785,
+      "loss": 0.6931,
+      "step": 38460
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00048606315789473685,
+      "loss": 0.7136,
+      "step": 38470
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00048598421052631574,
+      "loss": 0.7081,
+      "step": 38480
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004859052631578947,
+      "loss": 0.6948,
+      "step": 38490
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00048582631578947364,
+      "loss": 0.6915,
+      "step": 38500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004857473684210526,
+      "loss": 0.6917,
+      "step": 38510
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004856684210526315,
+      "loss": 0.6973,
+      "step": 38520
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004855894736842105,
+      "loss": 0.7004,
+      "step": 38530
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004855105263157895,
+      "loss": 0.6954,
+      "step": 38540
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004854315789473684,
+      "loss": 0.6859,
+      "step": 38550
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00048535263157894733,
+      "loss": 0.6932,
+      "step": 38560
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004852736842105263,
+      "loss": 0.6987,
+      "step": 38570
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004851947368421052,
+      "loss": 0.7112,
+      "step": 38580
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004851157894736842,
+      "loss": 0.6718,
+      "step": 38590
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004850368421052631,
+      "loss": 0.7012,
+      "step": 38600
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000484957894736842,
+      "loss": 0.7003,
+      "step": 38610
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000484878947368421,
+      "loss": 0.6851,
+      "step": 38620
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00048479999999999997,
+      "loss": 0.6972,
+      "step": 38630
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004847210526315789,
+      "loss": 0.7078,
+      "step": 38640
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00048464210526315786,
+      "loss": 0.7038,
+      "step": 38650
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004845631578947368,
+      "loss": 0.6966,
+      "step": 38660
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00048448421052631576,
+      "loss": 0.7014,
+      "step": 38670
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00048440526315789465,
+      "loss": 0.7019,
+      "step": 38680
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00048432631578947366,
+      "loss": 0.7057,
+      "step": 38690
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00048424736842105255,
+      "loss": 0.6955,
+      "step": 38700
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00048416842105263155,
+      "loss": 0.6929,
+      "step": 38710
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004840894736842105,
+      "loss": 0.696,
+      "step": 38720
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00048401052631578945,
+      "loss": 0.6963,
+      "step": 38730
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004839315789473684,
+      "loss": 0.6944,
+      "step": 38740
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004838526315789473,
+      "loss": 0.7043,
+      "step": 38750
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004837736842105263,
+      "loss": 0.6921,
+      "step": 38760
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004836947368421052,
+      "loss": 0.6853,
+      "step": 38770
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004836157894736842,
+      "loss": 0.6994,
+      "step": 38780
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004835368421052631,
+      "loss": 0.6953,
+      "step": 38790
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004834578947368421,
+      "loss": 0.6904,
+      "step": 38800
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00048337894736842104,
+      "loss": 0.6889,
+      "step": 38810
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00048329999999999993,
+      "loss": 0.6786,
+      "step": 38820
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00048322105263157893,
+      "loss": 0.6897,
+      "step": 38830
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004831421052631578,
+      "loss": 0.6931,
+      "step": 38840
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00048306315789473683,
+      "loss": 0.6884,
+      "step": 38850
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004829842105263157,
+      "loss": 0.6975,
+      "step": 38860
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004829052631578947,
+      "loss": 0.7129,
+      "step": 38870
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004828263157894736,
+      "loss": 0.7159,
+      "step": 38880
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00048274736842105257,
+      "loss": 0.6956,
+      "step": 38890
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004826684210526315,
+      "loss": 0.6964,
+      "step": 38900
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00048258947368421046,
+      "loss": 0.7002,
+      "step": 38910
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00048251052631578947,
+      "loss": 0.6982,
+      "step": 38920
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00048243157894736836,
+      "loss": 0.7088,
+      "step": 38930
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00048235263157894736,
+      "loss": 0.7087,
+      "step": 38940
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00048227368421052626,
+      "loss": 0.7199,
+      "step": 38950
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00048219473684210526,
+      "loss": 0.7218,
+      "step": 38960
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00048211578947368415,
+      "loss": 0.6993,
+      "step": 38970
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004820368421052631,
+      "loss": 0.6952,
+      "step": 38980
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00048195789473684205,
+      "loss": 0.6993,
+      "step": 38990
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000481878947368421,
+      "loss": 0.6955,
+      "step": 39000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004818,
+      "loss": 0.6888,
+      "step": 39010
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004817210526315789,
+      "loss": 0.6949,
+      "step": 39020
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004816421052631579,
+      "loss": 0.689,
+      "step": 39030
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004815710526315789,
+      "loss": 0.7001,
+      "step": 39040
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004814921052631578,
+      "loss": 0.6921,
+      "step": 39050
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004814131578947368,
+      "loss": 0.6863,
+      "step": 39060
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00048133421052631576,
+      "loss": 0.6886,
+      "step": 39070
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004812552631578947,
+      "loss": 0.6878,
+      "step": 39080
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00048117631578947366,
+      "loss": 0.6889,
+      "step": 39090
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00048109736842105255,
+      "loss": 0.6952,
+      "step": 39100
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00048101842105263156,
+      "loss": 0.7014,
+      "step": 39110
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00048093947368421045,
+      "loss": 0.687,
+      "step": 39120
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00048086052631578945,
+      "loss": 0.6916,
+      "step": 39130
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00048078157894736835,
+      "loss": 0.6913,
+      "step": 39140
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00048070263157894735,
+      "loss": 0.6941,
+      "step": 39150
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004806236842105263,
+      "loss": 0.6897,
+      "step": 39160
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004805447368421052,
+      "loss": 0.6934,
+      "step": 39170
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004804657894736842,
+      "loss": 0.6852,
+      "step": 39180
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004803868421052631,
+      "loss": 0.6965,
+      "step": 39190
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004803078947368421,
+      "loss": 0.6887,
+      "step": 39200
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000480228947368421,
+      "loss": 0.6916,
+      "step": 39210
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00048015,
+      "loss": 0.6985,
+      "step": 39220
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004800710526315789,
+      "loss": 0.6974,
+      "step": 39230
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00047999210526315783,
+      "loss": 0.6817,
+      "step": 39240
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00047991315789473683,
+      "loss": 0.6742,
+      "step": 39250
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00047983421052631573,
+      "loss": 0.681,
+      "step": 39260
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00047975526315789473,
+      "loss": 0.6815,
+      "step": 39270
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004796763157894736,
+      "loss": 0.674,
+      "step": 39280
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004795973684210526,
+      "loss": 0.6891,
+      "step": 39290
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004795184210526315,
+      "loss": 0.6903,
+      "step": 39300
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00047943947368421047,
+      "loss": 0.6884,
+      "step": 39310
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004793605263157894,
+      "loss": 0.7042,
+      "step": 39320
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00047928157894736837,
+      "loss": 0.6875,
+      "step": 39330
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004792026315789473,
+      "loss": 0.6812,
+      "step": 39340
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00047912368421052626,
+      "loss": 0.6878,
+      "step": 39350
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00047904473684210526,
+      "loss": 0.6839,
+      "step": 39360
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00047896578947368416,
+      "loss": 0.7026,
+      "step": 39370
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00047888684210526316,
+      "loss": 0.6946,
+      "step": 39380
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00047880789473684206,
+      "loss": 0.7005,
+      "step": 39390
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000478728947368421,
+      "loss": 0.7001,
+      "step": 39400
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00047864999999999995,
+      "loss": 0.7012,
+      "step": 39410
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004785710526315789,
+      "loss": 0.7053,
+      "step": 39420
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004784921052631578,
+      "loss": 0.7048,
+      "step": 39430
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004784131578947368,
+      "loss": 0.7034,
+      "step": 39440
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004783342105263158,
+      "loss": 0.7107,
+      "step": 39450
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004782552631578947,
+      "loss": 0.7069,
+      "step": 39460
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00047817631578947364,
+      "loss": 0.6973,
+      "step": 39470
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004780973684210526,
+      "loss": 0.7031,
+      "step": 39480
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00047801842105263154,
+      "loss": 0.7056,
+      "step": 39490
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004779394736842105,
+      "loss": 0.691,
+      "step": 39500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00047786052631578944,
+      "loss": 0.6971,
+      "step": 39510
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00047778157894736833,
+      "loss": 0.699,
+      "step": 39520
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00047770263157894733,
+      "loss": 0.7114,
+      "step": 39530
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004776236842105263,
+      "loss": 0.7069,
+      "step": 39540
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00047754473684210523,
+      "loss": 0.6815,
+      "step": 39550
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004774657894736842,
+      "loss": 0.7017,
+      "step": 39560
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004773868421052631,
+      "loss": 0.7047,
+      "step": 39570
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004773078947368421,
+      "loss": 0.6874,
+      "step": 39580
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00047722894736842097,
+      "loss": 0.7093,
+      "step": 39590
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00047714999999999997,
+      "loss": 0.695,
+      "step": 39600
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00047707105263157886,
+      "loss": 0.7006,
+      "step": 39610
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00047699210526315787,
+      "loss": 0.713,
+      "step": 39620
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004769131578947368,
+      "loss": 0.7064,
+      "step": 39630
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00047683421052631576,
+      "loss": 0.7092,
+      "step": 39640
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004767552631578947,
+      "loss": 0.708,
+      "step": 39650
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004766763157894736,
+      "loss": 0.6989,
+      "step": 39660
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004765973684210526,
+      "loss": 0.7091,
+      "step": 39670
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004765184210526315,
+      "loss": 0.6891,
+      "step": 39680
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004764394736842105,
+      "loss": 0.701,
+      "step": 39690
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004763605263157894,
+      "loss": 0.71,
+      "step": 39700
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004762815789473684,
+      "loss": 0.6988,
+      "step": 39710
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00047620263157894735,
+      "loss": 0.6897,
+      "step": 39720
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00047612368421052624,
+      "loss": 0.6949,
+      "step": 39730
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00047604473684210525,
+      "loss": 0.6938,
+      "step": 39740
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00047596578947368414,
+      "loss": 0.6925,
+      "step": 39750
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00047588684210526314,
+      "loss": 0.705,
+      "step": 39760
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00047580789473684204,
+      "loss": 0.6964,
+      "step": 39770
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00047572894736842104,
+      "loss": 0.6825,
+      "step": 39780
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00047564999999999993,
+      "loss": 0.6854,
+      "step": 39790
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004755710526315789,
+      "loss": 0.7001,
+      "step": 39800
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004754921052631579,
+      "loss": 0.6817,
+      "step": 39810
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004754131578947368,
+      "loss": 0.6928,
+      "step": 39820
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004753342105263158,
+      "loss": 0.6887,
+      "step": 39830
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004752552631578947,
+      "loss": 0.6972,
+      "step": 39840
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004751763157894737,
+      "loss": 0.6946,
+      "step": 39850
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00047509736842105257,
+      "loss": 0.6966,
+      "step": 39860
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004750184210526316,
+      "loss": 0.7063,
+      "step": 39870
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00047493947368421047,
+      "loss": 0.6945,
+      "step": 39880
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004748605263157894,
+      "loss": 0.6974,
+      "step": 39890
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00047478157894736836,
+      "loss": 0.6954,
+      "step": 39900
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004747026315789473,
+      "loss": 0.6986,
+      "step": 39910
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004746236842105263,
+      "loss": 0.7061,
+      "step": 39920
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004745447368421052,
+      "loss": 0.6927,
+      "step": 39930
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004744657894736842,
+      "loss": 0.6977,
+      "step": 39940
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004743868421052631,
+      "loss": 0.6994,
+      "step": 39950
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00047430789473684205,
+      "loss": 0.6969,
+      "step": 39960
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000474228947368421,
+      "loss": 0.707,
+      "step": 39970
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00047414999999999995,
+      "loss": 0.6954,
+      "step": 39980
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004740710526315789,
+      "loss": 0.7007,
+      "step": 39990
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00047399210526315785,
+      "loss": 0.6871,
+      "step": 40000
+    },
+    {
+      "epoch": 0.4,
+      "eval_accuracy": 0.8555457696762774,
+      "eval_loss": 0.6865234375,
+      "eval_runtime": 97.6539,
+      "eval_samples_per_second": 819.22,
+      "eval_steps_per_second": 1.608,
+      "step": 40000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00047391315789473685,
+      "loss": 0.6924,
+      "step": 40010
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00047383421052631574,
+      "loss": 0.6849,
+      "step": 40020
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004737552631578947,
+      "loss": 0.6881,
+      "step": 40030
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00047368421052631577,
+      "loss": 0.6896,
+      "step": 40040
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00047360526315789466,
+      "loss": 0.6836,
+      "step": 40050
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00047352631578947366,
+      "loss": 0.6836,
+      "step": 40060
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004734473684210526,
+      "loss": 0.6805,
+      "step": 40070
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004733684210526315,
+      "loss": 0.6913,
+      "step": 40080
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004732894736842105,
+      "loss": 0.6908,
+      "step": 40090
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004732105263157894,
+      "loss": 0.6953,
+      "step": 40100
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004731315789473684,
+      "loss": 0.6889,
+      "step": 40110
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004730526315789473,
+      "loss": 0.6899,
+      "step": 40120
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004729736842105263,
+      "loss": 0.7003,
+      "step": 40130
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004728947368421052,
+      "loss": 0.6891,
+      "step": 40140
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00047281578947368415,
+      "loss": 0.6883,
+      "step": 40150
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00047273684210526315,
+      "loss": 0.6824,
+      "step": 40160
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00047265789473684204,
+      "loss": 0.6874,
+      "step": 40170
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00047257894736842104,
+      "loss": 0.6972,
+      "step": 40180
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00047249999999999994,
+      "loss": 0.6928,
+      "step": 40190
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00047242105263157894,
+      "loss": 0.6907,
+      "step": 40200
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00047234210526315783,
+      "loss": 0.6792,
+      "step": 40210
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004722631578947368,
+      "loss": 0.702,
+      "step": 40220
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00047218421052631573,
+      "loss": 0.6839,
+      "step": 40230
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004721052631578947,
+      "loss": 0.6884,
+      "step": 40240
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004720263157894737,
+      "loss": 0.6862,
+      "step": 40250
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004719473684210526,
+      "loss": 0.6907,
+      "step": 40260
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004718684210526316,
+      "loss": 0.6965,
+      "step": 40270
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004717894736842105,
+      "loss": 0.6958,
+      "step": 40280
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004717105263157895,
+      "loss": 0.6888,
+      "step": 40290
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00047163157894736837,
+      "loss": 0.682,
+      "step": 40300
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004715526315789473,
+      "loss": 0.6933,
+      "step": 40310
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00047147368421052627,
+      "loss": 0.6881,
+      "step": 40320
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004713947368421052,
+      "loss": 0.6933,
+      "step": 40330
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004713157894736842,
+      "loss": 0.6976,
+      "step": 40340
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004712368421052631,
+      "loss": 0.6975,
+      "step": 40350
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004711578947368421,
+      "loss": 0.6913,
+      "step": 40360
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000471078947368421,
+      "loss": 0.6992,
+      "step": 40370
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00047099999999999996,
+      "loss": 0.6916,
+      "step": 40380
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004709210526315789,
+      "loss": 0.6828,
+      "step": 40390
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00047084210526315785,
+      "loss": 0.6956,
+      "step": 40400
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004707631578947368,
+      "loss": 0.6947,
+      "step": 40410
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00047068421052631575,
+      "loss": 0.6874,
+      "step": 40420
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00047060526315789464,
+      "loss": 0.6968,
+      "step": 40430
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00047052631578947365,
+      "loss": 0.7027,
+      "step": 40440
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004704473684210526,
+      "loss": 0.7011,
+      "step": 40450
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00047036842105263154,
+      "loss": 0.7095,
+      "step": 40460
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004702894736842105,
+      "loss": 0.694,
+      "step": 40470
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00047021052631578944,
+      "loss": 0.6841,
+      "step": 40480
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004701315789473684,
+      "loss": 0.6947,
+      "step": 40490
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004700526315789473,
+      "loss": 0.6903,
+      "step": 40500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004699736842105263,
+      "loss": 0.6838,
+      "step": 40510
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004698947368421052,
+      "loss": 0.6863,
+      "step": 40520
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004698157894736842,
+      "loss": 0.6855,
+      "step": 40530
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00046973684210526313,
+      "loss": 0.6874,
+      "step": 40540
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004696578947368421,
+      "loss": 0.7044,
+      "step": 40550
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000469578947368421,
+      "loss": 0.6961,
+      "step": 40560
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004694999999999999,
+      "loss": 0.695,
+      "step": 40570
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004694210526315789,
+      "loss": 0.6925,
+      "step": 40580
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004693421052631578,
+      "loss": 0.6871,
+      "step": 40590
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004692631578947368,
+      "loss": 0.6961,
+      "step": 40600
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004691842105263157,
+      "loss": 0.6917,
+      "step": 40610
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004691052631578947,
+      "loss": 0.7046,
+      "step": 40620
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00046902631578947366,
+      "loss": 0.6964,
+      "step": 40630
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00046894736842105256,
+      "loss": 0.7025,
+      "step": 40640
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00046886842105263156,
+      "loss": 0.697,
+      "step": 40650
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00046878947368421045,
+      "loss": 0.7004,
+      "step": 40660
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00046871052631578946,
+      "loss": 0.6905,
+      "step": 40670
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00046863157894736835,
+      "loss": 0.6974,
+      "step": 40680
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00046855263157894735,
+      "loss": 0.6892,
+      "step": 40690
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00046847368421052625,
+      "loss": 0.703,
+      "step": 40700
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004683947368421052,
+      "loss": 0.6959,
+      "step": 40710
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004683157894736842,
+      "loss": 0.7007,
+      "step": 40720
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004682368421052631,
+      "loss": 0.7028,
+      "step": 40730
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004681578947368421,
+      "loss": 0.6976,
+      "step": 40740
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000468078947368421,
+      "loss": 0.6759,
+      "step": 40750
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000468,
+      "loss": 0.687,
+      "step": 40760
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004679210526315789,
+      "loss": 0.691,
+      "step": 40770
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004678421052631579,
+      "loss": 0.7051,
+      "step": 40780
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004677631578947368,
+      "loss": 0.6868,
+      "step": 40790
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00046768421052631573,
+      "loss": 0.6897,
+      "step": 40800
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00046760526315789473,
+      "loss": 0.6961,
+      "step": 40810
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00046752631578947363,
+      "loss": 0.6947,
+      "step": 40820
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00046744736842105263,
+      "loss": 0.6867,
+      "step": 40830
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004673684210526315,
+      "loss": 0.7068,
+      "step": 40840
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004672894736842105,
+      "loss": 0.6843,
+      "step": 40850
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004672105263157894,
+      "loss": 0.6921,
+      "step": 40860
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00046713157894736837,
+      "loss": 0.6965,
+      "step": 40870
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004670526315789473,
+      "loss": 0.6805,
+      "step": 40880
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00046697368421052627,
+      "loss": 0.6899,
+      "step": 40890
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00046689473684210527,
+      "loss": 0.7008,
+      "step": 40900
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00046681578947368416,
+      "loss": 0.6849,
+      "step": 40910
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00046673684210526316,
+      "loss": 0.6882,
+      "step": 40920
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00046665789473684206,
+      "loss": 0.6899,
+      "step": 40930
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000466578947368421,
+      "loss": 0.6999,
+      "step": 40940
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00046649999999999996,
+      "loss": 0.6869,
+      "step": 40950
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004664210526315789,
+      "loss": 0.6805,
+      "step": 40960
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00046634210526315785,
+      "loss": 0.7013,
+      "step": 40970
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004662631578947368,
+      "loss": 0.7048,
+      "step": 40980
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004661842105263157,
+      "loss": 0.6954,
+      "step": 40990
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004661052631578947,
+      "loss": 0.6995,
+      "step": 41000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00046602631578947364,
+      "loss": 0.6981,
+      "step": 41010
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004659473684210526,
+      "loss": 0.6892,
+      "step": 41020
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00046586842105263154,
+      "loss": 0.7067,
+      "step": 41030
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004657973684210526,
+      "loss": 0.7051,
+      "step": 41040
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004657184210526315,
+      "loss": 0.6936,
+      "step": 41050
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00046563947368421046,
+      "loss": 0.6929,
+      "step": 41060
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00046556052631578946,
+      "loss": 0.6958,
+      "step": 41070
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00046548157894736836,
+      "loss": 0.689,
+      "step": 41080
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00046540263157894736,
+      "loss": 0.687,
+      "step": 41090
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00046532368421052625,
+      "loss": 0.6948,
+      "step": 41100
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00046524473684210525,
+      "loss": 0.6971,
+      "step": 41110
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00046516578947368415,
+      "loss": 0.7012,
+      "step": 41120
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00046508684210526315,
+      "loss": 0.6793,
+      "step": 41130
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00046500789473684205,
+      "loss": 0.6915,
+      "step": 41140
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000464928947368421,
+      "loss": 0.6676,
+      "step": 41150
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00046485,
+      "loss": 0.6703,
+      "step": 41160
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004647710526315789,
+      "loss": 0.6636,
+      "step": 41170
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004646921052631579,
+      "loss": 0.6597,
+      "step": 41180
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004646131578947368,
+      "loss": 0.6886,
+      "step": 41190
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004645342105263158,
+      "loss": 0.6793,
+      "step": 41200
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004644552631578947,
+      "loss": 0.69,
+      "step": 41210
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00046437631578947363,
+      "loss": 0.6747,
+      "step": 41220
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004642973684210526,
+      "loss": 0.686,
+      "step": 41230
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00046421842105263153,
+      "loss": 0.6928,
+      "step": 41240
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00046413947368421053,
+      "loss": 0.7013,
+      "step": 41250
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004640605263157894,
+      "loss": 0.695,
+      "step": 41260
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00046398157894736843,
+      "loss": 0.6906,
+      "step": 41270
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004639026315789473,
+      "loss": 0.7013,
+      "step": 41280
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00046382368421052627,
+      "loss": 0.6922,
+      "step": 41290
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004637447368421052,
+      "loss": 0.7055,
+      "step": 41300
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00046366578947368417,
+      "loss": 0.7107,
+      "step": 41310
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004635868421052631,
+      "loss": 0.6923,
+      "step": 41320
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00046350789473684206,
+      "loss": 0.7092,
+      "step": 41330
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00046342894736842107,
+      "loss": 0.6952,
+      "step": 41340
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00046334999999999996,
+      "loss": 0.6947,
+      "step": 41350
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004632710526315789,
+      "loss": 0.682,
+      "step": 41360
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00046319210526315786,
+      "loss": 0.6881,
+      "step": 41370
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004631131578947368,
+      "loss": 0.6981,
+      "step": 41380
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00046303421052631575,
+      "loss": 0.6984,
+      "step": 41390
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004629552631578947,
+      "loss": 0.6828,
+      "step": 41400
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004628763157894736,
+      "loss": 0.6994,
+      "step": 41410
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004627973684210526,
+      "loss": 0.7045,
+      "step": 41420
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00046271842105263155,
+      "loss": 0.6925,
+      "step": 41430
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004626394736842105,
+      "loss": 0.6868,
+      "step": 41440
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00046256052631578944,
+      "loss": 0.6992,
+      "step": 41450
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004624815789473684,
+      "loss": 0.7059,
+      "step": 41460
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00046240263157894734,
+      "loss": 0.7006,
+      "step": 41470
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00046232368421052623,
+      "loss": 0.7029,
+      "step": 41480
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00046224473684210524,
+      "loss": 0.6921,
+      "step": 41490
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00046216578947368413,
+      "loss": 0.6974,
+      "step": 41500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00046208684210526313,
+      "loss": 0.7081,
+      "step": 41510
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000462007894736842,
+      "loss": 0.6967,
+      "step": 41520
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00046192894736842103,
+      "loss": 0.6858,
+      "step": 41530
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00046185,
+      "loss": 0.6899,
+      "step": 41540
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00046177105263157887,
+      "loss": 0.7043,
+      "step": 41550
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0004616921052631579,
+      "loss": 0.7068,
+      "step": 41560
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00046161315789473677,
+      "loss": 0.6884,
+      "step": 41570
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00046153421052631577,
+      "loss": 0.6911,
+      "step": 41580
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00046145526315789466,
+      "loss": 0.6755,
+      "step": 41590
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00046137631578947367,
+      "loss": 0.6848,
+      "step": 41600
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00046129736842105256,
+      "loss": 0.6875,
+      "step": 41610
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00046121842105263156,
+      "loss": 0.6835,
+      "step": 41620
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0004611394736842105,
+      "loss": 0.6804,
+      "step": 41630
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0004610605263157894,
+      "loss": 0.7054,
+      "step": 41640
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0004609815789473684,
+      "loss": 0.7074,
+      "step": 41650
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0004609026315789473,
+      "loss": 0.7067,
+      "step": 41660
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0004608236842105263,
+      "loss": 0.689,
+      "step": 41670
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0004607447368421052,
+      "loss": 0.6833,
+      "step": 41680
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0004606657894736842,
+      "loss": 0.694,
+      "step": 41690
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0004605868421052631,
+      "loss": 0.6868,
+      "step": 41700
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00046050789473684204,
+      "loss": 0.6911,
+      "step": 41710
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00046042894736842105,
+      "loss": 0.6926,
+      "step": 41720
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00046034999999999994,
+      "loss": 0.7007,
+      "step": 41730
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00046027105263157894,
+      "loss": 0.6976,
+      "step": 41740
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00046019210526315784,
+      "loss": 0.6939,
+      "step": 41750
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00046011315789473684,
+      "loss": 0.6798,
+      "step": 41760
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00046003421052631573,
+      "loss": 0.6686,
+      "step": 41770
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0004599552631578947,
+      "loss": 0.6657,
+      "step": 41780
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00045987631578947363,
+      "loss": 0.6722,
+      "step": 41790
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0004597973684210526,
+      "loss": 0.667,
+      "step": 41800
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0004597184210526316,
+      "loss": 0.702,
+      "step": 41810
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0004596394736842105,
+      "loss": 0.6961,
+      "step": 41820
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0004595605263157895,
+      "loss": 0.7083,
+      "step": 41830
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00045948157894736837,
+      "loss": 0.7035,
+      "step": 41840
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0004594026315789473,
+      "loss": 0.6927,
+      "step": 41850
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00045932368421052627,
+      "loss": 0.7025,
+      "step": 41860
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0004592447368421052,
+      "loss": 0.7172,
+      "step": 41870
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00045916578947368417,
+      "loss": 0.689,
+      "step": 41880
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0004590868421052631,
+      "loss": 0.7104,
+      "step": 41890
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0004590078947368421,
+      "loss": 0.6763,
+      "step": 41900
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000458928947368421,
+      "loss": 0.6951,
+      "step": 41910
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00045884999999999996,
+      "loss": 0.6861,
+      "step": 41920
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0004587710526315789,
+      "loss": 0.6988,
+      "step": 41930
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00045869210526315786,
+      "loss": 0.705,
+      "step": 41940
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0004586131578947368,
+      "loss": 0.6976,
+      "step": 41950
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00045853421052631575,
+      "loss": 0.6956,
+      "step": 41960
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00045845526315789465,
+      "loss": 0.7138,
+      "step": 41970
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00045837631578947365,
+      "loss": 0.6899,
+      "step": 41980
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00045829736842105265,
+      "loss": 0.6924,
+      "step": 41990
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00045821842105263155,
+      "loss": 0.6994,
+      "step": 42000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0004581394736842105,
+      "loss": 0.6916,
+      "step": 42010
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00045806052631578944,
+      "loss": 0.6946,
+      "step": 42020
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0004579815789473684,
+      "loss": 0.6987,
+      "step": 42030
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00045791052631578947,
+      "loss": 0.6817,
+      "step": 42040
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00045783157894736836,
+      "loss": 0.6914,
+      "step": 42050
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0004577526315789473,
+      "loss": 0.6875,
+      "step": 42060
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0004576736842105263,
+      "loss": 0.6995,
+      "step": 42070
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0004575947368421052,
+      "loss": 0.6953,
+      "step": 42080
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0004575157894736842,
+      "loss": 0.6829,
+      "step": 42090
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0004574368421052631,
+      "loss": 0.6764,
+      "step": 42100
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0004573578947368421,
+      "loss": 0.6876,
+      "step": 42110
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000457278947368421,
+      "loss": 0.705,
+      "step": 42120
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00045719999999999995,
+      "loss": 0.6856,
+      "step": 42130
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0004571210526315789,
+      "loss": 0.6913,
+      "step": 42140
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00045704210526315784,
+      "loss": 0.6964,
+      "step": 42150
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00045696315789473684,
+      "loss": 0.6934,
+      "step": 42160
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00045688421052631574,
+      "loss": 0.6899,
+      "step": 42170
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00045680526315789474,
+      "loss": 0.693,
+      "step": 42180
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00045672631578947364,
+      "loss": 0.684,
+      "step": 42190
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0004566473684210526,
+      "loss": 0.6944,
+      "step": 42200
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00045656842105263153,
+      "loss": 0.6775,
+      "step": 42210
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0004564894736842105,
+      "loss": 0.6865,
+      "step": 42220
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00045641052631578943,
+      "loss": 0.6876,
+      "step": 42230
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0004563315789473684,
+      "loss": 0.6855,
+      "step": 42240
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0004562526315789474,
+      "loss": 0.6881,
+      "step": 42250
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0004561736842105263,
+      "loss": 0.6939,
+      "step": 42260
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0004560947368421052,
+      "loss": 0.6921,
+      "step": 42270
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00045601578947368417,
+      "loss": 0.6904,
+      "step": 42280
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0004559368421052631,
+      "loss": 0.6913,
+      "step": 42290
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00045585789473684207,
+      "loss": 0.6889,
+      "step": 42300
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000455778947368421,
+      "loss": 0.675,
+      "step": 42310
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0004556999999999999,
+      "loss": 0.6869,
+      "step": 42320
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0004556210526315789,
+      "loss": 0.6844,
+      "step": 42330
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00045554210526315786,
+      "loss": 0.6827,
+      "step": 42340
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0004554631578947368,
+      "loss": 0.6916,
+      "step": 42350
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00045538421052631576,
+      "loss": 0.6821,
+      "step": 42360
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0004553052631578947,
+      "loss": 0.6998,
+      "step": 42370
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00045522631578947365,
+      "loss": 0.6859,
+      "step": 42380
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00045514736842105255,
+      "loss": 0.6947,
+      "step": 42390
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00045506842105263155,
+      "loss": 0.685,
+      "step": 42400
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00045498947368421044,
+      "loss": 0.6817,
+      "step": 42410
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00045491052631578945,
+      "loss": 0.6796,
+      "step": 42420
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0004548315789473684,
+      "loss": 0.6856,
+      "step": 42430
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00045475263157894734,
+      "loss": 0.699,
+      "step": 42440
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0004546736842105263,
+      "loss": 0.6971,
+      "step": 42450
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0004545947368421052,
+      "loss": 0.6875,
+      "step": 42460
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0004545157894736842,
+      "loss": 0.6906,
+      "step": 42470
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0004544368421052631,
+      "loss": 0.6809,
+      "step": 42480
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0004543578947368421,
+      "loss": 0.7047,
+      "step": 42490
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000454278947368421,
+      "loss": 0.6838,
+      "step": 42500
+    },
+    {
+      "epoch": 0.42,
+      "eval_accuracy": 0.8575258035447281,
+      "eval_loss": 0.677734375,
+      "eval_runtime": 96.7246,
+      "eval_samples_per_second": 827.09,
+      "eval_steps_per_second": 1.623,
+      "step": 42500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004542,
+      "loss": 0.6802,
+      "step": 42510
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004541210526315789,
+      "loss": 0.6877,
+      "step": 42520
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004540421052631579,
+      "loss": 0.6861,
+      "step": 42530
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004539631578947368,
+      "loss": 0.6852,
+      "step": 42540
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004538842105263157,
+      "loss": 0.6799,
+      "step": 42550
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004538052631578947,
+      "loss": 0.6797,
+      "step": 42560
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004537263157894736,
+      "loss": 0.6812,
+      "step": 42570
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004536473684210526,
+      "loss": 0.6601,
+      "step": 42580
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004535684210526315,
+      "loss": 0.675,
+      "step": 42590
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004534894736842105,
+      "loss": 0.6841,
+      "step": 42600
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004534105263157894,
+      "loss": 0.6829,
+      "step": 42610
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00045333157894736836,
+      "loss": 0.6653,
+      "step": 42620
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00045325263157894736,
+      "loss": 0.6829,
+      "step": 42630
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00045317368421052626,
+      "loss": 0.6849,
+      "step": 42640
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00045309473684210526,
+      "loss": 0.6735,
+      "step": 42650
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00045301578947368415,
+      "loss": 0.6897,
+      "step": 42660
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00045293684210526315,
+      "loss": 0.6926,
+      "step": 42670
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00045285789473684205,
+      "loss": 0.6934,
+      "step": 42680
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000452778947368421,
+      "loss": 0.6712,
+      "step": 42690
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00045269999999999994,
+      "loss": 0.6946,
+      "step": 42700
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004526210526315789,
+      "loss": 0.673,
+      "step": 42710
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004525421052631579,
+      "loss": 0.6845,
+      "step": 42720
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004524631578947368,
+      "loss": 0.6793,
+      "step": 42730
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004523842105263158,
+      "loss": 0.6748,
+      "step": 42740
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004523052631578947,
+      "loss": 0.6857,
+      "step": 42750
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00045222631578947363,
+      "loss": 0.6747,
+      "step": 42760
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004521473684210526,
+      "loss": 0.6798,
+      "step": 42770
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00045206842105263153,
+      "loss": 0.6589,
+      "step": 42780
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004519894736842105,
+      "loss": 0.6727,
+      "step": 42790
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00045191052631578943,
+      "loss": 0.6831,
+      "step": 42800
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00045183157894736843,
+      "loss": 0.6783,
+      "step": 42810
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004517526315789473,
+      "loss": 0.6891,
+      "step": 42820
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00045167368421052627,
+      "loss": 0.6791,
+      "step": 42830
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004515947368421052,
+      "loss": 0.674,
+      "step": 42840
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00045151578947368417,
+      "loss": 0.6715,
+      "step": 42850
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004514368421052631,
+      "loss": 0.6802,
+      "step": 42860
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00045135789473684207,
+      "loss": 0.6719,
+      "step": 42870
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00045127894736842096,
+      "loss": 0.6785,
+      "step": 42880
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00045119999999999996,
+      "loss": 0.6824,
+      "step": 42890
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00045112105263157897,
+      "loss": 0.6907,
+      "step": 42900
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00045104210526315786,
+      "loss": 0.6926,
+      "step": 42910
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004509631578947368,
+      "loss": 0.6813,
+      "step": 42920
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00045088421052631576,
+      "loss": 0.6878,
+      "step": 42930
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004508052631578947,
+      "loss": 0.6756,
+      "step": 42940
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004507263157894736,
+      "loss": 0.6811,
+      "step": 42950
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004506473684210526,
+      "loss": 0.6798,
+      "step": 42960
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004505684210526315,
+      "loss": 0.6847,
+      "step": 42970
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004504894736842105,
+      "loss": 0.6842,
+      "step": 42980
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00045041052631578945,
+      "loss": 0.6751,
+      "step": 42990
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004503315789473684,
+      "loss": 0.6784,
+      "step": 43000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00045025263157894734,
+      "loss": 0.6883,
+      "step": 43010
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004501736842105263,
+      "loss": 0.676,
+      "step": 43020
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00045009473684210524,
+      "loss": 0.6871,
+      "step": 43030
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00045001578947368413,
+      "loss": 0.683,
+      "step": 43040
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004499447368421052,
+      "loss": 0.6733,
+      "step": 43050
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00044986578947368416,
+      "loss": 0.684,
+      "step": 43060
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00044978684210526316,
+      "loss": 0.6839,
+      "step": 43070
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00044970789473684205,
+      "loss": 0.6992,
+      "step": 43080
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00044962894736842106,
+      "loss": 0.6915,
+      "step": 43090
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00044954999999999995,
+      "loss": 0.6869,
+      "step": 43100
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004494710526315789,
+      "loss": 0.6782,
+      "step": 43110
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00044939210526315785,
+      "loss": 0.6847,
+      "step": 43120
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004493131578947368,
+      "loss": 0.6886,
+      "step": 43130
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00044923421052631574,
+      "loss": 0.6848,
+      "step": 43140
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004491552631578947,
+      "loss": 0.6659,
+      "step": 43150
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004490763157894737,
+      "loss": 0.686,
+      "step": 43160
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004489973684210526,
+      "loss": 0.678,
+      "step": 43170
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00044891842105263154,
+      "loss": 0.6871,
+      "step": 43180
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004488394736842105,
+      "loss": 0.6864,
+      "step": 43190
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00044876052631578943,
+      "loss": 0.6782,
+      "step": 43200
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004486815789473684,
+      "loss": 0.6772,
+      "step": 43210
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00044860263157894733,
+      "loss": 0.6797,
+      "step": 43220
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004485236842105262,
+      "loss": 0.6743,
+      "step": 43230
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004484447368421052,
+      "loss": 0.6811,
+      "step": 43240
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004483657894736842,
+      "loss": 0.6851,
+      "step": 43250
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004482868421052631,
+      "loss": 0.6763,
+      "step": 43260
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00044820789473684207,
+      "loss": 0.6883,
+      "step": 43270
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000448128947368421,
+      "loss": 0.6758,
+      "step": 43280
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00044804999999999997,
+      "loss": 0.6941,
+      "step": 43290
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00044797105263157886,
+      "loss": 0.6786,
+      "step": 43300
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00044789210526315786,
+      "loss": 0.6783,
+      "step": 43310
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00044781315789473676,
+      "loss": 0.6819,
+      "step": 43320
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00044773421052631576,
+      "loss": 0.6831,
+      "step": 43330
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004476552631578947,
+      "loss": 0.6831,
+      "step": 43340
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00044757631578947366,
+      "loss": 0.6886,
+      "step": 43350
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004474973684210526,
+      "loss": 0.6761,
+      "step": 43360
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004474184210526315,
+      "loss": 0.6773,
+      "step": 43370
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004473394736842105,
+      "loss": 0.684,
+      "step": 43380
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004472605263157894,
+      "loss": 0.6911,
+      "step": 43390
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004471815789473684,
+      "loss": 0.6846,
+      "step": 43400
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004471026315789473,
+      "loss": 0.6824,
+      "step": 43410
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004470236842105263,
+      "loss": 0.679,
+      "step": 43420
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00044694473684210524,
+      "loss": 0.6858,
+      "step": 43430
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004468657894736842,
+      "loss": 0.6865,
+      "step": 43440
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00044678684210526314,
+      "loss": 0.6831,
+      "step": 43450
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00044670789473684203,
+      "loss": 0.6761,
+      "step": 43460
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00044662894736842104,
+      "loss": 0.6728,
+      "step": 43470
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00044654999999999993,
+      "loss": 0.6763,
+      "step": 43480
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00044647105263157893,
+      "loss": 0.6937,
+      "step": 43490
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00044639210526315783,
+      "loss": 0.6782,
+      "step": 43500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00044631315789473683,
+      "loss": 0.6802,
+      "step": 43510
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0004462342105263158,
+      "loss": 0.6843,
+      "step": 43520
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00044615526315789467,
+      "loss": 0.6806,
+      "step": 43530
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0004460763157894737,
+      "loss": 0.6702,
+      "step": 43540
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00044599736842105257,
+      "loss": 0.673,
+      "step": 43550
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00044591842105263157,
+      "loss": 0.6804,
+      "step": 43560
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00044583947368421047,
+      "loss": 0.6717,
+      "step": 43570
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00044576052631578947,
+      "loss": 0.6729,
+      "step": 43580
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00044568157894736836,
+      "loss": 0.6647,
+      "step": 43590
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0004456026315789473,
+      "loss": 0.666,
+      "step": 43600
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00044552368421052626,
+      "loss": 0.6766,
+      "step": 43610
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0004454447368421052,
+      "loss": 0.6738,
+      "step": 43620
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0004453657894736842,
+      "loss": 0.6589,
+      "step": 43630
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0004452868421052631,
+      "loss": 0.6726,
+      "step": 43640
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0004452078947368421,
+      "loss": 0.6659,
+      "step": 43650
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000445128947368421,
+      "loss": 0.6782,
+      "step": 43660
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00044504999999999995,
+      "loss": 0.6578,
+      "step": 43670
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0004449710526315789,
+      "loss": 0.6786,
+      "step": 43680
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00044489210526315785,
+      "loss": 0.6603,
+      "step": 43690
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0004448131578947368,
+      "loss": 0.655,
+      "step": 43700
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00044473421052631574,
+      "loss": 0.6451,
+      "step": 43710
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00044465526315789474,
+      "loss": 0.6692,
+      "step": 43720
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00044457631578947364,
+      "loss": 0.6575,
+      "step": 43730
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0004444973684210526,
+      "loss": 0.6682,
+      "step": 43740
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00044441842105263154,
+      "loss": 0.6701,
+      "step": 43750
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0004443394736842105,
+      "loss": 0.681,
+      "step": 43760
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00044426052631578943,
+      "loss": 0.6744,
+      "step": 43770
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0004441815789473684,
+      "loss": 0.6838,
+      "step": 43780
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0004441026315789473,
+      "loss": 0.6789,
+      "step": 43790
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0004440236842105263,
+      "loss": 0.6821,
+      "step": 43800
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0004439447368421053,
+      "loss": 0.6697,
+      "step": 43810
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0004438657894736842,
+      "loss": 0.6769,
+      "step": 43820
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0004437868421052631,
+      "loss": 0.6616,
+      "step": 43830
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00044370789473684207,
+      "loss": 0.6701,
+      "step": 43840
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0004436368421052631,
+      "loss": 0.6669,
+      "step": 43850
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0004435578947368421,
+      "loss": 0.6583,
+      "step": 43860
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00044347894736842104,
+      "loss": 0.6521,
+      "step": 43870
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00044339999999999994,
+      "loss": 0.6759,
+      "step": 43880
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00044332105263157894,
+      "loss": 0.6624,
+      "step": 43890
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00044324210526315783,
+      "loss": 0.6608,
+      "step": 43900
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00044316315789473683,
+      "loss": 0.6528,
+      "step": 43910
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00044308421052631573,
+      "loss": 0.6748,
+      "step": 43920
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00044300526315789473,
+      "loss": 0.6503,
+      "step": 43930
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0004429263157894736,
+      "loss": 0.6682,
+      "step": 43940
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0004428473684210526,
+      "loss": 0.6718,
+      "step": 43950
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0004427684210526316,
+      "loss": 0.6745,
+      "step": 43960
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00044268947368421047,
+      "loss": 0.6619,
+      "step": 43970
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0004426105263157895,
+      "loss": 0.6703,
+      "step": 43980
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00044253157894736837,
+      "loss": 0.6618,
+      "step": 43990
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00044245263157894737,
+      "loss": 0.6658,
+      "step": 44000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00044237368421052626,
+      "loss": 0.6844,
+      "step": 44010
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0004422947368421052,
+      "loss": 0.6885,
+      "step": 44020
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00044221578947368416,
+      "loss": 0.6581,
+      "step": 44030
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0004421368421052631,
+      "loss": 0.6635,
+      "step": 44040
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0004420578947368421,
+      "loss": 0.6656,
+      "step": 44050
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000441978947368421,
+      "loss": 0.6661,
+      "step": 44060
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0004419,
+      "loss": 0.6651,
+      "step": 44070
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0004418210526315789,
+      "loss": 0.6614,
+      "step": 44080
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00044174210526315785,
+      "loss": 0.6768,
+      "step": 44090
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0004416631578947368,
+      "loss": 0.6879,
+      "step": 44100
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00044158421052631575,
+      "loss": 0.6754,
+      "step": 44110
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0004415052631578947,
+      "loss": 0.6872,
+      "step": 44120
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00044142631578947364,
+      "loss": 0.6782,
+      "step": 44130
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00044134736842105254,
+      "loss": 0.6851,
+      "step": 44140
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00044126842105263154,
+      "loss": 0.6729,
+      "step": 44150
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0004411894736842105,
+      "loss": 0.691,
+      "step": 44160
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00044111052631578944,
+      "loss": 0.6858,
+      "step": 44170
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0004410315789473684,
+      "loss": 0.6821,
+      "step": 44180
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00044095263157894733,
+      "loss": 0.6695,
+      "step": 44190
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0004408736842105263,
+      "loss": 0.6793,
+      "step": 44200
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0004407947368421052,
+      "loss": 0.6671,
+      "step": 44210
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0004407157894736842,
+      "loss": 0.6732,
+      "step": 44220
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00044063684210526307,
+      "loss": 0.6715,
+      "step": 44230
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0004405578947368421,
+      "loss": 0.6761,
+      "step": 44240
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000440478947368421,
+      "loss": 0.6714,
+      "step": 44250
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00044039999999999997,
+      "loss": 0.6802,
+      "step": 44260
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0004403210526315789,
+      "loss": 0.6855,
+      "step": 44270
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0004402421052631578,
+      "loss": 0.6826,
+      "step": 44280
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0004401631578947368,
+      "loss": 0.6783,
+      "step": 44290
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0004400842105263157,
+      "loss": 0.6795,
+      "step": 44300
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0004400052631578947,
+      "loss": 0.6872,
+      "step": 44310
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0004399263157894736,
+      "loss": 0.6884,
+      "step": 44320
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0004398473684210526,
+      "loss": 0.6806,
+      "step": 44330
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00043976842105263156,
+      "loss": 0.679,
+      "step": 44340
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0004396894736842105,
+      "loss": 0.6755,
+      "step": 44350
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00043961052631578945,
+      "loss": 0.692,
+      "step": 44360
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00043953157894736835,
+      "loss": 0.6801,
+      "step": 44370
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00043945263157894735,
+      "loss": 0.6728,
+      "step": 44380
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00043937368421052624,
+      "loss": 0.675,
+      "step": 44390
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00043929473684210525,
+      "loss": 0.669,
+      "step": 44400
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00043921578947368414,
+      "loss": 0.6866,
+      "step": 44410
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00043913684210526314,
+      "loss": 0.6831,
+      "step": 44420
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0004390578947368421,
+      "loss": 0.6796,
+      "step": 44430
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000438978947368421,
+      "loss": 0.69,
+      "step": 44440
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0004389,
+      "loss": 0.6708,
+      "step": 44450
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0004388210526315789,
+      "loss": 0.6892,
+      "step": 44460
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0004387421052631579,
+      "loss": 0.6883,
+      "step": 44470
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0004386631578947368,
+      "loss": 0.6838,
+      "step": 44480
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0004385842105263158,
+      "loss": 0.6874,
+      "step": 44490
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004385052631578947,
+      "loss": 0.7018,
+      "step": 44500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004384263157894736,
+      "loss": 0.6789,
+      "step": 44510
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004383473684210526,
+      "loss": 0.6831,
+      "step": 44520
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004382684210526315,
+      "loss": 0.6841,
+      "step": 44530
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004381894736842105,
+      "loss": 0.6891,
+      "step": 44540
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004381105263157894,
+      "loss": 0.676,
+      "step": 44550
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004380315789473684,
+      "loss": 0.6773,
+      "step": 44560
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004379526315789473,
+      "loss": 0.6804,
+      "step": 44570
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00043787368421052626,
+      "loss": 0.6738,
+      "step": 44580
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004377947368421052,
+      "loss": 0.6774,
+      "step": 44590
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00043771578947368416,
+      "loss": 0.6737,
+      "step": 44600
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00043763684210526316,
+      "loss": 0.6809,
+      "step": 44610
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00043755789473684206,
+      "loss": 0.6816,
+      "step": 44620
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00043747894736842106,
+      "loss": 0.6745,
+      "step": 44630
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00043739999999999995,
+      "loss": 0.6779,
+      "step": 44640
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004373210526315789,
+      "loss": 0.6817,
+      "step": 44650
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00043724210526315785,
+      "loss": 0.6768,
+      "step": 44660
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004371631578947368,
+      "loss": 0.6703,
+      "step": 44670
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00043708421052631575,
+      "loss": 0.674,
+      "step": 44680
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004370052631578947,
+      "loss": 0.6787,
+      "step": 44690
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004369263157894736,
+      "loss": 0.6882,
+      "step": 44700
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004368473684210526,
+      "loss": 0.6762,
+      "step": 44710
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004367684210526316,
+      "loss": 0.6687,
+      "step": 44720
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004366894736842105,
+      "loss": 0.6868,
+      "step": 44730
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00043661052631578944,
+      "loss": 0.6633,
+      "step": 44740
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004365315789473684,
+      "loss": 0.6732,
+      "step": 44750
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00043645263157894733,
+      "loss": 0.6595,
+      "step": 44760
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004363736842105262,
+      "loss": 0.6707,
+      "step": 44770
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00043629473684210523,
+      "loss": 0.6553,
+      "step": 44780
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004362157894736841,
+      "loss": 0.6675,
+      "step": 44790
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004361368421052631,
+      "loss": 0.6593,
+      "step": 44800
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004360578947368421,
+      "loss": 0.6639,
+      "step": 44810
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000435978947368421,
+      "loss": 0.6538,
+      "step": 44820
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00043589999999999997,
+      "loss": 0.6666,
+      "step": 44830
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004358210526315789,
+      "loss": 0.6583,
+      "step": 44840
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00043574210526315787,
+      "loss": 0.6791,
+      "step": 44850
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00043566315789473676,
+      "loss": 0.6504,
+      "step": 44860
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00043558421052631576,
+      "loss": 0.6677,
+      "step": 44870
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00043550526315789466,
+      "loss": 0.651,
+      "step": 44880
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00043542631578947366,
+      "loss": 0.6632,
+      "step": 44890
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004353473684210526,
+      "loss": 0.6563,
+      "step": 44900
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00043526842105263156,
+      "loss": 0.6656,
+      "step": 44910
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004351894736842105,
+      "loss": 0.6628,
+      "step": 44920
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004351105263157894,
+      "loss": 0.6647,
+      "step": 44930
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004350315789473684,
+      "loss": 0.6707,
+      "step": 44940
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004349526315789473,
+      "loss": 0.675,
+      "step": 44950
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004348736842105263,
+      "loss": 0.6905,
+      "step": 44960
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004347947368421052,
+      "loss": 0.6732,
+      "step": 44970
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004347157894736842,
+      "loss": 0.6757,
+      "step": 44980
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00043463684210526314,
+      "loss": 0.685,
+      "step": 44990
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00043455789473684204,
+      "loss": 0.6752,
+      "step": 45000
+    },
+    {
+      "epoch": 0.45,
+      "eval_accuracy": 0.8599255154050108,
+      "eval_loss": 0.6640625,
+      "eval_runtime": 97.1461,
+      "eval_samples_per_second": 823.502,
+      "eval_steps_per_second": 1.616,
+      "step": 45000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00043447894736842104,
+      "loss": 0.6799,
+      "step": 45010
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00043439999999999993,
+      "loss": 0.6723,
+      "step": 45020
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00043432105263157894,
+      "loss": 0.6637,
+      "step": 45030
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00043424210526315783,
+      "loss": 0.6702,
+      "step": 45040
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00043416315789473683,
+      "loss": 0.6607,
+      "step": 45050
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004340842105263157,
+      "loss": 0.6635,
+      "step": 45060
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004340052631578947,
+      "loss": 0.675,
+      "step": 45070
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004339263157894737,
+      "loss": 0.6741,
+      "step": 45080
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00043384736842105257,
+      "loss": 0.6663,
+      "step": 45090
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004337684210526316,
+      "loss": 0.6532,
+      "step": 45100
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00043368947368421047,
+      "loss": 0.6647,
+      "step": 45110
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00043361052631578947,
+      "loss": 0.6682,
+      "step": 45120
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00043353157894736837,
+      "loss": 0.658,
+      "step": 45130
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004334526315789473,
+      "loss": 0.6703,
+      "step": 45140
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00043337368421052626,
+      "loss": 0.6698,
+      "step": 45150
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004332947368421052,
+      "loss": 0.6698,
+      "step": 45160
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004332157894736842,
+      "loss": 0.6666,
+      "step": 45170
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004331368421052631,
+      "loss": 0.6747,
+      "step": 45180
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004330578947368421,
+      "loss": 0.6839,
+      "step": 45190
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000432978947368421,
+      "loss": 0.6637,
+      "step": 45200
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004329,
+      "loss": 0.6625,
+      "step": 45210
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004328210526315789,
+      "loss": 0.6749,
+      "step": 45220
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00043274210526315785,
+      "loss": 0.6785,
+      "step": 45230
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004326631578947368,
+      "loss": 0.6743,
+      "step": 45240
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00043258421052631574,
+      "loss": 0.6759,
+      "step": 45250
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00043250526315789464,
+      "loss": 0.6694,
+      "step": 45260
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00043242631578947364,
+      "loss": 0.6657,
+      "step": 45270
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00043234736842105264,
+      "loss": 0.6585,
+      "step": 45280
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00043226842105263154,
+      "loss": 0.668,
+      "step": 45290
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004321894736842105,
+      "loss": 0.6783,
+      "step": 45300
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00043211052631578943,
+      "loss": 0.6774,
+      "step": 45310
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004320315789473684,
+      "loss": 0.674,
+      "step": 45320
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00043195263157894733,
+      "loss": 0.6705,
+      "step": 45330
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004318736842105263,
+      "loss": 0.6672,
+      "step": 45340
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004317947368421052,
+      "loss": 0.6842,
+      "step": 45350
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004317157894736842,
+      "loss": 0.6807,
+      "step": 45360
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004316368421052631,
+      "loss": 0.6782,
+      "step": 45370
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00043155789473684207,
+      "loss": 0.6739,
+      "step": 45380
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000431478947368421,
+      "loss": 0.6886,
+      "step": 45390
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00043139999999999997,
+      "loss": 0.6809,
+      "step": 45400
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004313210526315789,
+      "loss": 0.667,
+      "step": 45410
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004312421052631578,
+      "loss": 0.6836,
+      "step": 45420
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004311631578947368,
+      "loss": 0.6772,
+      "step": 45430
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004310842105263157,
+      "loss": 0.6763,
+      "step": 45440
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004310052631578947,
+      "loss": 0.6701,
+      "step": 45450
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00043092631578947366,
+      "loss": 0.6785,
+      "step": 45460
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004308473684210526,
+      "loss": 0.6789,
+      "step": 45470
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00043076842105263156,
+      "loss": 0.6808,
+      "step": 45480
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00043068947368421045,
+      "loss": 0.6843,
+      "step": 45490
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00043061052631578945,
+      "loss": 0.6633,
+      "step": 45500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00043053157894736835,
+      "loss": 0.6578,
+      "step": 45510
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00043045263157894735,
+      "loss": 0.6717,
+      "step": 45520
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00043037368421052624,
+      "loss": 0.669,
+      "step": 45530
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00043029473684210525,
+      "loss": 0.6861,
+      "step": 45540
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004302157894736842,
+      "loss": 0.6764,
+      "step": 45550
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004301368421052631,
+      "loss": 0.6799,
+      "step": 45560
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004300578947368421,
+      "loss": 0.6657,
+      "step": 45570
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000429978947368421,
+      "loss": 0.6633,
+      "step": 45580
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004299,
+      "loss": 0.6667,
+      "step": 45590
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004298210526315789,
+      "loss": 0.6721,
+      "step": 45600
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004297421052631579,
+      "loss": 0.6818,
+      "step": 45610
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004296631578947368,
+      "loss": 0.6824,
+      "step": 45620
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004295842105263157,
+      "loss": 0.6818,
+      "step": 45630
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00042950526315789473,
+      "loss": 0.6684,
+      "step": 45640
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00042943421052631575,
+      "loss": 0.6763,
+      "step": 45650
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004293552631578947,
+      "loss": 0.6786,
+      "step": 45660
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00042927631578947365,
+      "loss": 0.6848,
+      "step": 45670
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004291973684210526,
+      "loss": 0.6851,
+      "step": 45680
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00042911842105263154,
+      "loss": 0.6897,
+      "step": 45690
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00042903947368421055,
+      "loss": 0.6916,
+      "step": 45700
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00042896052631578944,
+      "loss": 0.6804,
+      "step": 45710
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004288815789473684,
+      "loss": 0.6782,
+      "step": 45720
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00042880263157894734,
+      "loss": 0.6748,
+      "step": 45730
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004287236842105263,
+      "loss": 0.6799,
+      "step": 45740
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00042864473684210523,
+      "loss": 0.6882,
+      "step": 45750
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004285657894736842,
+      "loss": 0.6796,
+      "step": 45760
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004284868421052631,
+      "loss": 0.6627,
+      "step": 45770
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004284078947368421,
+      "loss": 0.6646,
+      "step": 45780
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00042832894736842097,
+      "loss": 0.6738,
+      "step": 45790
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00042825,
+      "loss": 0.6697,
+      "step": 45800
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004281710526315789,
+      "loss": 0.6758,
+      "step": 45810
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00042809210526315787,
+      "loss": 0.6837,
+      "step": 45820
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004280131578947368,
+      "loss": 0.6833,
+      "step": 45830
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004279342105263157,
+      "loss": 0.6777,
+      "step": 45840
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004278552631578947,
+      "loss": 0.6806,
+      "step": 45850
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004277763157894736,
+      "loss": 0.6843,
+      "step": 45860
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004276973684210526,
+      "loss": 0.6858,
+      "step": 45870
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004276184210526315,
+      "loss": 0.6962,
+      "step": 45880
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004275394736842105,
+      "loss": 0.6877,
+      "step": 45890
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00042746052631578946,
+      "loss": 0.684,
+      "step": 45900
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00042738157894736835,
+      "loss": 0.6886,
+      "step": 45910
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00042730263157894735,
+      "loss": 0.6726,
+      "step": 45920
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00042722368421052625,
+      "loss": 0.6592,
+      "step": 45930
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00042714473684210525,
+      "loss": 0.657,
+      "step": 45940
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00042706578947368414,
+      "loss": 0.6604,
+      "step": 45950
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00042698684210526315,
+      "loss": 0.6607,
+      "step": 45960
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00042690789473684204,
+      "loss": 0.6795,
+      "step": 45970
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000426828947368421,
+      "loss": 0.6574,
+      "step": 45980
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00042675,
+      "loss": 0.6729,
+      "step": 45990
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004266710526315789,
+      "loss": 0.6701,
+      "step": 46000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004265921052631579,
+      "loss": 0.6795,
+      "step": 46010
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004265131578947368,
+      "loss": 0.6854,
+      "step": 46020
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004264342105263158,
+      "loss": 0.6835,
+      "step": 46030
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004263552631578947,
+      "loss": 0.6778,
+      "step": 46040
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004262763157894737,
+      "loss": 0.6796,
+      "step": 46050
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004261973684210526,
+      "loss": 0.6714,
+      "step": 46060
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004261184210526315,
+      "loss": 0.6824,
+      "step": 46070
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004260394736842105,
+      "loss": 0.6705,
+      "step": 46080
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004259605263157894,
+      "loss": 0.6843,
+      "step": 46090
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004258815789473684,
+      "loss": 0.6873,
+      "step": 46100
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004258026315789473,
+      "loss": 0.6683,
+      "step": 46110
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004257236842105263,
+      "loss": 0.662,
+      "step": 46120
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004256447368421052,
+      "loss": 0.6848,
+      "step": 46130
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00042556578947368416,
+      "loss": 0.6905,
+      "step": 46140
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004254868421052631,
+      "loss": 0.6866,
+      "step": 46150
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00042540789473684206,
+      "loss": 0.6808,
+      "step": 46160
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00042532894736842106,
+      "loss": 0.6693,
+      "step": 46170
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00042524999999999996,
+      "loss": 0.6554,
+      "step": 46180
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00042517105263157896,
+      "loss": 0.6523,
+      "step": 46190
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00042509210526315785,
+      "loss": 0.6718,
+      "step": 46200
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004250131578947368,
+      "loss": 0.6625,
+      "step": 46210
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00042493421052631575,
+      "loss": 0.6728,
+      "step": 46220
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004248552631578947,
+      "loss": 0.6786,
+      "step": 46230
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00042477631578947365,
+      "loss": 0.6808,
+      "step": 46240
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004246973684210526,
+      "loss": 0.6762,
+      "step": 46250
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004246184210526316,
+      "loss": 0.6669,
+      "step": 46260
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004245394736842105,
+      "loss": 0.6685,
+      "step": 46270
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00042446052631578944,
+      "loss": 0.6691,
+      "step": 46280
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004243815789473684,
+      "loss": 0.6511,
+      "step": 46290
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00042430263157894733,
+      "loss": 0.6455,
+      "step": 46300
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004242236842105263,
+      "loss": 0.6656,
+      "step": 46310
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00042414473684210523,
+      "loss": 0.6635,
+      "step": 46320
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004240657894736841,
+      "loss": 0.6599,
+      "step": 46330
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00042398684210526313,
+      "loss": 0.6679,
+      "step": 46340
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000423907894736842,
+      "loss": 0.6538,
+      "step": 46350
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000423828947368421,
+      "loss": 0.6692,
+      "step": 46360
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00042375,
+      "loss": 0.6859,
+      "step": 46370
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004236710526315789,
+      "loss": 0.679,
+      "step": 46380
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00042359210526315787,
+      "loss": 0.6792,
+      "step": 46390
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00042351315789473676,
+      "loss": 0.6799,
+      "step": 46400
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00042343421052631577,
+      "loss": 0.6776,
+      "step": 46410
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00042335526315789466,
+      "loss": 0.6638,
+      "step": 46420
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00042327631578947366,
+      "loss": 0.6603,
+      "step": 46430
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00042319736842105256,
+      "loss": 0.6629,
+      "step": 46440
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00042311842105263156,
+      "loss": 0.6683,
+      "step": 46450
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004230394736842105,
+      "loss": 0.6739,
+      "step": 46460
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004229605263157894,
+      "loss": 0.6617,
+      "step": 46470
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004228815789473684,
+      "loss": 0.6792,
+      "step": 46480
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004228026315789473,
+      "loss": 0.6733,
+      "step": 46490
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0004227236842105263,
+      "loss": 0.6617,
+      "step": 46500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0004226447368421052,
+      "loss": 0.6792,
+      "step": 46510
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0004225657894736842,
+      "loss": 0.6797,
+      "step": 46520
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0004224868421052631,
+      "loss": 0.6739,
+      "step": 46530
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0004224078947368421,
+      "loss": 0.6763,
+      "step": 46540
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00042232894736842104,
+      "loss": 0.6887,
+      "step": 46550
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00042224999999999994,
+      "loss": 0.6853,
+      "step": 46560
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00042217105263157894,
+      "loss": 0.665,
+      "step": 46570
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00042209210526315783,
+      "loss": 0.6908,
+      "step": 46580
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00042201315789473684,
+      "loss": 0.6696,
+      "step": 46590
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00042193421052631573,
+      "loss": 0.6524,
+      "step": 46600
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00042185526315789473,
+      "loss": 0.6785,
+      "step": 46610
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0004217763157894736,
+      "loss": 0.6727,
+      "step": 46620
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0004216973684210526,
+      "loss": 0.6731,
+      "step": 46630
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0004216184210526316,
+      "loss": 0.6638,
+      "step": 46640
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00042153947368421047,
+      "loss": 0.6859,
+      "step": 46650
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0004214605263157895,
+      "loss": 0.6735,
+      "step": 46660
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00042138157894736837,
+      "loss": 0.6754,
+      "step": 46670
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00042130263157894737,
+      "loss": 0.6737,
+      "step": 46680
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00042122368421052626,
+      "loss": 0.6727,
+      "step": 46690
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0004211447368421052,
+      "loss": 0.6797,
+      "step": 46700
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00042106578947368416,
+      "loss": 0.6881,
+      "step": 46710
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0004209868421052631,
+      "loss": 0.6768,
+      "step": 46720
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0004209078947368421,
+      "loss": 0.6828,
+      "step": 46730
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000420828947368421,
+      "loss": 0.6682,
+      "step": 46740
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00042075,
+      "loss": 0.6918,
+      "step": 46750
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0004206710526315789,
+      "loss": 0.6486,
+      "step": 46760
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00042059210526315785,
+      "loss": 0.6764,
+      "step": 46770
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0004205131578947368,
+      "loss": 0.6808,
+      "step": 46780
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00042043421052631575,
+      "loss": 0.6777,
+      "step": 46790
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0004203552631578947,
+      "loss": 0.6736,
+      "step": 46800
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00042027631578947364,
+      "loss": 0.6823,
+      "step": 46810
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00042019736842105265,
+      "loss": 0.6693,
+      "step": 46820
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00042012631578947367,
+      "loss": 0.6722,
+      "step": 46830
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00042004736842105256,
+      "loss": 0.675,
+      "step": 46840
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00041996842105263156,
+      "loss": 0.6845,
+      "step": 46850
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00041988947368421046,
+      "loss": 0.6852,
+      "step": 46860
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00041981052631578946,
+      "loss": 0.6726,
+      "step": 46870
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00041973157894736835,
+      "loss": 0.6752,
+      "step": 46880
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0004196526315789473,
+      "loss": 0.6716,
+      "step": 46890
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0004195736842105263,
+      "loss": 0.6745,
+      "step": 46900
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0004194947368421052,
+      "loss": 0.6786,
+      "step": 46910
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0004194157894736842,
+      "loss": 0.671,
+      "step": 46920
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0004193368421052631,
+      "loss": 0.6709,
+      "step": 46930
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0004192578947368421,
+      "loss": 0.6724,
+      "step": 46940
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000419178947368421,
+      "loss": 0.6708,
+      "step": 46950
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0004191,
+      "loss": 0.6692,
+      "step": 46960
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0004190210526315789,
+      "loss": 0.6705,
+      "step": 46970
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00041894210526315784,
+      "loss": 0.6702,
+      "step": 46980
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00041886315789473684,
+      "loss": 0.6726,
+      "step": 46990
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00041878421052631573,
+      "loss": 0.6681,
+      "step": 47000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00041870526315789474,
+      "loss": 0.679,
+      "step": 47010
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00041862631578947363,
+      "loss": 0.6913,
+      "step": 47020
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00041854736842105263,
+      "loss": 0.6748,
+      "step": 47030
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00041846842105263153,
+      "loss": 0.6694,
+      "step": 47040
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0004183894736842105,
+      "loss": 0.6742,
+      "step": 47050
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0004183105263157894,
+      "loss": 0.6874,
+      "step": 47060
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00041823157894736837,
+      "loss": 0.6867,
+      "step": 47070
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0004181526315789474,
+      "loss": 0.6798,
+      "step": 47080
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00041807368421052627,
+      "loss": 0.6688,
+      "step": 47090
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00041799473684210527,
+      "loss": 0.6667,
+      "step": 47100
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00041791578947368417,
+      "loss": 0.6644,
+      "step": 47110
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0004178368421052631,
+      "loss": 0.6694,
+      "step": 47120
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00041775789473684206,
+      "loss": 0.675,
+      "step": 47130
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000417678947368421,
+      "loss": 0.6737,
+      "step": 47140
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00041759999999999996,
+      "loss": 0.6563,
+      "step": 47150
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0004175210526315789,
+      "loss": 0.6591,
+      "step": 47160
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0004174421052631579,
+      "loss": 0.6632,
+      "step": 47170
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0004173631578947368,
+      "loss": 0.663,
+      "step": 47180
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00041728421052631575,
+      "loss": 0.6693,
+      "step": 47190
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0004172052631578947,
+      "loss": 0.6683,
+      "step": 47200
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00041712631578947365,
+      "loss": 0.6762,
+      "step": 47210
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0004170473684210526,
+      "loss": 0.6671,
+      "step": 47220
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00041696842105263155,
+      "loss": 0.6758,
+      "step": 47230
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00041688947368421044,
+      "loss": 0.6842,
+      "step": 47240
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00041681052631578944,
+      "loss": 0.6697,
+      "step": 47250
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0004167315789473684,
+      "loss": 0.6771,
+      "step": 47260
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00041665263157894734,
+      "loss": 0.6809,
+      "step": 47270
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0004165736842105263,
+      "loss": 0.681,
+      "step": 47280
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00041649473684210524,
+      "loss": 0.6644,
+      "step": 47290
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0004164157894736842,
+      "loss": 0.6846,
+      "step": 47300
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0004163368421052631,
+      "loss": 0.6707,
+      "step": 47310
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0004162578947368421,
+      "loss": 0.6552,
+      "step": 47320
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000416178947368421,
+      "loss": 0.6606,
+      "step": 47330
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0004161,
+      "loss": 0.6753,
+      "step": 47340
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0004160210526315789,
+      "loss": 0.6682,
+      "step": 47350
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0004159421052631579,
+      "loss": 0.6725,
+      "step": 47360
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0004158631578947368,
+      "loss": 0.666,
+      "step": 47370
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0004157842105263157,
+      "loss": 0.6741,
+      "step": 47380
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0004157052631578947,
+      "loss": 0.6695,
+      "step": 47390
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0004156263157894736,
+      "loss": 0.6607,
+      "step": 47400
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0004155473684210526,
+      "loss": 0.6789,
+      "step": 47410
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0004154684210526315,
+      "loss": 0.6655,
+      "step": 47420
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0004153894736842105,
+      "loss": 0.6711,
+      "step": 47430
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0004153105263157894,
+      "loss": 0.6638,
+      "step": 47440
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0004152315789473684,
+      "loss": 0.6634,
+      "step": 47450
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00041515263157894736,
+      "loss": 0.6739,
+      "step": 47460
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00041507368421052625,
+      "loss": 0.6712,
+      "step": 47470
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00041499473684210525,
+      "loss": 0.6556,
+      "step": 47480
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00041491578947368415,
+      "loss": 0.6629,
+      "step": 47490
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00041483684210526315,
+      "loss": 0.6603,
+      "step": 47500
+    },
+    {
+      "epoch": 0.47,
+      "eval_accuracy": 0.8611843722171862,
+      "eval_loss": 0.65625,
+      "eval_runtime": 97.2871,
+      "eval_samples_per_second": 822.308,
+      "eval_steps_per_second": 1.614,
+      "step": 47500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00041475789473684204,
+      "loss": 0.6665,
+      "step": 47510
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00041467894736842105,
+      "loss": 0.6721,
+      "step": 47520
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00041459999999999994,
+      "loss": 0.6617,
+      "step": 47530
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0004145210526315789,
+      "loss": 0.662,
+      "step": 47540
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0004144421052631579,
+      "loss": 0.6685,
+      "step": 47550
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0004143631578947368,
+      "loss": 0.665,
+      "step": 47560
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0004142842105263158,
+      "loss": 0.6763,
+      "step": 47570
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0004142052631578947,
+      "loss": 0.6729,
+      "step": 47580
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0004141263157894737,
+      "loss": 0.6688,
+      "step": 47590
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0004140473684210526,
+      "loss": 0.6775,
+      "step": 47600
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0004139684210526315,
+      "loss": 0.6608,
+      "step": 47610
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0004138894736842105,
+      "loss": 0.6663,
+      "step": 47620
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0004138105263157894,
+      "loss": 0.681,
+      "step": 47630
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0004137315789473684,
+      "loss": 0.6751,
+      "step": 47640
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0004136526315789473,
+      "loss": 0.6745,
+      "step": 47650
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0004135736842105263,
+      "loss": 0.6808,
+      "step": 47660
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0004134947368421052,
+      "loss": 0.6837,
+      "step": 47670
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00041341578947368416,
+      "loss": 0.6747,
+      "step": 47680
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0004133368421052631,
+      "loss": 0.6863,
+      "step": 47690
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00041325789473684206,
+      "loss": 0.6741,
+      "step": 47700
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000413178947368421,
+      "loss": 0.6686,
+      "step": 47710
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00041309999999999996,
+      "loss": 0.6508,
+      "step": 47720
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00041302105263157896,
+      "loss": 0.6655,
+      "step": 47730
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00041294210526315785,
+      "loss": 0.6534,
+      "step": 47740
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0004128631578947368,
+      "loss": 0.6614,
+      "step": 47750
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00041278421052631575,
+      "loss": 0.6596,
+      "step": 47760
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0004127052631578947,
+      "loss": 0.6649,
+      "step": 47770
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00041262631578947365,
+      "loss": 0.6662,
+      "step": 47780
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0004125473684210526,
+      "loss": 0.6594,
+      "step": 47790
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0004124684210526315,
+      "loss": 0.6673,
+      "step": 47800
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0004123894736842105,
+      "loss": 0.6615,
+      "step": 47810
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0004123105263157895,
+      "loss": 0.6646,
+      "step": 47820
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0004122315789473684,
+      "loss": 0.6659,
+      "step": 47830
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00041215263157894734,
+      "loss": 0.6614,
+      "step": 47840
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0004120736842105263,
+      "loss": 0.6622,
+      "step": 47850
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00041199473684210523,
+      "loss": 0.6761,
+      "step": 47860
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00041191578947368413,
+      "loss": 0.6659,
+      "step": 47870
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00041183684210526313,
+      "loss": 0.6695,
+      "step": 47880
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000411757894736842,
+      "loss": 0.6684,
+      "step": 47890
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00041167894736842103,
+      "loss": 0.6677,
+      "step": 47900
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0004116,
+      "loss": 0.6579,
+      "step": 47910
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0004115210526315789,
+      "loss": 0.6563,
+      "step": 47920
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00041144210526315787,
+      "loss": 0.6725,
+      "step": 47930
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0004113631578947368,
+      "loss": 0.6655,
+      "step": 47940
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00041128421052631577,
+      "loss": 0.658,
+      "step": 47950
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00041120526315789466,
+      "loss": 0.683,
+      "step": 47960
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00041112631578947367,
+      "loss": 0.6687,
+      "step": 47970
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00041104736842105256,
+      "loss": 0.6673,
+      "step": 47980
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00041096842105263156,
+      "loss": 0.6745,
+      "step": 47990
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00041088947368421046,
+      "loss": 0.6717,
+      "step": 48000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00041081052631578946,
+      "loss": 0.6718,
+      "step": 48010
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0004107315789473684,
+      "loss": 0.6721,
+      "step": 48020
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0004106526315789473,
+      "loss": 0.6806,
+      "step": 48030
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0004105736842105263,
+      "loss": 0.677,
+      "step": 48040
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0004104947368421052,
+      "loss": 0.6742,
+      "step": 48050
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0004104157894736842,
+      "loss": 0.6732,
+      "step": 48060
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0004103368421052631,
+      "loss": 0.6678,
+      "step": 48070
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0004102578947368421,
+      "loss": 0.6677,
+      "step": 48080
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000410178947368421,
+      "loss": 0.6641,
+      "step": 48090
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00041009999999999994,
+      "loss": 0.68,
+      "step": 48100
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00041002105263157894,
+      "loss": 0.656,
+      "step": 48110
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00040994210526315784,
+      "loss": 0.6642,
+      "step": 48120
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00040986315789473684,
+      "loss": 0.6736,
+      "step": 48130
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00040978421052631573,
+      "loss": 0.6775,
+      "step": 48140
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00040970526315789474,
+      "loss": 0.6898,
+      "step": 48150
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00040962631578947363,
+      "loss": 0.6685,
+      "step": 48160
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0004095473684210526,
+      "loss": 0.6727,
+      "step": 48170
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0004094684210526315,
+      "loss": 0.6671,
+      "step": 48180
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0004093894736842105,
+      "loss": 0.6635,
+      "step": 48190
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0004093105263157895,
+      "loss": 0.6664,
+      "step": 48200
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00040923157894736837,
+      "loss": 0.6647,
+      "step": 48210
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0004091526315789474,
+      "loss": 0.6597,
+      "step": 48220
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00040907368421052627,
+      "loss": 0.6663,
+      "step": 48230
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0004089947368421052,
+      "loss": 0.6622,
+      "step": 48240
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00040891578947368416,
+      "loss": 0.6684,
+      "step": 48250
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0004088368421052631,
+      "loss": 0.6645,
+      "step": 48260
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00040875789473684206,
+      "loss": 0.6604,
+      "step": 48270
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000408678947368421,
+      "loss": 0.6632,
+      "step": 48280
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0004086,
+      "loss": 0.6635,
+      "step": 48290
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0004085210526315789,
+      "loss": 0.6639,
+      "step": 48300
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0004084421052631579,
+      "loss": 0.6704,
+      "step": 48310
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0004083631578947368,
+      "loss": 0.6704,
+      "step": 48320
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00040828421052631575,
+      "loss": 0.6678,
+      "step": 48330
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0004082052631578947,
+      "loss": 0.6555,
+      "step": 48340
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00040812631578947365,
+      "loss": 0.6691,
+      "step": 48350
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00040804736842105254,
+      "loss": 0.6684,
+      "step": 48360
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00040796842105263154,
+      "loss": 0.6829,
+      "step": 48370
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00040788947368421055,
+      "loss": 0.6842,
+      "step": 48380
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00040781052631578944,
+      "loss": 0.6686,
+      "step": 48390
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0004077315789473684,
+      "loss": 0.6714,
+      "step": 48400
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00040765263157894734,
+      "loss": 0.6656,
+      "step": 48410
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0004075736842105263,
+      "loss": 0.6594,
+      "step": 48420
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00040749473684210523,
+      "loss": 0.6602,
+      "step": 48430
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0004074157894736842,
+      "loss": 0.6531,
+      "step": 48440
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0004073368421052631,
+      "loss": 0.6682,
+      "step": 48450
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0004072578947368421,
+      "loss": 0.6538,
+      "step": 48460
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000407178947368421,
+      "loss": 0.6633,
+      "step": 48470
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0004071,
+      "loss": 0.6666,
+      "step": 48480
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0004070210526315789,
+      "loss": 0.677,
+      "step": 48490
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00040694210526315787,
+      "loss": 0.6798,
+      "step": 48500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0004068631578947368,
+      "loss": 0.6707,
+      "step": 48510
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0004067842105263157,
+      "loss": 0.6508,
+      "step": 48520
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0004067052631578947,
+      "loss": 0.6545,
+      "step": 48530
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0004066263157894736,
+      "loss": 0.6535,
+      "step": 48540
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0004065473684210526,
+      "loss": 0.655,
+      "step": 48550
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00040646842105263156,
+      "loss": 0.6534,
+      "step": 48560
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0004063894736842105,
+      "loss": 0.6458,
+      "step": 48570
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00040631052631578946,
+      "loss": 0.6362,
+      "step": 48580
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00040623157894736835,
+      "loss": 0.6534,
+      "step": 48590
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00040615263157894735,
+      "loss": 0.6359,
+      "step": 48600
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00040607368421052625,
+      "loss": 0.6409,
+      "step": 48610
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00040599473684210525,
+      "loss": 0.6255,
+      "step": 48620
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00040591578947368415,
+      "loss": 0.6389,
+      "step": 48630
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00040583684210526315,
+      "loss": 0.6409,
+      "step": 48640
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00040575789473684204,
+      "loss": 0.6467,
+      "step": 48650
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000405678947368421,
+      "loss": 0.6451,
+      "step": 48660
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0004056,
+      "loss": 0.6524,
+      "step": 48670
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0004055210526315789,
+      "loss": 0.6358,
+      "step": 48680
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0004054421052631579,
+      "loss": 0.6621,
+      "step": 48690
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0004053631578947368,
+      "loss": 0.6634,
+      "step": 48700
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0004052842105263158,
+      "loss": 0.6623,
+      "step": 48710
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0004052052631578947,
+      "loss": 0.674,
+      "step": 48720
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0004051342105263158,
+      "loss": 0.6669,
+      "step": 48730
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0004050552631578947,
+      "loss": 0.6658,
+      "step": 48740
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00040497631578947365,
+      "loss": 0.666,
+      "step": 48750
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0004048973684210526,
+      "loss": 0.67,
+      "step": 48760
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00040481842105263155,
+      "loss": 0.6487,
+      "step": 48770
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00040473947368421044,
+      "loss": 0.6497,
+      "step": 48780
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00040466052631578944,
+      "loss": 0.6421,
+      "step": 48790
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00040458157894736834,
+      "loss": 0.6627,
+      "step": 48800
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00040450263157894734,
+      "loss": 0.6718,
+      "step": 48810
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0004044236842105263,
+      "loss": 0.6636,
+      "step": 48820
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00040434473684210524,
+      "loss": 0.6705,
+      "step": 48830
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0004042657894736842,
+      "loss": 0.6773,
+      "step": 48840
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00040418684210526313,
+      "loss": 0.6702,
+      "step": 48850
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0004041078947368421,
+      "loss": 0.658,
+      "step": 48860
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000404028947368421,
+      "loss": 0.6717,
+      "step": 48870
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00040395,
+      "loss": 0.6848,
+      "step": 48880
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0004038710526315789,
+      "loss": 0.6769,
+      "step": 48890
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0004037921052631579,
+      "loss": 0.6567,
+      "step": 48900
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0004037131578947368,
+      "loss": 0.6604,
+      "step": 48910
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00040363421052631577,
+      "loss": 0.6588,
+      "step": 48920
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0004035552631578947,
+      "loss": 0.6665,
+      "step": 48930
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0004034763157894736,
+      "loss": 0.6531,
+      "step": 48940
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0004033973684210526,
+      "loss": 0.6615,
+      "step": 48950
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0004033184210526315,
+      "loss": 0.6544,
+      "step": 48960
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0004032394736842105,
+      "loss": 0.6442,
+      "step": 48970
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0004031605263157894,
+      "loss": 0.6489,
+      "step": 48980
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0004030815789473684,
+      "loss": 0.6303,
+      "step": 48990
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00040300263157894736,
+      "loss": 0.6541,
+      "step": 49000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00040292368421052625,
+      "loss": 0.6591,
+      "step": 49010
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00040284473684210526,
+      "loss": 0.6793,
+      "step": 49020
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00040276578947368415,
+      "loss": 0.6784,
+      "step": 49030
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00040268684210526315,
+      "loss": 0.658,
+      "step": 49040
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00040260789473684205,
+      "loss": 0.6489,
+      "step": 49050
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00040252894736842105,
+      "loss": 0.6641,
+      "step": 49060
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00040244999999999994,
+      "loss": 0.6475,
+      "step": 49070
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0004023710526315789,
+      "loss": 0.6513,
+      "step": 49080
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00040229210526315784,
+      "loss": 0.6385,
+      "step": 49090
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0004022131578947368,
+      "loss": 0.6425,
+      "step": 49100
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0004021342105263158,
+      "loss": 0.6633,
+      "step": 49110
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0004020552631578947,
+      "loss": 0.6402,
+      "step": 49120
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0004019763157894737,
+      "loss": 0.644,
+      "step": 49130
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0004018973684210526,
+      "loss": 0.665,
+      "step": 49140
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00040181842105263153,
+      "loss": 0.6602,
+      "step": 49150
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0004017394736842105,
+      "loss": 0.6461,
+      "step": 49160
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0004016605263157894,
+      "loss": 0.6612,
+      "step": 49170
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0004015815789473684,
+      "loss": 0.6735,
+      "step": 49180
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0004015026315789473,
+      "loss": 0.6647,
+      "step": 49190
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0004014236842105263,
+      "loss": 0.6665,
+      "step": 49200
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0004013447368421052,
+      "loss": 0.6698,
+      "step": 49210
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0004012657894736842,
+      "loss": 0.6686,
+      "step": 49220
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0004011868421052631,
+      "loss": 0.6732,
+      "step": 49230
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00040110789473684206,
+      "loss": 0.6715,
+      "step": 49240
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000401028947368421,
+      "loss": 0.6625,
+      "step": 49250
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00040094999999999996,
+      "loss": 0.6672,
+      "step": 49260
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00040087105263157886,
+      "loss": 0.6581,
+      "step": 49270
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00040079210526315786,
+      "loss": 0.6664,
+      "step": 49280
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00040071315789473686,
+      "loss": 0.6461,
+      "step": 49290
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00040063421052631575,
+      "loss": 0.6499,
+      "step": 49300
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0004005552631578947,
+      "loss": 0.6513,
+      "step": 49310
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00040047631578947365,
+      "loss": 0.65,
+      "step": 49320
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0004003973684210526,
+      "loss": 0.6534,
+      "step": 49330
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00040031842105263155,
+      "loss": 0.6588,
+      "step": 49340
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0004002394736842105,
+      "loss": 0.6739,
+      "step": 49350
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0004001605263157894,
+      "loss": 0.667,
+      "step": 49360
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0004000815789473684,
+      "loss": 0.6504,
+      "step": 49370
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00040000263157894734,
+      "loss": 0.6691,
+      "step": 49380
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003999236842105263,
+      "loss": 0.66,
+      "step": 49390
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00039984473684210524,
+      "loss": 0.6655,
+      "step": 49400
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003997657894736842,
+      "loss": 0.6522,
+      "step": 49410
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00039968684210526313,
+      "loss": 0.6568,
+      "step": 49420
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00039960789473684203,
+      "loss": 0.6421,
+      "step": 49430
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00039952894736842103,
+      "loss": 0.6648,
+      "step": 49440
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003994499999999999,
+      "loss": 0.6596,
+      "step": 49450
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003993710526315789,
+      "loss": 0.6566,
+      "step": 49460
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003992921052631579,
+      "loss": 0.6488,
+      "step": 49470
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003992131578947368,
+      "loss": 0.6351,
+      "step": 49480
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00039913421052631577,
+      "loss": 0.6388,
+      "step": 49490
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00039905526315789467,
+      "loss": 0.6648,
+      "step": 49500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00039897631578947367,
+      "loss": 0.6592,
+      "step": 49510
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00039889736842105256,
+      "loss": 0.6631,
+      "step": 49520
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00039881842105263157,
+      "loss": 0.663,
+      "step": 49530
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00039873947368421046,
+      "loss": 0.6561,
+      "step": 49540
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00039866052631578946,
+      "loss": 0.6616,
+      "step": 49550
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003985815789473684,
+      "loss": 0.6621,
+      "step": 49560
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003985026315789473,
+      "loss": 0.6695,
+      "step": 49570
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003984236842105263,
+      "loss": 0.6611,
+      "step": 49580
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003983447368421052,
+      "loss": 0.6498,
+      "step": 49590
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003982657894736842,
+      "loss": 0.6701,
+      "step": 49600
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003981868421052631,
+      "loss": 0.6551,
+      "step": 49610
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003981078947368421,
+      "loss": 0.6552,
+      "step": 49620
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000398028947368421,
+      "loss": 0.6551,
+      "step": 49630
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00039794999999999994,
+      "loss": 0.6551,
+      "step": 49640
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003978710526315789,
+      "loss": 0.6727,
+      "step": 49650
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00039779210526315784,
+      "loss": 0.6609,
+      "step": 49660
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00039771315789473684,
+      "loss": 0.6655,
+      "step": 49670
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00039763421052631574,
+      "loss": 0.659,
+      "step": 49680
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00039755526315789474,
+      "loss": 0.6646,
+      "step": 49690
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00039747631578947363,
+      "loss": 0.6639,
+      "step": 49700
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00039739736842105263,
+      "loss": 0.6631,
+      "step": 49710
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00039731842105263153,
+      "loss": 0.6661,
+      "step": 49720
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003972394736842105,
+      "loss": 0.6596,
+      "step": 49730
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003971605263157894,
+      "loss": 0.6568,
+      "step": 49740
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003970815789473684,
+      "loss": 0.6572,
+      "step": 49750
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003970026315789474,
+      "loss": 0.6661,
+      "step": 49760
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00039692368421052627,
+      "loss": 0.6573,
+      "step": 49770
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00039684473684210527,
+      "loss": 0.6563,
+      "step": 49780
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00039676578947368417,
+      "loss": 0.6427,
+      "step": 49790
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003966868421052631,
+      "loss": 0.6553,
+      "step": 49800
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00039660789473684206,
+      "loss": 0.6391,
+      "step": 49810
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000396528947368421,
+      "loss": 0.6505,
+      "step": 49820
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00039644999999999996,
+      "loss": 0.6568,
+      "step": 49830
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003963710526315789,
+      "loss": 0.6554,
+      "step": 49840
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003962921052631579,
+      "loss": 0.6532,
+      "step": 49850
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003962131578947368,
+      "loss": 0.6553,
+      "step": 49860
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00039613421052631575,
+      "loss": 0.6671,
+      "step": 49870
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003960552631578947,
+      "loss": 0.6608,
+      "step": 49880
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00039597631578947365,
+      "loss": 0.6586,
+      "step": 49890
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003958973684210526,
+      "loss": 0.6551,
+      "step": 49900
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00039581842105263155,
+      "loss": 0.6553,
+      "step": 49910
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00039573947368421044,
+      "loss": 0.6585,
+      "step": 49920
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00039566052631578944,
+      "loss": 0.6604,
+      "step": 49930
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003955815789473684,
+      "loss": 0.6602,
+      "step": 49940
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00039550263157894734,
+      "loss": 0.6503,
+      "step": 49950
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003954236842105263,
+      "loss": 0.6615,
+      "step": 49960
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00039534473684210524,
+      "loss": 0.6618,
+      "step": 49970
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003952657894736842,
+      "loss": 0.6545,
+      "step": 49980
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003951868421052631,
+      "loss": 0.6545,
+      "step": 49990
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003951078947368421,
+      "loss": 0.6467,
+      "step": 50000
+    },
+    {
+      "epoch": 0.5,
+      "eval_accuracy": 0.8627790655085218,
+      "eval_loss": 0.6474609375,
+      "eval_runtime": 97.2319,
+      "eval_samples_per_second": 822.775,
+      "eval_steps_per_second": 1.615,
+      "step": 50000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000395028947368421,
+      "loss": 0.6478,
+      "step": 50010
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00039495,
+      "loss": 0.6406,
+      "step": 50020
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003948710526315789,
+      "loss": 0.6548,
+      "step": 50030
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003947921052631579,
+      "loss": 0.6629,
+      "step": 50040
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003947131578947368,
+      "loss": 0.6648,
+      "step": 50050
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003946342105263157,
+      "loss": 0.6518,
+      "step": 50060
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003945552631578947,
+      "loss": 0.6682,
+      "step": 50070
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003944763157894736,
+      "loss": 0.6668,
+      "step": 50080
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003943973684210526,
+      "loss": 0.6572,
+      "step": 50090
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003943184210526315,
+      "loss": 0.6682,
+      "step": 50100
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003942394736842105,
+      "loss": 0.6581,
+      "step": 50110
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00039416052631578946,
+      "loss": 0.644,
+      "step": 50120
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00039408157894736835,
+      "loss": 0.6578,
+      "step": 50130
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00039400263157894736,
+      "loss": 0.6699,
+      "step": 50140
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00039392368421052625,
+      "loss": 0.6599,
+      "step": 50150
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00039384473684210525,
+      "loss": 0.6607,
+      "step": 50160
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00039376578947368415,
+      "loss": 0.6576,
+      "step": 50170
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00039368684210526315,
+      "loss": 0.6477,
+      "step": 50180
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00039360789473684204,
+      "loss": 0.6685,
+      "step": 50190
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00039352894736842105,
+      "loss": 0.6593,
+      "step": 50200
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00039345,
+      "loss": 0.6451,
+      "step": 50210
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003933710526315789,
+      "loss": 0.6577,
+      "step": 50220
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003932921052631579,
+      "loss": 0.6486,
+      "step": 50230
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003932131578947368,
+      "loss": 0.6487,
+      "step": 50240
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003931342105263158,
+      "loss": 0.6562,
+      "step": 50250
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003930552631578947,
+      "loss": 0.6566,
+      "step": 50260
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003929763157894737,
+      "loss": 0.6451,
+      "step": 50270
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003928973684210526,
+      "loss": 0.6526,
+      "step": 50280
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00039281842105263153,
+      "loss": 0.65,
+      "step": 50290
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003927394736842105,
+      "loss": 0.6615,
+      "step": 50300
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003926605263157894,
+      "loss": 0.6484,
+      "step": 50310
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003925815789473684,
+      "loss": 0.6562,
+      "step": 50320
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003925026315789473,
+      "loss": 0.6456,
+      "step": 50330
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003924236842105263,
+      "loss": 0.6401,
+      "step": 50340
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003923447368421052,
+      "loss": 0.6466,
+      "step": 50350
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00039226578947368417,
+      "loss": 0.656,
+      "step": 50360
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003921868421052631,
+      "loss": 0.6617,
+      "step": 50370
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00039210789473684206,
+      "loss": 0.6593,
+      "step": 50380
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000392028947368421,
+      "loss": 0.6499,
+      "step": 50390
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00039194999999999996,
+      "loss": 0.6606,
+      "step": 50400
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000391878947368421,
+      "loss": 0.6769,
+      "step": 50410
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003918,
+      "loss": 0.6655,
+      "step": 50420
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003917210526315789,
+      "loss": 0.6597,
+      "step": 50430
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003916421052631579,
+      "loss": 0.6629,
+      "step": 50440
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003915631578947368,
+      "loss": 0.6632,
+      "step": 50450
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003914842105263158,
+      "loss": 0.6738,
+      "step": 50460
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003914052631578947,
+      "loss": 0.6579,
+      "step": 50470
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003913263157894736,
+      "loss": 0.6583,
+      "step": 50480
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003912473684210526,
+      "loss": 0.6537,
+      "step": 50490
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003911684210526315,
+      "loss": 0.6484,
+      "step": 50500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003910894736842105,
+      "loss": 0.657,
+      "step": 50510
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003910105263157894,
+      "loss": 0.6687,
+      "step": 50520
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003909315789473684,
+      "loss": 0.6594,
+      "step": 50530
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003908526315789473,
+      "loss": 0.6434,
+      "step": 50540
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00039077368421052626,
+      "loss": 0.6427,
+      "step": 50550
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00039069473684210526,
+      "loss": 0.6594,
+      "step": 50560
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00039061578947368415,
+      "loss": 0.6624,
+      "step": 50570
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00039053684210526316,
+      "loss": 0.6478,
+      "step": 50580
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00039045789473684205,
+      "loss": 0.6563,
+      "step": 50590
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00039037894736842105,
+      "loss": 0.6623,
+      "step": 50600
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00039029999999999995,
+      "loss": 0.645,
+      "step": 50610
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00039022105263157895,
+      "loss": 0.6657,
+      "step": 50620
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00039014210526315784,
+      "loss": 0.6548,
+      "step": 50630
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003900631578947368,
+      "loss": 0.6546,
+      "step": 50640
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003899842105263158,
+      "loss": 0.65,
+      "step": 50650
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003899052631578947,
+      "loss": 0.6426,
+      "step": 50660
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003898263157894737,
+      "loss": 0.6462,
+      "step": 50670
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003897473684210526,
+      "loss": 0.647,
+      "step": 50680
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003896684210526316,
+      "loss": 0.6356,
+      "step": 50690
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003895894736842105,
+      "loss": 0.6479,
+      "step": 50700
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00038951052631578943,
+      "loss": 0.6553,
+      "step": 50710
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003894315789473684,
+      "loss": 0.6536,
+      "step": 50720
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003893526315789473,
+      "loss": 0.6348,
+      "step": 50730
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003892736842105263,
+      "loss": 0.639,
+      "step": 50740
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003891947368421052,
+      "loss": 0.6415,
+      "step": 50750
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003891157894736842,
+      "loss": 0.6507,
+      "step": 50760
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003890368421052631,
+      "loss": 0.6582,
+      "step": 50770
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00038895789473684207,
+      "loss": 0.6578,
+      "step": 50780
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000388878947368421,
+      "loss": 0.6626,
+      "step": 50790
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00038879999999999996,
+      "loss": 0.6634,
+      "step": 50800
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003887210526315789,
+      "loss": 0.6446,
+      "step": 50810
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00038864210526315786,
+      "loss": 0.6516,
+      "step": 50820
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00038856315789473675,
+      "loss": 0.6617,
+      "step": 50830
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00038848421052631576,
+      "loss": 0.65,
+      "step": 50840
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003884052631578947,
+      "loss": 0.6599,
+      "step": 50850
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00038832631578947365,
+      "loss": 0.6413,
+      "step": 50860
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003882473684210526,
+      "loss": 0.6392,
+      "step": 50870
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00038816842105263155,
+      "loss": 0.6411,
+      "step": 50880
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003880894736842105,
+      "loss": 0.648,
+      "step": 50890
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003880105263157894,
+      "loss": 0.6573,
+      "step": 50900
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003879315789473684,
+      "loss": 0.6687,
+      "step": 50910
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003878526315789473,
+      "loss": 0.6561,
+      "step": 50920
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003877736842105263,
+      "loss": 0.6528,
+      "step": 50930
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00038769473684210524,
+      "loss": 0.6566,
+      "step": 50940
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003876157894736842,
+      "loss": 0.6565,
+      "step": 50950
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00038753684210526314,
+      "loss": 0.655,
+      "step": 50960
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00038745789473684203,
+      "loss": 0.6358,
+      "step": 50970
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00038737894736842103,
+      "loss": 0.6547,
+      "step": 50980
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00038729999999999993,
+      "loss": 0.6545,
+      "step": 50990
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00038722105263157893,
+      "loss": 0.641,
+      "step": 51000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003871421052631578,
+      "loss": 0.6476,
+      "step": 51010
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003870631578947368,
+      "loss": 0.6519,
+      "step": 51020
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003869842105263158,
+      "loss": 0.6684,
+      "step": 51030
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003869052631578947,
+      "loss": 0.64,
+      "step": 51040
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00038682631578947367,
+      "loss": 0.6486,
+      "step": 51050
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00038674736842105257,
+      "loss": 0.6557,
+      "step": 51060
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00038666842105263157,
+      "loss": 0.6658,
+      "step": 51070
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00038658947368421046,
+      "loss": 0.6544,
+      "step": 51080
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00038651052631578946,
+      "loss": 0.6711,
+      "step": 51090
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00038643157894736836,
+      "loss": 0.6641,
+      "step": 51100
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00038635263157894736,
+      "loss": 0.6559,
+      "step": 51110
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003862736842105263,
+      "loss": 0.6534,
+      "step": 51120
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003861947368421052,
+      "loss": 0.6714,
+      "step": 51130
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003861157894736842,
+      "loss": 0.6672,
+      "step": 51140
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003860368421052631,
+      "loss": 0.6625,
+      "step": 51150
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003859578947368421,
+      "loss": 0.6477,
+      "step": 51160
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000385878947368421,
+      "loss": 0.6416,
+      "step": 51170
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003858,
+      "loss": 0.6352,
+      "step": 51180
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003857210526315789,
+      "loss": 0.6331,
+      "step": 51190
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00038564210526315784,
+      "loss": 0.644,
+      "step": 51200
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00038556315789473684,
+      "loss": 0.6553,
+      "step": 51210
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00038548421052631574,
+      "loss": 0.655,
+      "step": 51220
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00038540526315789474,
+      "loss": 0.6516,
+      "step": 51230
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00038532631578947363,
+      "loss": 0.658,
+      "step": 51240
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00038524736842105264,
+      "loss": 0.6545,
+      "step": 51250
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00038516842105263153,
+      "loss": 0.6655,
+      "step": 51260
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003850894736842105,
+      "loss": 0.66,
+      "step": 51270
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00038501052631578943,
+      "loss": 0.6617,
+      "step": 51280
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003849315789473684,
+      "loss": 0.6322,
+      "step": 51290
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003848526315789474,
+      "loss": 0.6551,
+      "step": 51300
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003847736842105263,
+      "loss": 0.6413,
+      "step": 51310
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003846947368421053,
+      "loss": 0.6444,
+      "step": 51320
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00038461578947368417,
+      "loss": 0.6408,
+      "step": 51330
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003845368421052631,
+      "loss": 0.643,
+      "step": 51340
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00038445789473684207,
+      "loss": 0.6361,
+      "step": 51350
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000384378947368421,
+      "loss": 0.6386,
+      "step": 51360
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00038429999999999996,
+      "loss": 0.6516,
+      "step": 51370
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003842210526315789,
+      "loss": 0.6461,
+      "step": 51380
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003841421052631578,
+      "loss": 0.6445,
+      "step": 51390
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003840631578947368,
+      "loss": 0.652,
+      "step": 51400
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00038398421052631576,
+      "loss": 0.646,
+      "step": 51410
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003839052631578947,
+      "loss": 0.6523,
+      "step": 51420
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00038382631578947365,
+      "loss": 0.651,
+      "step": 51430
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003837473684210526,
+      "loss": 0.6525,
+      "step": 51440
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00038366842105263155,
+      "loss": 0.6591,
+      "step": 51450
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00038358947368421044,
+      "loss": 0.6548,
+      "step": 51460
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00038351052631578945,
+      "loss": 0.6574,
+      "step": 51470
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00038343157894736834,
+      "loss": 0.6645,
+      "step": 51480
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00038335263157894734,
+      "loss": 0.6542,
+      "step": 51490
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003832736842105263,
+      "loss": 0.6521,
+      "step": 51500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00038319473684210524,
+      "loss": 0.6574,
+      "step": 51510
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003831157894736842,
+      "loss": 0.6604,
+      "step": 51520
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00038303684210526314,
+      "loss": 0.6465,
+      "step": 51530
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003829578947368421,
+      "loss": 0.6578,
+      "step": 51540
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000382878947368421,
+      "loss": 0.6521,
+      "step": 51550
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003828,
+      "loss": 0.6559,
+      "step": 51560
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003827210526315789,
+      "loss": 0.6621,
+      "step": 51570
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003826421052631579,
+      "loss": 0.6728,
+      "step": 51580
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003825631578947368,
+      "loss": 0.663,
+      "step": 51590
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003824842105263158,
+      "loss": 0.6561,
+      "step": 51600
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003824052631578947,
+      "loss": 0.653,
+      "step": 51610
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003823263157894736,
+      "loss": 0.6648,
+      "step": 51620
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003822473684210526,
+      "loss": 0.6565,
+      "step": 51630
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003821684210526315,
+      "loss": 0.672,
+      "step": 51640
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003820894736842105,
+      "loss": 0.6618,
+      "step": 51650
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003820105263157894,
+      "loss": 0.666,
+      "step": 51660
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003819315789473684,
+      "loss": 0.6613,
+      "step": 51670
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00038185263157894736,
+      "loss": 0.6443,
+      "step": 51680
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00038177368421052625,
+      "loss": 0.6462,
+      "step": 51690
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00038169473684210526,
+      "loss": 0.6637,
+      "step": 51700
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00038161578947368415,
+      "loss": 0.6504,
+      "step": 51710
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00038153684210526315,
+      "loss": 0.6587,
+      "step": 51720
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00038145789473684205,
+      "loss": 0.6591,
+      "step": 51730
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00038137894736842105,
+      "loss": 0.6515,
+      "step": 51740
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00038129999999999994,
+      "loss": 0.6455,
+      "step": 51750
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003812210526315789,
+      "loss": 0.6511,
+      "step": 51760
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003811421052631579,
+      "loss": 0.6576,
+      "step": 51770
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003810631578947368,
+      "loss": 0.6549,
+      "step": 51780
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003809842105263158,
+      "loss": 0.6584,
+      "step": 51790
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003809052631578947,
+      "loss": 0.6557,
+      "step": 51800
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003808263157894737,
+      "loss": 0.6624,
+      "step": 51810
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003807473684210526,
+      "loss": 0.6569,
+      "step": 51820
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00038066842105263153,
+      "loss": 0.6588,
+      "step": 51830
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003805894736842105,
+      "loss": 0.6647,
+      "step": 51840
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00038051052631578943,
+      "loss": 0.6738,
+      "step": 51850
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00038043157894736843,
+      "loss": 0.6485,
+      "step": 51860
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003803526315789473,
+      "loss": 0.6389,
+      "step": 51870
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003802736842105263,
+      "loss": 0.6487,
+      "step": 51880
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003801947368421052,
+      "loss": 0.6574,
+      "step": 51890
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003801157894736842,
+      "loss": 0.6427,
+      "step": 51900
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003800368421052631,
+      "loss": 0.6439,
+      "step": 51910
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00037995789473684207,
+      "loss": 0.6496,
+      "step": 51920
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000379878947368421,
+      "loss": 0.6539,
+      "step": 51930
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00037979999999999996,
+      "loss": 0.6414,
+      "step": 51940
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00037972105263157886,
+      "loss": 0.6469,
+      "step": 51950
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00037964210526315786,
+      "loss": 0.6624,
+      "step": 51960
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00037956315789473686,
+      "loss": 0.6523,
+      "step": 51970
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00037948421052631576,
+      "loss": 0.6602,
+      "step": 51980
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003794052631578947,
+      "loss": 0.6606,
+      "step": 51990
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00037932631578947365,
+      "loss": 0.6581,
+      "step": 52000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003792473684210526,
+      "loss": 0.661,
+      "step": 52010
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00037916842105263155,
+      "loss": 0.6523,
+      "step": 52020
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003790894736842105,
+      "loss": 0.6457,
+      "step": 52030
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003790105263157894,
+      "loss": 0.6389,
+      "step": 52040
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003789315789473684,
+      "loss": 0.6602,
+      "step": 52050
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00037885263157894734,
+      "loss": 0.6364,
+      "step": 52060
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003787736842105263,
+      "loss": 0.6482,
+      "step": 52070
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00037869473684210524,
+      "loss": 0.6579,
+      "step": 52080
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003786157894736842,
+      "loss": 0.6532,
+      "step": 52090
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00037853684210526313,
+      "loss": 0.6534,
+      "step": 52100
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00037845789473684203,
+      "loss": 0.6596,
+      "step": 52110
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00037837894736842103,
+      "loss": 0.6409,
+      "step": 52120
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00037830789473684205,
+      "loss": 0.6608,
+      "step": 52130
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00037822894736842105,
+      "loss": 0.6557,
+      "step": 52140
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00037814999999999995,
+      "loss": 0.6693,
+      "step": 52150
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00037807105263157895,
+      "loss": 0.6585,
+      "step": 52160
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00037799210526315785,
+      "loss": 0.6476,
+      "step": 52170
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003779131578947368,
+      "loss": 0.6557,
+      "step": 52180
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00037783421052631574,
+      "loss": 0.66,
+      "step": 52190
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003777552631578947,
+      "loss": 0.653,
+      "step": 52200
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003776763157894737,
+      "loss": 0.6686,
+      "step": 52210
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003775973684210526,
+      "loss": 0.6333,
+      "step": 52220
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003775184210526316,
+      "loss": 0.6293,
+      "step": 52230
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003774394736842105,
+      "loss": 0.6291,
+      "step": 52240
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00037736052631578943,
+      "loss": 0.6429,
+      "step": 52250
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003772815789473684,
+      "loss": 0.6422,
+      "step": 52260
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00037720263157894733,
+      "loss": 0.634,
+      "step": 52270
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003771236842105263,
+      "loss": 0.6351,
+      "step": 52280
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003770447368421052,
+      "loss": 0.6347,
+      "step": 52290
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00037696578947368423,
+      "loss": 0.631,
+      "step": 52300
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003768868421052631,
+      "loss": 0.6298,
+      "step": 52310
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003768078947368421,
+      "loss": 0.642,
+      "step": 52320
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000376728947368421,
+      "loss": 0.6384,
+      "step": 52330
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00037664999999999997,
+      "loss": 0.6417,
+      "step": 52340
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003765710526315789,
+      "loss": 0.6363,
+      "step": 52350
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00037649210526315786,
+      "loss": 0.6414,
+      "step": 52360
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00037641315789473676,
+      "loss": 0.6366,
+      "step": 52370
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00037633421052631576,
+      "loss": 0.6137,
+      "step": 52380
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00037625526315789465,
+      "loss": 0.6338,
+      "step": 52390
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00037617631578947366,
+      "loss": 0.6376,
+      "step": 52400
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003760973684210526,
+      "loss": 0.6323,
+      "step": 52410
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00037601842105263155,
+      "loss": 0.6304,
+      "step": 52420
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003759394736842105,
+      "loss": 0.6433,
+      "step": 52430
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00037586052631578945,
+      "loss": 0.6372,
+      "step": 52440
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003757815789473684,
+      "loss": 0.6331,
+      "step": 52450
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003757026315789473,
+      "loss": 0.6432,
+      "step": 52460
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003756236842105263,
+      "loss": 0.655,
+      "step": 52470
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003755447368421052,
+      "loss": 0.6527,
+      "step": 52480
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003754657894736842,
+      "loss": 0.6574,
+      "step": 52490
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00037538684210526314,
+      "loss": 0.665,
+      "step": 52500
+    },
+    {
+      "epoch": 0.53,
+      "eval_accuracy": 0.8649771990754928,
+      "eval_loss": 0.6337890625,
+      "eval_runtime": 97.2311,
+      "eval_samples_per_second": 822.782,
+      "eval_steps_per_second": 1.615,
+      "step": 52500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003753078947368421,
+      "loss": 0.6597,
+      "step": 52510
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00037522894736842104,
+      "loss": 0.6546,
+      "step": 52520
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00037514999999999993,
+      "loss": 0.6526,
+      "step": 52530
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00037507105263157893,
+      "loss": 0.6492,
+      "step": 52540
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003749921052631578,
+      "loss": 0.6548,
+      "step": 52550
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00037491315789473683,
+      "loss": 0.6473,
+      "step": 52560
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003748342105263157,
+      "loss": 0.6521,
+      "step": 52570
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003747552631578947,
+      "loss": 0.6439,
+      "step": 52580
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003746763157894737,
+      "loss": 0.6484,
+      "step": 52590
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00037459736842105257,
+      "loss": 0.653,
+      "step": 52600
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00037451842105263157,
+      "loss": 0.6508,
+      "step": 52610
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00037443947368421046,
+      "loss": 0.6658,
+      "step": 52620
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00037436052631578947,
+      "loss": 0.6619,
+      "step": 52630
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00037428157894736836,
+      "loss": 0.6419,
+      "step": 52640
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00037420263157894736,
+      "loss": 0.6504,
+      "step": 52650
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00037412368421052626,
+      "loss": 0.66,
+      "step": 52660
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003740447368421052,
+      "loss": 0.6504,
+      "step": 52670
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003739657894736842,
+      "loss": 0.6578,
+      "step": 52680
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003738868421052631,
+      "loss": 0.6597,
+      "step": 52690
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003738078947368421,
+      "loss": 0.6533,
+      "step": 52700
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000373728947368421,
+      "loss": 0.662,
+      "step": 52710
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00037365,
+      "loss": 0.6571,
+      "step": 52720
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003735710526315789,
+      "loss": 0.6505,
+      "step": 52730
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00037349210526315784,
+      "loss": 0.6408,
+      "step": 52740
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003734131578947368,
+      "loss": 0.6477,
+      "step": 52750
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00037333421052631574,
+      "loss": 0.6433,
+      "step": 52760
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00037325526315789474,
+      "loss": 0.661,
+      "step": 52770
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00037317631578947364,
+      "loss": 0.6455,
+      "step": 52780
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00037309736842105264,
+      "loss": 0.6468,
+      "step": 52790
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00037301842105263153,
+      "loss": 0.6506,
+      "step": 52800
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00037293947368421054,
+      "loss": 0.6513,
+      "step": 52810
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00037286052631578943,
+      "loss": 0.651,
+      "step": 52820
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003727815789473684,
+      "loss": 0.6478,
+      "step": 52830
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00037270263157894733,
+      "loss": 0.6523,
+      "step": 52840
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003726236842105263,
+      "loss": 0.6505,
+      "step": 52850
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003725447368421053,
+      "loss": 0.6493,
+      "step": 52860
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00037246578947368417,
+      "loss": 0.6512,
+      "step": 52870
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003723868421052632,
+      "loss": 0.6598,
+      "step": 52880
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00037230789473684207,
+      "loss": 0.6508,
+      "step": 52890
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000372228947368421,
+      "loss": 0.6491,
+      "step": 52900
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00037214999999999997,
+      "loss": 0.6432,
+      "step": 52910
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003720710526315789,
+      "loss": 0.6409,
+      "step": 52920
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00037199210526315786,
+      "loss": 0.6495,
+      "step": 52930
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003719131578947368,
+      "loss": 0.6614,
+      "step": 52940
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003718342105263158,
+      "loss": 0.6442,
+      "step": 52950
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003717552631578947,
+      "loss": 0.6475,
+      "step": 52960
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00037167631578947366,
+      "loss": 0.6479,
+      "step": 52970
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003715973684210526,
+      "loss": 0.6461,
+      "step": 52980
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00037151842105263155,
+      "loss": 0.6597,
+      "step": 52990
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003714394736842105,
+      "loss": 0.6391,
+      "step": 53000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00037136052631578945,
+      "loss": 0.662,
+      "step": 53010
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00037128157894736834,
+      "loss": 0.657,
+      "step": 53020
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00037120263157894735,
+      "loss": 0.6601,
+      "step": 53030
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00037112368421052624,
+      "loss": 0.6527,
+      "step": 53040
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00037104473684210524,
+      "loss": 0.6575,
+      "step": 53050
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003709657894736842,
+      "loss": 0.6459,
+      "step": 53060
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00037088684210526314,
+      "loss": 0.6423,
+      "step": 53070
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003708078947368421,
+      "loss": 0.6438,
+      "step": 53080
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000370728947368421,
+      "loss": 0.6432,
+      "step": 53090
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00037065,
+      "loss": 0.6415,
+      "step": 53100
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003705710526315789,
+      "loss": 0.6409,
+      "step": 53110
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003704921052631579,
+      "loss": 0.6501,
+      "step": 53120
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003704131578947368,
+      "loss": 0.6508,
+      "step": 53130
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003703342105263158,
+      "loss": 0.6429,
+      "step": 53140
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003702552631578947,
+      "loss": 0.6593,
+      "step": 53150
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003701763157894736,
+      "loss": 0.6582,
+      "step": 53160
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003700973684210526,
+      "loss": 0.6425,
+      "step": 53170
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003700184210526315,
+      "loss": 0.6582,
+      "step": 53180
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003699394736842105,
+      "loss": 0.6554,
+      "step": 53190
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003698605263157894,
+      "loss": 0.6587,
+      "step": 53200
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00036978947368421054,
+      "loss": 0.6533,
+      "step": 53210
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00036971052631578944,
+      "loss": 0.656,
+      "step": 53220
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00036963157894736844,
+      "loss": 0.6619,
+      "step": 53230
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00036955263157894733,
+      "loss": 0.6465,
+      "step": 53240
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003694736842105263,
+      "loss": 0.6526,
+      "step": 53250
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00036939473684210523,
+      "loss": 0.6582,
+      "step": 53260
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003693157894736842,
+      "loss": 0.6478,
+      "step": 53270
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00036923684210526307,
+      "loss": 0.6473,
+      "step": 53280
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003691578947368421,
+      "loss": 0.6559,
+      "step": 53290
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003690789473684211,
+      "loss": 0.6522,
+      "step": 53300
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00036899999999999997,
+      "loss": 0.6699,
+      "step": 53310
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003689210526315789,
+      "loss": 0.6507,
+      "step": 53320
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00036884210526315787,
+      "loss": 0.6577,
+      "step": 53330
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003687631578947368,
+      "loss": 0.6501,
+      "step": 53340
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00036868421052631576,
+      "loss": 0.6583,
+      "step": 53350
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003686052631578947,
+      "loss": 0.6705,
+      "step": 53360
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003685263157894736,
+      "loss": 0.6437,
+      "step": 53370
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003684473684210526,
+      "loss": 0.6594,
+      "step": 53380
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00036836842105263156,
+      "loss": 0.669,
+      "step": 53390
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003682894736842105,
+      "loss": 0.6464,
+      "step": 53400
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00036821052631578945,
+      "loss": 0.6407,
+      "step": 53410
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003681315789473684,
+      "loss": 0.638,
+      "step": 53420
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00036805263157894735,
+      "loss": 0.6348,
+      "step": 53430
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00036797368421052624,
+      "loss": 0.6266,
+      "step": 53440
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00036789473684210525,
+      "loss": 0.6391,
+      "step": 53450
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00036781578947368414,
+      "loss": 0.6362,
+      "step": 53460
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00036773684210526314,
+      "loss": 0.6494,
+      "step": 53470
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00036765789473684204,
+      "loss": 0.6467,
+      "step": 53480
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00036757894736842104,
+      "loss": 0.6497,
+      "step": 53490
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003675,
+      "loss": 0.6531,
+      "step": 53500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003674210526315789,
+      "loss": 0.6574,
+      "step": 53510
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003673421052631579,
+      "loss": 0.6492,
+      "step": 53520
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003672631578947368,
+      "loss": 0.6602,
+      "step": 53530
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003671842105263158,
+      "loss": 0.6606,
+      "step": 53540
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003671052631578947,
+      "loss": 0.6547,
+      "step": 53550
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003670263157894737,
+      "loss": 0.6544,
+      "step": 53560
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00036694736842105257,
+      "loss": 0.6526,
+      "step": 53570
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003668684210526315,
+      "loss": 0.6398,
+      "step": 53580
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003667894736842105,
+      "loss": 0.6504,
+      "step": 53590
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003667105263157894,
+      "loss": 0.6406,
+      "step": 53600
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003666315789473684,
+      "loss": 0.6389,
+      "step": 53610
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003665526315789473,
+      "loss": 0.6466,
+      "step": 53620
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003664736842105263,
+      "loss": 0.6452,
+      "step": 53630
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003663947368421052,
+      "loss": 0.6412,
+      "step": 53640
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00036631578947368416,
+      "loss": 0.6413,
+      "step": 53650
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003662368421052631,
+      "loss": 0.6475,
+      "step": 53660
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00036615789473684206,
+      "loss": 0.6469,
+      "step": 53670
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00036607894736842106,
+      "loss": 0.6475,
+      "step": 53680
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00036599999999999995,
+      "loss": 0.6378,
+      "step": 53690
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00036592105263157895,
+      "loss": 0.6466,
+      "step": 53700
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00036584210526315785,
+      "loss": 0.6557,
+      "step": 53710
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00036576315789473685,
+      "loss": 0.64,
+      "step": 53720
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00036568421052631574,
+      "loss": 0.6391,
+      "step": 53730
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003656052631578947,
+      "loss": 0.6496,
+      "step": 53740
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00036552631578947364,
+      "loss": 0.6429,
+      "step": 53750
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003654473684210526,
+      "loss": 0.6265,
+      "step": 53760
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003653684210526316,
+      "loss": 0.6269,
+      "step": 53770
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003652894736842105,
+      "loss": 0.6271,
+      "step": 53780
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003652105263157895,
+      "loss": 0.6333,
+      "step": 53790
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003651315789473684,
+      "loss": 0.6269,
+      "step": 53800
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00036505263157894733,
+      "loss": 0.6305,
+      "step": 53810
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003649736842105263,
+      "loss": 0.6251,
+      "step": 53820
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00036489473684210523,
+      "loss": 0.6266,
+      "step": 53830
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003648157894736842,
+      "loss": 0.6291,
+      "step": 53840
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003647368421052631,
+      "loss": 0.6326,
+      "step": 53850
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003646578947368421,
+      "loss": 0.6255,
+      "step": 53860
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000364578947368421,
+      "loss": 0.6416,
+      "step": 53870
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00036449999999999997,
+      "loss": 0.6301,
+      "step": 53880
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003644210526315789,
+      "loss": 0.6356,
+      "step": 53890
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00036434210526315787,
+      "loss": 0.6396,
+      "step": 53900
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003642631578947368,
+      "loss": 0.6397,
+      "step": 53910
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00036418421052631576,
+      "loss": 0.624,
+      "step": 53920
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00036410526315789466,
+      "loss": 0.6294,
+      "step": 53930
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00036402631578947366,
+      "loss": 0.6254,
+      "step": 53940
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003639473684210526,
+      "loss": 0.6404,
+      "step": 53950
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00036386842105263156,
+      "loss": 0.6266,
+      "step": 53960
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003637894736842105,
+      "loss": 0.6206,
+      "step": 53970
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00036371052631578945,
+      "loss": 0.6368,
+      "step": 53980
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003636315789473684,
+      "loss": 0.6368,
+      "step": 53990
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003635526315789473,
+      "loss": 0.6188,
+      "step": 54000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003634736842105263,
+      "loss": 0.6486,
+      "step": 54010
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003633947368421052,
+      "loss": 0.6524,
+      "step": 54020
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003633157894736842,
+      "loss": 0.652,
+      "step": 54030
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003632368421052631,
+      "loss": 0.6692,
+      "step": 54040
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003631578947368421,
+      "loss": 0.6525,
+      "step": 54050
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00036307894736842104,
+      "loss": 0.66,
+      "step": 54060
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00036299999999999993,
+      "loss": 0.6666,
+      "step": 54070
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00036292105263157894,
+      "loss": 0.6658,
+      "step": 54080
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00036284210526315783,
+      "loss": 0.6607,
+      "step": 54090
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00036276315789473683,
+      "loss": 0.644,
+      "step": 54100
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003626842105263157,
+      "loss": 0.6426,
+      "step": 54110
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00036260526315789473,
+      "loss": 0.6392,
+      "step": 54120
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003625263157894736,
+      "loss": 0.6323,
+      "step": 54130
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00036244736842105257,
+      "loss": 0.6458,
+      "step": 54140
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003623684210526316,
+      "loss": 0.6532,
+      "step": 54150
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00036228947368421047,
+      "loss": 0.6577,
+      "step": 54160
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00036221052631578947,
+      "loss": 0.643,
+      "step": 54170
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00036213157894736836,
+      "loss": 0.6336,
+      "step": 54180
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00036205263157894737,
+      "loss": 0.6455,
+      "step": 54190
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00036197368421052626,
+      "loss": 0.6565,
+      "step": 54200
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00036189473684210526,
+      "loss": 0.6515,
+      "step": 54210
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00036181578947368416,
+      "loss": 0.6487,
+      "step": 54220
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003617368421052631,
+      "loss": 0.6596,
+      "step": 54230
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003616578947368421,
+      "loss": 0.6494,
+      "step": 54240
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000361578947368421,
+      "loss": 0.6481,
+      "step": 54250
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003615,
+      "loss": 0.6374,
+      "step": 54260
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003614210526315789,
+      "loss": 0.6385,
+      "step": 54270
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003613421052631579,
+      "loss": 0.6185,
+      "step": 54280
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003612631578947368,
+      "loss": 0.6492,
+      "step": 54290
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00036118421052631574,
+      "loss": 0.6324,
+      "step": 54300
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003611052631578947,
+      "loss": 0.6396,
+      "step": 54310
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00036102631578947364,
+      "loss": 0.6416,
+      "step": 54320
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00036094736842105264,
+      "loss": 0.6386,
+      "step": 54330
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00036086842105263154,
+      "loss": 0.6363,
+      "step": 54340
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00036078947368421054,
+      "loss": 0.6421,
+      "step": 54350
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00036071052631578943,
+      "loss": 0.6554,
+      "step": 54360
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003606315789473684,
+      "loss": 0.6553,
+      "step": 54370
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00036055263157894733,
+      "loss": 0.6402,
+      "step": 54380
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003604736842105263,
+      "loss": 0.6519,
+      "step": 54390
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003603947368421052,
+      "loss": 0.657,
+      "step": 54400
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003603157894736842,
+      "loss": 0.6611,
+      "step": 54410
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003602368421052632,
+      "loss": 0.6669,
+      "step": 54420
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00036015789473684207,
+      "loss": 0.6622,
+      "step": 54430
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000360078947368421,
+      "loss": 0.6494,
+      "step": 54440
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00035999999999999997,
+      "loss": 0.6571,
+      "step": 54450
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003599210526315789,
+      "loss": 0.6397,
+      "step": 54460
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00035984210526315787,
+      "loss": 0.6506,
+      "step": 54470
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003597631578947368,
+      "loss": 0.6398,
+      "step": 54480
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003596842105263157,
+      "loss": 0.6269,
+      "step": 54490
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003596052631578947,
+      "loss": 0.6423,
+      "step": 54500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00035952631578947366,
+      "loss": 0.6452,
+      "step": 54510
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00035945526315789473,
+      "loss": 0.644,
+      "step": 54520
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00035937631578947363,
+      "loss": 0.6432,
+      "step": 54530
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00035929736842105263,
+      "loss": 0.6419,
+      "step": 54540
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003592184210526315,
+      "loss": 0.6584,
+      "step": 54550
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00035913947368421047,
+      "loss": 0.649,
+      "step": 54560
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003590605263157894,
+      "loss": 0.6497,
+      "step": 54570
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00035898157894736837,
+      "loss": 0.6494,
+      "step": 54580
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00035890263157894737,
+      "loss": 0.6523,
+      "step": 54590
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00035882368421052627,
+      "loss": 0.6405,
+      "step": 54600
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00035874473684210527,
+      "loss": 0.6463,
+      "step": 54610
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00035866578947368416,
+      "loss": 0.6429,
+      "step": 54620
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00035858684210526316,
+      "loss": 0.6452,
+      "step": 54630
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00035850789473684206,
+      "loss": 0.639,
+      "step": 54640
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000358428947368421,
+      "loss": 0.6324,
+      "step": 54650
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00035834999999999996,
+      "loss": 0.6429,
+      "step": 54660
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003582710526315789,
+      "loss": 0.6392,
+      "step": 54670
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003581921052631579,
+      "loss": 0.6248,
+      "step": 54680
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003581131578947368,
+      "loss": 0.6205,
+      "step": 54690
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003580342105263158,
+      "loss": 0.6234,
+      "step": 54700
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003579552631578947,
+      "loss": 0.6364,
+      "step": 54710
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00035787631578947365,
+      "loss": 0.6511,
+      "step": 54720
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003577973684210526,
+      "loss": 0.6481,
+      "step": 54730
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00035771842105263154,
+      "loss": 0.6538,
+      "step": 54740
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003576394736842105,
+      "loss": 0.6412,
+      "step": 54750
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00035756052631578944,
+      "loss": 0.6432,
+      "step": 54760
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00035748157894736844,
+      "loss": 0.656,
+      "step": 54770
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00035740263157894734,
+      "loss": 0.6436,
+      "step": 54780
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003573236842105263,
+      "loss": 0.6444,
+      "step": 54790
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00035724473684210523,
+      "loss": 0.636,
+      "step": 54800
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003571657894736842,
+      "loss": 0.6347,
+      "step": 54810
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00035708684210526313,
+      "loss": 0.6237,
+      "step": 54820
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003570078947368421,
+      "loss": 0.6384,
+      "step": 54830
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00035692894736842097,
+      "loss": 0.6369,
+      "step": 54840
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00035685,
+      "loss": 0.649,
+      "step": 54850
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003567710526315789,
+      "loss": 0.6448,
+      "step": 54860
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00035669210526315787,
+      "loss": 0.6391,
+      "step": 54870
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003566131578947368,
+      "loss": 0.6459,
+      "step": 54880
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00035653421052631577,
+      "loss": 0.6546,
+      "step": 54890
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003564552631578947,
+      "loss": 0.6441,
+      "step": 54900
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003563763157894736,
+      "loss": 0.6526,
+      "step": 54910
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003562973684210526,
+      "loss": 0.6456,
+      "step": 54920
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003562184210526315,
+      "loss": 0.6433,
+      "step": 54930
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003561394736842105,
+      "loss": 0.6512,
+      "step": 54940
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00035606052631578946,
+      "loss": 0.6423,
+      "step": 54950
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003559815789473684,
+      "loss": 0.6436,
+      "step": 54960
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00035590263157894735,
+      "loss": 0.6341,
+      "step": 54970
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00035582368421052625,
+      "loss": 0.6416,
+      "step": 54980
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00035574473684210525,
+      "loss": 0.6439,
+      "step": 54990
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00035566578947368414,
+      "loss": 0.625,
+      "step": 55000
+    },
+    {
+      "epoch": 0.55,
+      "eval_accuracy": 0.8664423590057567,
+      "eval_loss": 0.62548828125,
+      "eval_runtime": 97.6676,
+      "eval_samples_per_second": 819.104,
+      "eval_steps_per_second": 1.607,
+      "step": 55000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00035558684210526315,
+      "loss": 0.644,
+      "step": 55010
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00035550789473684204,
+      "loss": 0.6439,
+      "step": 55020
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00035542894736842104,
+      "loss": 0.646,
+      "step": 55030
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00035535,
+      "loss": 0.6523,
+      "step": 55040
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003552710526315789,
+      "loss": 0.6389,
+      "step": 55050
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003551921052631579,
+      "loss": 0.6624,
+      "step": 55060
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003551131578947368,
+      "loss": 0.6395,
+      "step": 55070
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003550342105263158,
+      "loss": 0.644,
+      "step": 55080
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003549552631578947,
+      "loss": 0.6422,
+      "step": 55090
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003548763157894737,
+      "loss": 0.6481,
+      "step": 55100
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003547973684210526,
+      "loss": 0.6472,
+      "step": 55110
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003547184210526316,
+      "loss": 0.649,
+      "step": 55120
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00035463947368421047,
+      "loss": 0.6428,
+      "step": 55130
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003545605263157894,
+      "loss": 0.6383,
+      "step": 55140
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003544815789473684,
+      "loss": 0.6402,
+      "step": 55150
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003544026315789473,
+      "loss": 0.6516,
+      "step": 55160
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003543236842105263,
+      "loss": 0.6429,
+      "step": 55170
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003542447368421052,
+      "loss": 0.6501,
+      "step": 55180
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003541657894736842,
+      "loss": 0.647,
+      "step": 55190
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003540868421052631,
+      "loss": 0.659,
+      "step": 55200
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00035400789473684206,
+      "loss": 0.6417,
+      "step": 55210
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000353928947368421,
+      "loss": 0.6509,
+      "step": 55220
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00035384999999999995,
+      "loss": 0.6471,
+      "step": 55230
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00035377105263157896,
+      "loss": 0.6479,
+      "step": 55240
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00035369210526315785,
+      "loss": 0.6385,
+      "step": 55250
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00035361315789473685,
+      "loss": 0.6525,
+      "step": 55260
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00035353421052631575,
+      "loss": 0.6489,
+      "step": 55270
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003534552631578947,
+      "loss": 0.6523,
+      "step": 55280
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00035337631578947364,
+      "loss": 0.657,
+      "step": 55290
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003532973684210526,
+      "loss": 0.6315,
+      "step": 55300
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00035321842105263154,
+      "loss": 0.6426,
+      "step": 55310
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003531394736842105,
+      "loss": 0.6449,
+      "step": 55320
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003530605263157895,
+      "loss": 0.6538,
+      "step": 55330
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003529815789473684,
+      "loss": 0.6431,
+      "step": 55340
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00035290263157894733,
+      "loss": 0.6439,
+      "step": 55350
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003528236842105263,
+      "loss": 0.6624,
+      "step": 55360
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00035274473684210523,
+      "loss": 0.6426,
+      "step": 55370
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003526657894736842,
+      "loss": 0.6441,
+      "step": 55380
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00035258684210526313,
+      "loss": 0.6522,
+      "step": 55390
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000352507894736842,
+      "loss": 0.6543,
+      "step": 55400
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000352428947368421,
+      "loss": 0.6452,
+      "step": 55410
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00035234999999999997,
+      "loss": 0.6691,
+      "step": 55420
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003522710526315789,
+      "loss": 0.6552,
+      "step": 55430
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00035219210526315787,
+      "loss": 0.6545,
+      "step": 55440
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003521131578947368,
+      "loss": 0.6435,
+      "step": 55450
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00035203421052631577,
+      "loss": 0.6458,
+      "step": 55460
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00035195526315789466,
+      "loss": 0.6472,
+      "step": 55470
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00035187631578947366,
+      "loss": 0.6469,
+      "step": 55480
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00035179736842105256,
+      "loss": 0.6351,
+      "step": 55490
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00035171842105263156,
+      "loss": 0.6439,
+      "step": 55500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003516394736842105,
+      "loss": 0.6513,
+      "step": 55510
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00035156052631578946,
+      "loss": 0.6383,
+      "step": 55520
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003514815789473684,
+      "loss": 0.6356,
+      "step": 55530
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003514026315789473,
+      "loss": 0.6444,
+      "step": 55540
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003513236842105263,
+      "loss": 0.6421,
+      "step": 55550
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003512447368421052,
+      "loss": 0.6535,
+      "step": 55560
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003511657894736842,
+      "loss": 0.6477,
+      "step": 55570
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003510868421052631,
+      "loss": 0.6432,
+      "step": 55580
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003510078947368421,
+      "loss": 0.6337,
+      "step": 55590
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00035092894736842104,
+      "loss": 0.6365,
+      "step": 55600
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00035085,
+      "loss": 0.6457,
+      "step": 55610
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00035077105263157894,
+      "loss": 0.6544,
+      "step": 55620
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00035069210526315783,
+      "loss": 0.6516,
+      "step": 55630
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00035061315789473683,
+      "loss": 0.6408,
+      "step": 55640
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00035053421052631573,
+      "loss": 0.6473,
+      "step": 55650
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00035045526315789473,
+      "loss": 0.6427,
+      "step": 55660
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003503763157894736,
+      "loss": 0.6376,
+      "step": 55670
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00035029736842105263,
+      "loss": 0.6378,
+      "step": 55680
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003502184210526316,
+      "loss": 0.6464,
+      "step": 55690
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00035013947368421047,
+      "loss": 0.6271,
+      "step": 55700
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003500605263157895,
+      "loss": 0.6344,
+      "step": 55710
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00034998157894736837,
+      "loss": 0.648,
+      "step": 55720
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00034990263157894737,
+      "loss": 0.643,
+      "step": 55730
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00034982368421052626,
+      "loss": 0.631,
+      "step": 55740
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00034974473684210527,
+      "loss": 0.6502,
+      "step": 55750
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00034966578947368416,
+      "loss": 0.6507,
+      "step": 55760
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003495868421052631,
+      "loss": 0.6426,
+      "step": 55770
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00034950789473684206,
+      "loss": 0.6424,
+      "step": 55780
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000349428947368421,
+      "loss": 0.6392,
+      "step": 55790
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00034935,
+      "loss": 0.6421,
+      "step": 55800
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003492710526315789,
+      "loss": 0.6324,
+      "step": 55810
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003491921052631579,
+      "loss": 0.6444,
+      "step": 55820
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003491131578947368,
+      "loss": 0.6447,
+      "step": 55830
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00034903421052631575,
+      "loss": 0.6435,
+      "step": 55840
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003489552631578947,
+      "loss": 0.644,
+      "step": 55850
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00034887631578947364,
+      "loss": 0.646,
+      "step": 55860
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003487973684210526,
+      "loss": 0.6475,
+      "step": 55870
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00034871842105263154,
+      "loss": 0.6389,
+      "step": 55880
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00034863947368421054,
+      "loss": 0.6506,
+      "step": 55890
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00034856052631578944,
+      "loss": 0.6348,
+      "step": 55900
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003484815789473684,
+      "loss": 0.6432,
+      "step": 55910
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00034840263157894733,
+      "loss": 0.6391,
+      "step": 55920
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003483236842105263,
+      "loss": 0.6329,
+      "step": 55930
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00034824473684210523,
+      "loss": 0.6301,
+      "step": 55940
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003481657894736842,
+      "loss": 0.6469,
+      "step": 55950
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00034808684210526307,
+      "loss": 0.6357,
+      "step": 55960
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003480078947368421,
+      "loss": 0.6221,
+      "step": 55970
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003479289473684211,
+      "loss": 0.6339,
+      "step": 55980
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00034784999999999997,
+      "loss": 0.6337,
+      "step": 55990
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003477710526315789,
+      "loss": 0.6332,
+      "step": 56000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00034769210526315787,
+      "loss": 0.6302,
+      "step": 56010
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003476131578947368,
+      "loss": 0.6371,
+      "step": 56020
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003475342105263157,
+      "loss": 0.6307,
+      "step": 56030
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003474552631578947,
+      "loss": 0.639,
+      "step": 56040
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003473763157894736,
+      "loss": 0.6372,
+      "step": 56050
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003472973684210526,
+      "loss": 0.6372,
+      "step": 56060
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00034721842105263156,
+      "loss": 0.63,
+      "step": 56070
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003471394736842105,
+      "loss": 0.632,
+      "step": 56080
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00034706052631578945,
+      "loss": 0.6431,
+      "step": 56090
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003469815789473684,
+      "loss": 0.6313,
+      "step": 56100
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00034690263157894735,
+      "loss": 0.6261,
+      "step": 56110
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00034682368421052625,
+      "loss": 0.6483,
+      "step": 56120
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00034674473684210525,
+      "loss": 0.6419,
+      "step": 56130
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00034666578947368414,
+      "loss": 0.6399,
+      "step": 56140
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00034658684210526314,
+      "loss": 0.6251,
+      "step": 56150
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003465078947368421,
+      "loss": 0.6423,
+      "step": 56160
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00034642894736842104,
+      "loss": 0.6186,
+      "step": 56170
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00034635,
+      "loss": 0.641,
+      "step": 56180
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003462710526315789,
+      "loss": 0.6452,
+      "step": 56190
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003461921052631579,
+      "loss": 0.6445,
+      "step": 56200
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003461131578947368,
+      "loss": 0.6446,
+      "step": 56210
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003460342105263158,
+      "loss": 0.629,
+      "step": 56220
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003459552631578947,
+      "loss": 0.644,
+      "step": 56230
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003458763157894737,
+      "loss": 0.6362,
+      "step": 56240
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00034579736842105263,
+      "loss": 0.6241,
+      "step": 56250
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003457184210526315,
+      "loss": 0.6325,
+      "step": 56260
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003456394736842105,
+      "loss": 0.6376,
+      "step": 56270
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003455605263157894,
+      "loss": 0.6394,
+      "step": 56280
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003454815789473684,
+      "loss": 0.6292,
+      "step": 56290
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003454026315789473,
+      "loss": 0.6353,
+      "step": 56300
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003453236842105263,
+      "loss": 0.6216,
+      "step": 56310
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003452447368421052,
+      "loss": 0.6399,
+      "step": 56320
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00034516578947368416,
+      "loss": 0.6352,
+      "step": 56330
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003450868421052631,
+      "loss": 0.6526,
+      "step": 56340
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003450157894736842,
+      "loss": 0.6381,
+      "step": 56350
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00034493684210526313,
+      "loss": 0.6281,
+      "step": 56360
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003448578947368421,
+      "loss": 0.6319,
+      "step": 56370
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000344778947368421,
+      "loss": 0.6304,
+      "step": 56380
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003447,
+      "loss": 0.6291,
+      "step": 56390
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00034462105263157887,
+      "loss": 0.6208,
+      "step": 56400
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00034454210526315787,
+      "loss": 0.6203,
+      "step": 56410
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003444631578947368,
+      "loss": 0.6251,
+      "step": 56420
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00034438421052631577,
+      "loss": 0.6147,
+      "step": 56430
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003443052631578947,
+      "loss": 0.6122,
+      "step": 56440
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00034422631578947367,
+      "loss": 0.6184,
+      "step": 56450
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003441473684210526,
+      "loss": 0.6262,
+      "step": 56460
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003440684210526315,
+      "loss": 0.6301,
+      "step": 56470
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003439894736842105,
+      "loss": 0.61,
+      "step": 56480
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003439105263157894,
+      "loss": 0.6155,
+      "step": 56490
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003438315789473684,
+      "loss": 0.6194,
+      "step": 56500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00034375263157894736,
+      "loss": 0.6331,
+      "step": 56510
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003436736842105263,
+      "loss": 0.6275,
+      "step": 56520
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00034359473684210525,
+      "loss": 0.6382,
+      "step": 56530
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00034351578947368415,
+      "loss": 0.61,
+      "step": 56540
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00034343684210526315,
+      "loss": 0.6213,
+      "step": 56550
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00034335789473684204,
+      "loss": 0.6242,
+      "step": 56560
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00034327894736842105,
+      "loss": 0.6303,
+      "step": 56570
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00034319999999999994,
+      "loss": 0.6167,
+      "step": 56580
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00034312105263157894,
+      "loss": 0.6226,
+      "step": 56590
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003430421052631579,
+      "loss": 0.6151,
+      "step": 56600
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003429631578947368,
+      "loss": 0.6189,
+      "step": 56610
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003428842105263158,
+      "loss": 0.6126,
+      "step": 56620
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003428052631578947,
+      "loss": 0.6307,
+      "step": 56630
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003427263157894737,
+      "loss": 0.6203,
+      "step": 56640
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003426473684210526,
+      "loss": 0.6186,
+      "step": 56650
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003425684210526316,
+      "loss": 0.6192,
+      "step": 56660
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003424894736842105,
+      "loss": 0.6345,
+      "step": 56670
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003424105263157894,
+      "loss": 0.6189,
+      "step": 56680
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003423315789473684,
+      "loss": 0.6174,
+      "step": 56690
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003422526315789473,
+      "loss": 0.6186,
+      "step": 56700
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003421736842105263,
+      "loss": 0.6288,
+      "step": 56710
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003420947368421052,
+      "loss": 0.61,
+      "step": 56720
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003420157894736842,
+      "loss": 0.6311,
+      "step": 56730
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003419368421052631,
+      "loss": 0.6098,
+      "step": 56740
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00034185789473684206,
+      "loss": 0.6229,
+      "step": 56750
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000341778947368421,
+      "loss": 0.6144,
+      "step": 56760
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00034169999999999996,
+      "loss": 0.6202,
+      "step": 56770
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003416210526315789,
+      "loss": 0.6269,
+      "step": 56780
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00034154210526315785,
+      "loss": 0.6207,
+      "step": 56790
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00034146315789473686,
+      "loss": 0.6011,
+      "step": 56800
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00034138421052631575,
+      "loss": 0.6114,
+      "step": 56810
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00034130526315789475,
+      "loss": 0.6172,
+      "step": 56820
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00034122631578947365,
+      "loss": 0.619,
+      "step": 56830
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003411473684210526,
+      "loss": 0.6259,
+      "step": 56840
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00034106842105263154,
+      "loss": 0.6353,
+      "step": 56850
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003409894736842105,
+      "loss": 0.6145,
+      "step": 56860
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003409105263157894,
+      "loss": 0.6306,
+      "step": 56870
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003408315789473684,
+      "loss": 0.6264,
+      "step": 56880
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003407526315789474,
+      "loss": 0.636,
+      "step": 56890
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003406736842105263,
+      "loss": 0.6438,
+      "step": 56900
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00034059473684210523,
+      "loss": 0.6417,
+      "step": 56910
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003405157894736842,
+      "loss": 0.6303,
+      "step": 56920
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00034043684210526313,
+      "loss": 0.6296,
+      "step": 56930
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003403578947368421,
+      "loss": 0.6394,
+      "step": 56940
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000340278947368421,
+      "loss": 0.6436,
+      "step": 56950
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003401999999999999,
+      "loss": 0.6471,
+      "step": 56960
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003401210526315789,
+      "loss": 0.645,
+      "step": 56970
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00034004210526315787,
+      "loss": 0.6309,
+      "step": 56980
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003399631578947368,
+      "loss": 0.6183,
+      "step": 56990
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00033988421052631577,
+      "loss": 0.6312,
+      "step": 57000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003398052631578947,
+      "loss": 0.6371,
+      "step": 57010
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00033972631578947366,
+      "loss": 0.6384,
+      "step": 57020
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00033964736842105256,
+      "loss": 0.6484,
+      "step": 57030
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00033956842105263156,
+      "loss": 0.6298,
+      "step": 57040
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00033948947368421046,
+      "loss": 0.6307,
+      "step": 57050
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00033941052631578946,
+      "loss": 0.6307,
+      "step": 57060
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003393315789473684,
+      "loss": 0.6244,
+      "step": 57070
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00033925263157894735,
+      "loss": 0.631,
+      "step": 57080
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003391736842105263,
+      "loss": 0.638,
+      "step": 57090
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003390947368421052,
+      "loss": 0.6357,
+      "step": 57100
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003390157894736842,
+      "loss": 0.6357,
+      "step": 57110
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003389368421052631,
+      "loss": 0.6268,
+      "step": 57120
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003388578947368421,
+      "loss": 0.6304,
+      "step": 57130
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000338778947368421,
+      "loss": 0.6377,
+      "step": 57140
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003387,
+      "loss": 0.6362,
+      "step": 57150
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00033862105263157894,
+      "loss": 0.6273,
+      "step": 57160
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00033854210526315784,
+      "loss": 0.6422,
+      "step": 57170
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00033846315789473684,
+      "loss": 0.6449,
+      "step": 57180
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00033838421052631573,
+      "loss": 0.6322,
+      "step": 57190
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00033830526315789473,
+      "loss": 0.6326,
+      "step": 57200
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00033822631578947363,
+      "loss": 0.6288,
+      "step": 57210
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00033814736842105263,
+      "loss": 0.6461,
+      "step": 57220
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003380684210526315,
+      "loss": 0.637,
+      "step": 57230
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003379894736842105,
+      "loss": 0.637,
+      "step": 57240
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003379105263157895,
+      "loss": 0.6544,
+      "step": 57250
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00033783157894736837,
+      "loss": 0.6205,
+      "step": 57260
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00033775263157894737,
+      "loss": 0.6239,
+      "step": 57270
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00033767368421052627,
+      "loss": 0.6405,
+      "step": 57280
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00033759473684210527,
+      "loss": 0.6491,
+      "step": 57290
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00033751578947368416,
+      "loss": 0.636,
+      "step": 57300
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00033743684210526317,
+      "loss": 0.633,
+      "step": 57310
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00033735789473684206,
+      "loss": 0.6305,
+      "step": 57320
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000337278947368421,
+      "loss": 0.634,
+      "step": 57330
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003372,
+      "loss": 0.6291,
+      "step": 57340
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003371210526315789,
+      "loss": 0.6426,
+      "step": 57350
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003370421052631579,
+      "loss": 0.6482,
+      "step": 57360
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003369631578947368,
+      "loss": 0.6401,
+      "step": 57370
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003368842105263158,
+      "loss": 0.6438,
+      "step": 57380
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003368052631578947,
+      "loss": 0.6434,
+      "step": 57390
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00033672631578947365,
+      "loss": 0.6377,
+      "step": 57400
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003366473684210526,
+      "loss": 0.6433,
+      "step": 57410
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00033656842105263154,
+      "loss": 0.6356,
+      "step": 57420
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003364894736842105,
+      "loss": 0.6279,
+      "step": 57430
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00033641052631578944,
+      "loss": 0.634,
+      "step": 57440
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00033633157894736844,
+      "loss": 0.6353,
+      "step": 57450
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00033625263157894734,
+      "loss": 0.6357,
+      "step": 57460
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003361736842105263,
+      "loss": 0.6419,
+      "step": 57470
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00033609473684210523,
+      "loss": 0.6388,
+      "step": 57480
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003360157894736842,
+      "loss": 0.6252,
+      "step": 57490
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00033593684210526313,
+      "loss": 0.6428,
+      "step": 57500
+    },
+    {
+      "epoch": 0.57,
+      "eval_accuracy": 0.8676777163075269,
+      "eval_loss": 0.61767578125,
+      "eval_runtime": 96.8045,
+      "eval_samples_per_second": 826.408,
+      "eval_steps_per_second": 1.622,
+      "step": 57500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003358578947368421,
+      "loss": 0.6304,
+      "step": 57510
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00033577894736842097,
+      "loss": 0.6299,
+      "step": 57520
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003357,
+      "loss": 0.6365,
+      "step": 57530
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003356210526315789,
+      "loss": 0.6266,
+      "step": 57540
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00033554210526315787,
+      "loss": 0.6205,
+      "step": 57550
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003354631578947368,
+      "loss": 0.6284,
+      "step": 57560
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00033538421052631577,
+      "loss": 0.6286,
+      "step": 57570
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003353052631578947,
+      "loss": 0.6389,
+      "step": 57580
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003352263157894736,
+      "loss": 0.6443,
+      "step": 57590
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003351473684210526,
+      "loss": 0.6421,
+      "step": 57600
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003350684210526315,
+      "loss": 0.6325,
+      "step": 57610
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003349894736842105,
+      "loss": 0.6377,
+      "step": 57620
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00033491052631578946,
+      "loss": 0.6283,
+      "step": 57630
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003348315789473684,
+      "loss": 0.6312,
+      "step": 57640
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00033475263157894735,
+      "loss": 0.6421,
+      "step": 57650
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00033467368421052625,
+      "loss": 0.6334,
+      "step": 57660
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00033459473684210525,
+      "loss": 0.6284,
+      "step": 57670
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00033451578947368414,
+      "loss": 0.6297,
+      "step": 57680
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00033443684210526315,
+      "loss": 0.6359,
+      "step": 57690
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00033435789473684204,
+      "loss": 0.6484,
+      "step": 57700
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00033427894736842104,
+      "loss": 0.6333,
+      "step": 57710
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003342,
+      "loss": 0.6478,
+      "step": 57720
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003341210526315789,
+      "loss": 0.6257,
+      "step": 57730
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003340421052631579,
+      "loss": 0.6177,
+      "step": 57740
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003339631578947368,
+      "loss": 0.6136,
+      "step": 57750
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003338842105263158,
+      "loss": 0.6088,
+      "step": 57760
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003338052631578947,
+      "loss": 0.6189,
+      "step": 57770
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003337263157894737,
+      "loss": 0.6258,
+      "step": 57780
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003336473684210526,
+      "loss": 0.623,
+      "step": 57790
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003335684210526316,
+      "loss": 0.6204,
+      "step": 57800
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003334894736842105,
+      "loss": 0.6175,
+      "step": 57810
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003334105263157894,
+      "loss": 0.6306,
+      "step": 57820
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003333315789473684,
+      "loss": 0.641,
+      "step": 57830
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003332526315789473,
+      "loss": 0.6432,
+      "step": 57840
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003331736842105263,
+      "loss": 0.6333,
+      "step": 57850
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003330947368421052,
+      "loss": 0.6337,
+      "step": 57860
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003330157894736842,
+      "loss": 0.6331,
+      "step": 57870
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003329368421052631,
+      "loss": 0.6361,
+      "step": 57880
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00033285789473684206,
+      "loss": 0.6286,
+      "step": 57890
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00033277894736842106,
+      "loss": 0.6331,
+      "step": 57900
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00033269999999999996,
+      "loss": 0.6294,
+      "step": 57910
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00033262105263157896,
+      "loss": 0.6184,
+      "step": 57920
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00033254210526315785,
+      "loss": 0.6229,
+      "step": 57930
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00033246315789473685,
+      "loss": 0.6278,
+      "step": 57940
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00033238421052631575,
+      "loss": 0.6325,
+      "step": 57950
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003323052631578947,
+      "loss": 0.6175,
+      "step": 57960
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00033222631578947365,
+      "loss": 0.6318,
+      "step": 57970
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003321473684210526,
+      "loss": 0.6412,
+      "step": 57980
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00033206842105263154,
+      "loss": 0.6204,
+      "step": 57990
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003319894736842105,
+      "loss": 0.6178,
+      "step": 58000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003319105263157895,
+      "loss": 0.631,
+      "step": 58010
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003318315789473684,
+      "loss": 0.6335,
+      "step": 58020
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00033175263157894734,
+      "loss": 0.629,
+      "step": 58030
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003316736842105263,
+      "loss": 0.6225,
+      "step": 58040
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00033159473684210523,
+      "loss": 0.6384,
+      "step": 58050
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003315157894736842,
+      "loss": 0.6271,
+      "step": 58060
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00033143684210526313,
+      "loss": 0.6272,
+      "step": 58070
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000331357894736842,
+      "loss": 0.6344,
+      "step": 58080
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000331278947368421,
+      "loss": 0.6287,
+      "step": 58090
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003312,
+      "loss": 0.6213,
+      "step": 58100
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003311210526315789,
+      "loss": 0.6273,
+      "step": 58110
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00033104210526315787,
+      "loss": 0.63,
+      "step": 58120
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003309631578947368,
+      "loss": 0.6309,
+      "step": 58130
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00033088421052631577,
+      "loss": 0.6349,
+      "step": 58140
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00033080526315789466,
+      "loss": 0.6352,
+      "step": 58150
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00033072631578947366,
+      "loss": 0.6297,
+      "step": 58160
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00033064736842105256,
+      "loss": 0.6297,
+      "step": 58170
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00033056842105263156,
+      "loss": 0.6383,
+      "step": 58180
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003304894736842105,
+      "loss": 0.6287,
+      "step": 58190
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00033041052631578946,
+      "loss": 0.6298,
+      "step": 58200
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003303315789473684,
+      "loss": 0.6346,
+      "step": 58210
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003302526315789473,
+      "loss": 0.6338,
+      "step": 58220
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003301736842105263,
+      "loss": 0.6466,
+      "step": 58230
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003300947368421052,
+      "loss": 0.6371,
+      "step": 58240
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003300157894736842,
+      "loss": 0.6309,
+      "step": 58250
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003299368421052631,
+      "loss": 0.6383,
+      "step": 58260
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003298578947368421,
+      "loss": 0.6353,
+      "step": 58270
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00032977894736842104,
+      "loss": 0.6262,
+      "step": 58280
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003297,
+      "loss": 0.6326,
+      "step": 58290
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00032962105263157894,
+      "loss": 0.6334,
+      "step": 58300
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00032954210526315783,
+      "loss": 0.6404,
+      "step": 58310
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00032946315789473684,
+      "loss": 0.6286,
+      "step": 58320
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00032938421052631573,
+      "loss": 0.6268,
+      "step": 58330
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00032930526315789473,
+      "loss": 0.6331,
+      "step": 58340
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00032923421052631575,
+      "loss": 0.6318,
+      "step": 58350
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00032915526315789476,
+      "loss": 0.6397,
+      "step": 58360
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00032907631578947365,
+      "loss": 0.6216,
+      "step": 58370
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003289973684210526,
+      "loss": 0.6413,
+      "step": 58380
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00032891842105263155,
+      "loss": 0.6298,
+      "step": 58390
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003288394736842105,
+      "loss": 0.6345,
+      "step": 58400
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00032876052631578944,
+      "loss": 0.6322,
+      "step": 58410
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003286815789473684,
+      "loss": 0.638,
+      "step": 58420
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003286026315789474,
+      "loss": 0.6288,
+      "step": 58430
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003285236842105263,
+      "loss": 0.6297,
+      "step": 58440
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00032844473684210524,
+      "loss": 0.6264,
+      "step": 58450
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003283657894736842,
+      "loss": 0.6328,
+      "step": 58460
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00032828684210526313,
+      "loss": 0.6424,
+      "step": 58470
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003282078947368421,
+      "loss": 0.6358,
+      "step": 58480
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00032812894736842103,
+      "loss": 0.6158,
+      "step": 58490
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003280499999999999,
+      "loss": 0.6232,
+      "step": 58500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003279710526315789,
+      "loss": 0.6338,
+      "step": 58510
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003278921052631578,
+      "loss": 0.6393,
+      "step": 58520
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003278131578947368,
+      "loss": 0.6383,
+      "step": 58530
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00032773421052631577,
+      "loss": 0.6386,
+      "step": 58540
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003276552631578947,
+      "loss": 0.6394,
+      "step": 58550
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00032757631578947367,
+      "loss": 0.6503,
+      "step": 58560
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00032749736842105256,
+      "loss": 0.6437,
+      "step": 58570
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00032741842105263156,
+      "loss": 0.6389,
+      "step": 58580
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003273473684210526,
+      "loss": 0.6397,
+      "step": 58590
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003272684210526316,
+      "loss": 0.6371,
+      "step": 58600
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003271894736842105,
+      "loss": 0.6372,
+      "step": 58610
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003271105263157895,
+      "loss": 0.6464,
+      "step": 58620
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003270315789473684,
+      "loss": 0.6482,
+      "step": 58630
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003269526315789474,
+      "loss": 0.6394,
+      "step": 58640
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003268736842105263,
+      "loss": 0.6346,
+      "step": 58650
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003267947368421052,
+      "loss": 0.6476,
+      "step": 58660
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00032671578947368417,
+      "loss": 0.6379,
+      "step": 58670
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003266368421052631,
+      "loss": 0.6384,
+      "step": 58680
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003265578947368421,
+      "loss": 0.6229,
+      "step": 58690
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000326478947368421,
+      "loss": 0.6252,
+      "step": 58700
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003264,
+      "loss": 0.6342,
+      "step": 58710
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003263210526315789,
+      "loss": 0.6275,
+      "step": 58720
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00032624210526315786,
+      "loss": 0.6402,
+      "step": 58730
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003261631578947368,
+      "loss": 0.6292,
+      "step": 58740
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00032608421052631576,
+      "loss": 0.6141,
+      "step": 58750
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003260052631578947,
+      "loss": 0.6227,
+      "step": 58760
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00032592631578947365,
+      "loss": 0.636,
+      "step": 58770
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00032584736842105266,
+      "loss": 0.6358,
+      "step": 58780
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00032576842105263155,
+      "loss": 0.6214,
+      "step": 58790
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003256894736842105,
+      "loss": 0.6314,
+      "step": 58800
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00032561052631578945,
+      "loss": 0.6404,
+      "step": 58810
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003255315789473684,
+      "loss": 0.6314,
+      "step": 58820
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00032545263157894734,
+      "loss": 0.6287,
+      "step": 58830
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003253736842105263,
+      "loss": 0.635,
+      "step": 58840
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003252947368421052,
+      "loss": 0.632,
+      "step": 58850
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003252157894736842,
+      "loss": 0.6224,
+      "step": 58860
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00032513684210526314,
+      "loss": 0.6334,
+      "step": 58870
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003250578947368421,
+      "loss": 0.6179,
+      "step": 58880
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00032497894736842103,
+      "loss": 0.6306,
+      "step": 58890
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003249,
+      "loss": 0.6364,
+      "step": 58900
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00032482105263157893,
+      "loss": 0.6397,
+      "step": 58910
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003247421052631578,
+      "loss": 0.6311,
+      "step": 58920
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00032466315789473683,
+      "loss": 0.6396,
+      "step": 58930
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003245842105263157,
+      "loss": 0.6295,
+      "step": 58940
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003245052631578947,
+      "loss": 0.6268,
+      "step": 58950
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003244263157894736,
+      "loss": 0.6296,
+      "step": 58960
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003243473684210526,
+      "loss": 0.6345,
+      "step": 58970
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00032426842105263157,
+      "loss": 0.6227,
+      "step": 58980
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00032418947368421046,
+      "loss": 0.6243,
+      "step": 58990
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00032411052631578947,
+      "loss": 0.6189,
+      "step": 59000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00032403157894736836,
+      "loss": 0.6153,
+      "step": 59010
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00032395263157894736,
+      "loss": 0.6217,
+      "step": 59020
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00032387368421052626,
+      "loss": 0.6358,
+      "step": 59030
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00032379473684210526,
+      "loss": 0.6353,
+      "step": 59040
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00032371578947368415,
+      "loss": 0.6306,
+      "step": 59050
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003236368421052631,
+      "loss": 0.627,
+      "step": 59060
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003235578947368421,
+      "loss": 0.6333,
+      "step": 59070
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000323478947368421,
+      "loss": 0.6257,
+      "step": 59080
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003234,
+      "loss": 0.6248,
+      "step": 59090
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003233210526315789,
+      "loss": 0.6364,
+      "step": 59100
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003232421052631579,
+      "loss": 0.6307,
+      "step": 59110
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003231631578947368,
+      "loss": 0.622,
+      "step": 59120
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003230842105263158,
+      "loss": 0.6147,
+      "step": 59130
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003230052631578947,
+      "loss": 0.6204,
+      "step": 59140
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00032292631578947364,
+      "loss": 0.6294,
+      "step": 59150
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00032284736842105264,
+      "loss": 0.6309,
+      "step": 59160
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00032276842105263153,
+      "loss": 0.6282,
+      "step": 59170
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00032268947368421054,
+      "loss": 0.6147,
+      "step": 59180
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00032261052631578943,
+      "loss": 0.6206,
+      "step": 59190
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00032253157894736843,
+      "loss": 0.6234,
+      "step": 59200
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003224526315789473,
+      "loss": 0.6262,
+      "step": 59210
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003223736842105263,
+      "loss": 0.6348,
+      "step": 59220
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003222947368421052,
+      "loss": 0.6257,
+      "step": 59230
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00032221578947368417,
+      "loss": 0.6284,
+      "step": 59240
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003221368421052632,
+      "loss": 0.6237,
+      "step": 59250
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00032205789473684207,
+      "loss": 0.6259,
+      "step": 59260
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00032197894736842107,
+      "loss": 0.6308,
+      "step": 59270
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00032189999999999996,
+      "loss": 0.6322,
+      "step": 59280
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003218210526315789,
+      "loss": 0.6329,
+      "step": 59290
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00032174210526315786,
+      "loss": 0.6196,
+      "step": 59300
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003216631578947368,
+      "loss": 0.619,
+      "step": 59310
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00032158421052631576,
+      "loss": 0.6221,
+      "step": 59320
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003215052631578947,
+      "loss": 0.6213,
+      "step": 59330
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003214263157894737,
+      "loss": 0.6261,
+      "step": 59340
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003213473684210526,
+      "loss": 0.6266,
+      "step": 59350
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00032126842105263155,
+      "loss": 0.6212,
+      "step": 59360
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003211894736842105,
+      "loss": 0.6213,
+      "step": 59370
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00032111052631578945,
+      "loss": 0.6246,
+      "step": 59380
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003210315789473684,
+      "loss": 0.6276,
+      "step": 59390
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00032095263157894734,
+      "loss": 0.6341,
+      "step": 59400
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00032087368421052624,
+      "loss": 0.6288,
+      "step": 59410
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00032079473684210524,
+      "loss": 0.6212,
+      "step": 59420
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003207157894736842,
+      "loss": 0.633,
+      "step": 59430
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00032063684210526314,
+      "loss": 0.6285,
+      "step": 59440
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003205578947368421,
+      "loss": 0.6212,
+      "step": 59450
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00032047894736842103,
+      "loss": 0.6343,
+      "step": 59460
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003204,
+      "loss": 0.6328,
+      "step": 59470
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003203210526315789,
+      "loss": 0.6272,
+      "step": 59480
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003202421052631579,
+      "loss": 0.6278,
+      "step": 59490
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00032016315789473677,
+      "loss": 0.6287,
+      "step": 59500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003200842105263158,
+      "loss": 0.6277,
+      "step": 59510
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00032000526315789467,
+      "loss": 0.6342,
+      "step": 59520
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00031992631578947367,
+      "loss": 0.6318,
+      "step": 59530
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003198473684210526,
+      "loss": 0.6406,
+      "step": 59540
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003197684210526315,
+      "loss": 0.6313,
+      "step": 59550
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003196894736842105,
+      "loss": 0.6356,
+      "step": 59560
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003196105263157894,
+      "loss": 0.6193,
+      "step": 59570
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003195315789473684,
+      "loss": 0.6304,
+      "step": 59580
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003194526315789473,
+      "loss": 0.6199,
+      "step": 59590
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003193736842105263,
+      "loss": 0.6353,
+      "step": 59600
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003192947368421052,
+      "loss": 0.613,
+      "step": 59610
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003192157894736842,
+      "loss": 0.6261,
+      "step": 59620
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00031913684210526315,
+      "loss": 0.6297,
+      "step": 59630
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00031905789473684205,
+      "loss": 0.6254,
+      "step": 59640
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00031897894736842105,
+      "loss": 0.6246,
+      "step": 59650
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00031889999999999995,
+      "loss": 0.6192,
+      "step": 59660
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00031882105263157895,
+      "loss": 0.6128,
+      "step": 59670
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00031874210526315784,
+      "loss": 0.6194,
+      "step": 59680
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00031866315789473684,
+      "loss": 0.6397,
+      "step": 59690
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00031858421052631574,
+      "loss": 0.635,
+      "step": 59700
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003185052631578947,
+      "loss": 0.6295,
+      "step": 59710
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003184263157894737,
+      "loss": 0.6401,
+      "step": 59720
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003183473684210526,
+      "loss": 0.6183,
+      "step": 59730
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00031827631578947366,
+      "loss": 0.6335,
+      "step": 59740
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003181973684210526,
+      "loss": 0.6204,
+      "step": 59750
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003181184210526315,
+      "loss": 0.616,
+      "step": 59760
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003180394736842105,
+      "loss": 0.6197,
+      "step": 59770
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00031796052631578945,
+      "loss": 0.6293,
+      "step": 59780
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003178815789473684,
+      "loss": 0.6145,
+      "step": 59790
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00031780263157894735,
+      "loss": 0.6298,
+      "step": 59800
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003177236842105263,
+      "loss": 0.6333,
+      "step": 59810
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00031764473684210524,
+      "loss": 0.6434,
+      "step": 59820
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00031756578947368414,
+      "loss": 0.6417,
+      "step": 59830
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00031748684210526314,
+      "loss": 0.637,
+      "step": 59840
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00031740789473684204,
+      "loss": 0.6227,
+      "step": 59850
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00031732894736842104,
+      "loss": 0.6152,
+      "step": 59860
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00031725,
+      "loss": 0.633,
+      "step": 59870
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00031717105263157893,
+      "loss": 0.6391,
+      "step": 59880
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003170921052631579,
+      "loss": 0.623,
+      "step": 59890
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003170131578947368,
+      "loss": 0.6313,
+      "step": 59900
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003169342105263158,
+      "loss": 0.6346,
+      "step": 59910
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003168552631578947,
+      "loss": 0.6354,
+      "step": 59920
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003167763157894737,
+      "loss": 0.6216,
+      "step": 59930
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00031669736842105257,
+      "loss": 0.6221,
+      "step": 59940
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00031661842105263157,
+      "loss": 0.613,
+      "step": 59950
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003165394736842105,
+      "loss": 0.6269,
+      "step": 59960
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003164605263157894,
+      "loss": 0.6301,
+      "step": 59970
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003163815789473684,
+      "loss": 0.6229,
+      "step": 59980
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003163026315789473,
+      "loss": 0.6208,
+      "step": 59990
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003162236842105263,
+      "loss": 0.6275,
+      "step": 60000
+    },
+    {
+      "epoch": 0.6,
+      "eval_accuracy": 0.8692195892541361,
+      "eval_loss": 0.61181640625,
+      "eval_runtime": 97.6328,
+      "eval_samples_per_second": 819.396,
+      "eval_steps_per_second": 1.608,
+      "step": 60000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003161447368421052,
+      "loss": 0.632,
+      "step": 60010
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003160657894736842,
+      "loss": 0.637,
+      "step": 60020
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003159868421052631,
+      "loss": 0.6263,
+      "step": 60030
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003159078947368421,
+      "loss": 0.6303,
+      "step": 60040
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000315828947368421,
+      "loss": 0.6338,
+      "step": 60050
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00031574999999999995,
+      "loss": 0.6278,
+      "step": 60060
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00031567105263157895,
+      "loss": 0.6364,
+      "step": 60070
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00031559210526315785,
+      "loss": 0.6261,
+      "step": 60080
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00031551315789473685,
+      "loss": 0.6224,
+      "step": 60090
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00031543421052631574,
+      "loss": 0.6252,
+      "step": 60100
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00031535526315789475,
+      "loss": 0.6245,
+      "step": 60110
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00031527631578947364,
+      "loss": 0.6357,
+      "step": 60120
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003151973684210526,
+      "loss": 0.6277,
+      "step": 60130
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00031511842105263154,
+      "loss": 0.6235,
+      "step": 60140
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003150394736842105,
+      "loss": 0.6275,
+      "step": 60150
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003149605263157895,
+      "loss": 0.625,
+      "step": 60160
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003148815789473684,
+      "loss": 0.6379,
+      "step": 60170
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003148026315789474,
+      "loss": 0.6249,
+      "step": 60180
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003147236842105263,
+      "loss": 0.633,
+      "step": 60190
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003146447368421052,
+      "loss": 0.6364,
+      "step": 60200
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003145657894736842,
+      "loss": 0.6279,
+      "step": 60210
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003144868421052631,
+      "loss": 0.6266,
+      "step": 60220
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00031440789473684207,
+      "loss": 0.6171,
+      "step": 60230
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000314328947368421,
+      "loss": 0.6098,
+      "step": 60240
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00031425,
+      "loss": 0.6407,
+      "step": 60250
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003141710526315789,
+      "loss": 0.627,
+      "step": 60260
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00031409210526315786,
+      "loss": 0.6257,
+      "step": 60270
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003140131578947368,
+      "loss": 0.618,
+      "step": 60280
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00031393421052631576,
+      "loss": 0.6326,
+      "step": 60290
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003138552631578947,
+      "loss": 0.634,
+      "step": 60300
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00031377631578947366,
+      "loss": 0.6198,
+      "step": 60310
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00031369736842105255,
+      "loss": 0.6212,
+      "step": 60320
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00031361842105263155,
+      "loss": 0.6295,
+      "step": 60330
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003135394736842105,
+      "loss": 0.6275,
+      "step": 60340
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00031346052631578945,
+      "loss": 0.6225,
+      "step": 60350
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003133815789473684,
+      "loss": 0.6308,
+      "step": 60360
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00031330263157894735,
+      "loss": 0.6313,
+      "step": 60370
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003132236842105263,
+      "loss": 0.6352,
+      "step": 60380
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003131447368421052,
+      "loss": 0.6164,
+      "step": 60390
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003130657894736842,
+      "loss": 0.623,
+      "step": 60400
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003129868421052631,
+      "loss": 0.6265,
+      "step": 60410
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003129078947368421,
+      "loss": 0.6195,
+      "step": 60420
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00031282894736842104,
+      "loss": 0.6345,
+      "step": 60430
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00031275,
+      "loss": 0.6402,
+      "step": 60440
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00031267105263157893,
+      "loss": 0.6309,
+      "step": 60450
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00031259210526315783,
+      "loss": 0.609,
+      "step": 60460
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00031251315789473683,
+      "loss": 0.6225,
+      "step": 60470
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003124342105263157,
+      "loss": 0.6204,
+      "step": 60480
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003123552631578947,
+      "loss": 0.6171,
+      "step": 60490
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003122763157894736,
+      "loss": 0.6381,
+      "step": 60500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003121973684210526,
+      "loss": 0.6226,
+      "step": 60510
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00031211842105263157,
+      "loss": 0.6345,
+      "step": 60520
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003120394736842105,
+      "loss": 0.6231,
+      "step": 60530
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00031196052631578947,
+      "loss": 0.6288,
+      "step": 60540
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00031188157894736836,
+      "loss": 0.6224,
+      "step": 60550
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00031180263157894737,
+      "loss": 0.6392,
+      "step": 60560
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00031172368421052626,
+      "loss": 0.6262,
+      "step": 60570
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00031164473684210526,
+      "loss": 0.6266,
+      "step": 60580
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00031156578947368416,
+      "loss": 0.6389,
+      "step": 60590
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00031148684210526316,
+      "loss": 0.64,
+      "step": 60600
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00031140789473684205,
+      "loss": 0.6286,
+      "step": 60610
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000311328947368421,
+      "loss": 0.6186,
+      "step": 60620
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00031125,
+      "loss": 0.6166,
+      "step": 60630
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003111710526315789,
+      "loss": 0.6251,
+      "step": 60640
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003110921052631579,
+      "loss": 0.6116,
+      "step": 60650
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003110131578947368,
+      "loss": 0.6275,
+      "step": 60660
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003109342105263158,
+      "loss": 0.6184,
+      "step": 60670
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003108552631578947,
+      "loss": 0.6228,
+      "step": 60680
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00031077631578947364,
+      "loss": 0.613,
+      "step": 60690
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003106973684210526,
+      "loss": 0.6364,
+      "step": 60700
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00031061842105263154,
+      "loss": 0.6166,
+      "step": 60710
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00031053947368421054,
+      "loss": 0.6254,
+      "step": 60720
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00031046052631578943,
+      "loss": 0.6301,
+      "step": 60730
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00031038157894736843,
+      "loss": 0.6314,
+      "step": 60740
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00031030263157894733,
+      "loss": 0.6388,
+      "step": 60750
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003102236842105263,
+      "loss": 0.6364,
+      "step": 60760
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003101447368421052,
+      "loss": 0.6293,
+      "step": 60770
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003100657894736842,
+      "loss": 0.6234,
+      "step": 60780
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003099868421052631,
+      "loss": 0.6111,
+      "step": 60790
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00030990789473684207,
+      "loss": 0.6188,
+      "step": 60800
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00030982894736842107,
+      "loss": 0.6154,
+      "step": 60810
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00030974999999999997,
+      "loss": 0.6339,
+      "step": 60820
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003096710526315789,
+      "loss": 0.6194,
+      "step": 60830
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00030959210526315786,
+      "loss": 0.6195,
+      "step": 60840
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003095131578947368,
+      "loss": 0.6186,
+      "step": 60850
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00030943421052631576,
+      "loss": 0.6165,
+      "step": 60860
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003093552631578947,
+      "loss": 0.6314,
+      "step": 60870
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003092763157894736,
+      "loss": 0.6307,
+      "step": 60880
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003091973684210526,
+      "loss": 0.6341,
+      "step": 60890
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003091184210526316,
+      "loss": 0.6156,
+      "step": 60900
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003090394736842105,
+      "loss": 0.6256,
+      "step": 60910
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00030896052631578945,
+      "loss": 0.63,
+      "step": 60920
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003088815789473684,
+      "loss": 0.632,
+      "step": 60930
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00030880263157894735,
+      "loss": 0.6306,
+      "step": 60940
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00030872368421052624,
+      "loss": 0.6324,
+      "step": 60950
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00030864473684210524,
+      "loss": 0.6291,
+      "step": 60960
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00030856578947368414,
+      "loss": 0.6204,
+      "step": 60970
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00030848684210526314,
+      "loss": 0.6199,
+      "step": 60980
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003084078947368421,
+      "loss": 0.6295,
+      "step": 60990
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00030832894736842104,
+      "loss": 0.6233,
+      "step": 61000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00030825,
+      "loss": 0.6229,
+      "step": 61010
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00030817105263157893,
+      "loss": 0.6079,
+      "step": 61020
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003080921052631579,
+      "loss": 0.6135,
+      "step": 61030
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003080131578947368,
+      "loss": 0.6028,
+      "step": 61040
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003079342105263158,
+      "loss": 0.6226,
+      "step": 61050
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00030785526315789467,
+      "loss": 0.6309,
+      "step": 61060
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003077763157894737,
+      "loss": 0.6358,
+      "step": 61070
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003076973684210526,
+      "loss": 0.6278,
+      "step": 61080
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00030761842105263157,
+      "loss": 0.6392,
+      "step": 61090
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003075394736842105,
+      "loss": 0.6313,
+      "step": 61100
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003074605263157894,
+      "loss": 0.6166,
+      "step": 61110
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003073815789473684,
+      "loss": 0.6287,
+      "step": 61120
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003073026315789473,
+      "loss": 0.6237,
+      "step": 61130
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003072236842105263,
+      "loss": 0.6303,
+      "step": 61140
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00030715263157894733,
+      "loss": 0.6295,
+      "step": 61150
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00030707368421052634,
+      "loss": 0.6245,
+      "step": 61160
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00030699473684210523,
+      "loss": 0.6352,
+      "step": 61170
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003069157894736842,
+      "loss": 0.6223,
+      "step": 61180
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003068368421052631,
+      "loss": 0.6271,
+      "step": 61190
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003067578947368421,
+      "loss": 0.6274,
+      "step": 61200
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000306678947368421,
+      "loss": 0.6205,
+      "step": 61210
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00030659999999999997,
+      "loss": 0.6145,
+      "step": 61220
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00030652105263157887,
+      "loss": 0.6268,
+      "step": 61230
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00030644210526315787,
+      "loss": 0.6302,
+      "step": 61240
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00030636315789473687,
+      "loss": 0.6325,
+      "step": 61250
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00030628421052631576,
+      "loss": 0.6452,
+      "step": 61260
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003062052631578947,
+      "loss": 0.6215,
+      "step": 61270
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00030612631578947366,
+      "loss": 0.6305,
+      "step": 61280
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003060473684210526,
+      "loss": 0.6314,
+      "step": 61290
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003059684210526315,
+      "loss": 0.6411,
+      "step": 61300
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003058894736842105,
+      "loss": 0.639,
+      "step": 61310
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003058105263157894,
+      "loss": 0.6268,
+      "step": 61320
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003057315789473684,
+      "loss": 0.6194,
+      "step": 61330
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00030565263157894735,
+      "loss": 0.6243,
+      "step": 61340
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003055736842105263,
+      "loss": 0.6305,
+      "step": 61350
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00030549473684210525,
+      "loss": 0.6338,
+      "step": 61360
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003054157894736842,
+      "loss": 0.6212,
+      "step": 61370
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00030533684210526314,
+      "loss": 0.622,
+      "step": 61380
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00030525789473684204,
+      "loss": 0.6349,
+      "step": 61390
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00030517894736842104,
+      "loss": 0.6355,
+      "step": 61400
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00030509999999999994,
+      "loss": 0.6128,
+      "step": 61410
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00030502105263157894,
+      "loss": 0.6311,
+      "step": 61420
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003049421052631579,
+      "loss": 0.6299,
+      "step": 61430
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00030486315789473683,
+      "loss": 0.6296,
+      "step": 61440
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003047842105263158,
+      "loss": 0.6288,
+      "step": 61450
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003047052631578947,
+      "loss": 0.6322,
+      "step": 61460
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003046263157894737,
+      "loss": 0.642,
+      "step": 61470
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003045473684210526,
+      "loss": 0.6117,
+      "step": 61480
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003044684210526316,
+      "loss": 0.6149,
+      "step": 61490
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00030438947368421047,
+      "loss": 0.6307,
+      "step": 61500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00030431052631578947,
+      "loss": 0.6317,
+      "step": 61510
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003042315789473684,
+      "loss": 0.6253,
+      "step": 61520
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003041526315789473,
+      "loss": 0.6289,
+      "step": 61530
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003040736842105263,
+      "loss": 0.6247,
+      "step": 61540
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003039947368421052,
+      "loss": 0.6246,
+      "step": 61550
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003039157894736842,
+      "loss": 0.6224,
+      "step": 61560
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003038368421052631,
+      "loss": 0.6232,
+      "step": 61570
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003037578947368421,
+      "loss": 0.632,
+      "step": 61580
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000303678947368421,
+      "loss": 0.6228,
+      "step": 61590
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00030359999999999995,
+      "loss": 0.6365,
+      "step": 61600
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00030352105263157896,
+      "loss": 0.632,
+      "step": 61610
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00030344210526315785,
+      "loss": 0.6316,
+      "step": 61620
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00030336315789473685,
+      "loss": 0.639,
+      "step": 61630
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00030328421052631575,
+      "loss": 0.6287,
+      "step": 61640
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00030320526315789475,
+      "loss": 0.622,
+      "step": 61650
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00030312631578947364,
+      "loss": 0.6274,
+      "step": 61660
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003030473684210526,
+      "loss": 0.6117,
+      "step": 61670
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00030296842105263154,
+      "loss": 0.6234,
+      "step": 61680
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003028894736842105,
+      "loss": 0.6216,
+      "step": 61690
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00030281052631578944,
+      "loss": 0.6374,
+      "step": 61700
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003027315789473684,
+      "loss": 0.6377,
+      "step": 61710
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003026526315789474,
+      "loss": 0.6358,
+      "step": 61720
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003025736842105263,
+      "loss": 0.6196,
+      "step": 61730
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003024947368421053,
+      "loss": 0.6232,
+      "step": 61740
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003024157894736842,
+      "loss": 0.6274,
+      "step": 61750
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003023368421052631,
+      "loss": 0.6398,
+      "step": 61760
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003022578947368421,
+      "loss": 0.6381,
+      "step": 61770
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000302178947368421,
+      "loss": 0.6327,
+      "step": 61780
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003020999999999999,
+      "loss": 0.621,
+      "step": 61790
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003020210526315789,
+      "loss": 0.6335,
+      "step": 61800
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003019421052631579,
+      "loss": 0.6233,
+      "step": 61810
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003018631578947368,
+      "loss": 0.629,
+      "step": 61820
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00030178421052631576,
+      "loss": 0.6417,
+      "step": 61830
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003017052631578947,
+      "loss": 0.627,
+      "step": 61840
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00030162631578947366,
+      "loss": 0.6304,
+      "step": 61850
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003015473684210526,
+      "loss": 0.6376,
+      "step": 61860
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00030146842105263156,
+      "loss": 0.6327,
+      "step": 61870
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00030138947368421045,
+      "loss": 0.619,
+      "step": 61880
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00030131052631578945,
+      "loss": 0.6299,
+      "step": 61890
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003012315789473684,
+      "loss": 0.631,
+      "step": 61900
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00030115263157894735,
+      "loss": 0.6322,
+      "step": 61910
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003010736842105263,
+      "loss": 0.6164,
+      "step": 61920
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00030099473684210525,
+      "loss": 0.621,
+      "step": 61930
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003009157894736842,
+      "loss": 0.631,
+      "step": 61940
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003008368421052631,
+      "loss": 0.6438,
+      "step": 61950
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003007578947368421,
+      "loss": 0.6347,
+      "step": 61960
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000300678947368421,
+      "loss": 0.634,
+      "step": 61970
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003006,
+      "loss": 0.6197,
+      "step": 61980
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00030052105263157894,
+      "loss": 0.5923,
+      "step": 61990
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003004421052631579,
+      "loss": 0.6237,
+      "step": 62000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00030036315789473683,
+      "loss": 0.6013,
+      "step": 62010
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00030028421052631573,
+      "loss": 0.6049,
+      "step": 62020
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00030020526315789473,
+      "loss": 0.5971,
+      "step": 62030
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003001263157894736,
+      "loss": 0.6033,
+      "step": 62040
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003000473684210526,
+      "loss": 0.6055,
+      "step": 62050
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002999684210526316,
+      "loss": 0.6101,
+      "step": 62060
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002998894736842105,
+      "loss": 0.6141,
+      "step": 62070
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029981052631578947,
+      "loss": 0.6111,
+      "step": 62080
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029973157894736837,
+      "loss": 0.6125,
+      "step": 62090
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002996526315789473,
+      "loss": 0.6139,
+      "step": 62100
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029957368421052626,
+      "loss": 0.5965,
+      "step": 62110
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002994947368421052,
+      "loss": 0.6188,
+      "step": 62120
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002994157894736842,
+      "loss": 0.6104,
+      "step": 62130
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029933684210526316,
+      "loss": 0.6224,
+      "step": 62140
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002992578947368421,
+      "loss": 0.6328,
+      "step": 62150
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000299178947368421,
+      "loss": 0.6341,
+      "step": 62160
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029909999999999995,
+      "loss": 0.6263,
+      "step": 62170
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002990210526315789,
+      "loss": 0.6255,
+      "step": 62180
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029894210526315785,
+      "loss": 0.6184,
+      "step": 62190
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002988631578947368,
+      "loss": 0.6249,
+      "step": 62200
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029878421052631575,
+      "loss": 0.6208,
+      "step": 62210
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029870526315789475,
+      "loss": 0.6298,
+      "step": 62220
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002986263157894737,
+      "loss": 0.621,
+      "step": 62230
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002985473684210526,
+      "loss": 0.6203,
+      "step": 62240
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029846842105263154,
+      "loss": 0.6323,
+      "step": 62250
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002983894736842105,
+      "loss": 0.6298,
+      "step": 62260
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029831052631578943,
+      "loss": 0.6397,
+      "step": 62270
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002982315789473684,
+      "loss": 0.6296,
+      "step": 62280
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029815263157894733,
+      "loss": 0.6239,
+      "step": 62290
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002980736842105263,
+      "loss": 0.6316,
+      "step": 62300
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029799473684210523,
+      "loss": 0.6312,
+      "step": 62310
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002979157894736842,
+      "loss": 0.6491,
+      "step": 62320
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002978368421052631,
+      "loss": 0.6374,
+      "step": 62330
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002977578947368421,
+      "loss": 0.6282,
+      "step": 62340
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000297678947368421,
+      "loss": 0.6291,
+      "step": 62350
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029759999999999997,
+      "loss": 0.6272,
+      "step": 62360
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002975210526315789,
+      "loss": 0.6113,
+      "step": 62370
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029744210526315787,
+      "loss": 0.6234,
+      "step": 62380
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002973631578947368,
+      "loss": 0.6311,
+      "step": 62390
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029728421052631576,
+      "loss": 0.6298,
+      "step": 62400
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002972052631578947,
+      "loss": 0.6361,
+      "step": 62410
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029712631578947366,
+      "loss": 0.6314,
+      "step": 62420
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002970473684210526,
+      "loss": 0.6183,
+      "step": 62430
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029696842105263156,
+      "loss": 0.6201,
+      "step": 62440
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002968894736842105,
+      "loss": 0.6338,
+      "step": 62450
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029681052631578945,
+      "loss": 0.6498,
+      "step": 62460
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002967315789473684,
+      "loss": 0.6313,
+      "step": 62470
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029665263157894735,
+      "loss": 0.616,
+      "step": 62480
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002965736842105263,
+      "loss": 0.6071,
+      "step": 62490
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029649473684210525,
+      "loss": 0.6063,
+      "step": 62500
+    },
+    {
+      "epoch": 0.62,
+      "eval_accuracy": 0.8706084189495512,
+      "eval_loss": 0.60107421875,
+      "eval_runtime": 97.0491,
+      "eval_samples_per_second": 824.325,
+      "eval_steps_per_second": 1.618,
+      "step": 62500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002964157894736842,
+      "loss": 0.5934,
+      "step": 62510
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00029633684210526314,
+      "loss": 0.6107,
+      "step": 62520
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002962578947368421,
+      "loss": 0.6027,
+      "step": 62530
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00029617894736842104,
+      "loss": 0.6043,
+      "step": 62540
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002961,
+      "loss": 0.602,
+      "step": 62550
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00029602105263157894,
+      "loss": 0.6035,
+      "step": 62560
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002959421052631579,
+      "loss": 0.6028,
+      "step": 62570
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002958631578947368,
+      "loss": 0.6066,
+      "step": 62580
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002957842105263158,
+      "loss": 0.5998,
+      "step": 62590
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00029570526315789473,
+      "loss": 0.6178,
+      "step": 62600
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002956263157894737,
+      "loss": 0.619,
+      "step": 62610
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002955473684210526,
+      "loss": 0.6166,
+      "step": 62620
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002954684210526316,
+      "loss": 0.6262,
+      "step": 62630
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002953894736842105,
+      "loss": 0.632,
+      "step": 62640
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002953105263157894,
+      "loss": 0.6159,
+      "step": 62650
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00029523157894736836,
+      "loss": 0.6267,
+      "step": 62660
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002951526315789473,
+      "loss": 0.632,
+      "step": 62670
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00029507368421052626,
+      "loss": 0.6263,
+      "step": 62680
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00029499473684210526,
+      "loss": 0.6209,
+      "step": 62690
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002949157894736842,
+      "loss": 0.6242,
+      "step": 62700
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00029483684210526316,
+      "loss": 0.6247,
+      "step": 62710
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002947578947368421,
+      "loss": 0.6175,
+      "step": 62720
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000294678947368421,
+      "loss": 0.6334,
+      "step": 62730
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00029459999999999995,
+      "loss": 0.6237,
+      "step": 62740
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002945210526315789,
+      "loss": 0.6233,
+      "step": 62750
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00029444210526315785,
+      "loss": 0.6153,
+      "step": 62760
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002943631578947368,
+      "loss": 0.6262,
+      "step": 62770
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002942842105263158,
+      "loss": 0.6317,
+      "step": 62780
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00029420526315789475,
+      "loss": 0.6091,
+      "step": 62790
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00029412631578947364,
+      "loss": 0.6173,
+      "step": 62800
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002940473684210526,
+      "loss": 0.6332,
+      "step": 62810
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00029396842105263154,
+      "loss": 0.6217,
+      "step": 62820
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002938894736842105,
+      "loss": 0.623,
+      "step": 62830
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00029381052631578943,
+      "loss": 0.6223,
+      "step": 62840
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002937315789473684,
+      "loss": 0.6251,
+      "step": 62850
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00029365263157894733,
+      "loss": 0.6207,
+      "step": 62860
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002935736842105263,
+      "loss": 0.6127,
+      "step": 62870
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00029349473684210523,
+      "loss": 0.6234,
+      "step": 62880
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002934157894736842,
+      "loss": 0.6109,
+      "step": 62890
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002933368421052631,
+      "loss": 0.6243,
+      "step": 62900
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00029325789473684207,
+      "loss": 0.6196,
+      "step": 62910
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000293178947368421,
+      "loss": 0.6146,
+      "step": 62920
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00029309999999999997,
+      "loss": 0.6201,
+      "step": 62930
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002930210526315789,
+      "loss": 0.6172,
+      "step": 62940
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00029294210526315787,
+      "loss": 0.6087,
+      "step": 62950
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002928631578947368,
+      "loss": 0.6242,
+      "step": 62960
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00029278421052631576,
+      "loss": 0.6154,
+      "step": 62970
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00029271315789473684,
+      "loss": 0.6189,
+      "step": 62980
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002926342105263158,
+      "loss": 0.6134,
+      "step": 62990
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002925552631578947,
+      "loss": 0.6109,
+      "step": 63000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00029247631578947363,
+      "loss": 0.6258,
+      "step": 63010
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002923973684210526,
+      "loss": 0.6197,
+      "step": 63020
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002923184210526316,
+      "loss": 0.615,
+      "step": 63030
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002922394736842105,
+      "loss": 0.6197,
+      "step": 63040
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002921605263157895,
+      "loss": 0.6159,
+      "step": 63050
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002920815789473684,
+      "loss": 0.6103,
+      "step": 63060
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002920026315789473,
+      "loss": 0.607,
+      "step": 63070
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00029192368421052627,
+      "loss": 0.6194,
+      "step": 63080
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002918447368421052,
+      "loss": 0.6207,
+      "step": 63090
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00029176578947368416,
+      "loss": 0.6235,
+      "step": 63100
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002916868421052631,
+      "loss": 0.6125,
+      "step": 63110
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002916078947368421,
+      "loss": 0.6209,
+      "step": 63120
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00029152894736842106,
+      "loss": 0.6281,
+      "step": 63130
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00029145,
+      "loss": 0.608,
+      "step": 63140
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002913710526315789,
+      "loss": 0.6118,
+      "step": 63150
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00029129210526315785,
+      "loss": 0.6085,
+      "step": 63160
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002912131578947368,
+      "loss": 0.6079,
+      "step": 63170
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00029113421052631575,
+      "loss": 0.6188,
+      "step": 63180
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002910552631578947,
+      "loss": 0.6259,
+      "step": 63190
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00029097631578947365,
+      "loss": 0.6247,
+      "step": 63200
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002908973684210526,
+      "loss": 0.6283,
+      "step": 63210
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00029081842105263154,
+      "loss": 0.6154,
+      "step": 63220
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002907394736842105,
+      "loss": 0.6136,
+      "step": 63230
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00029066052631578944,
+      "loss": 0.6131,
+      "step": 63240
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002905815789473684,
+      "loss": 0.6044,
+      "step": 63250
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00029050263157894734,
+      "loss": 0.6216,
+      "step": 63260
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002904236842105263,
+      "loss": 0.6183,
+      "step": 63270
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00029034473684210523,
+      "loss": 0.6204,
+      "step": 63280
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002902657894736842,
+      "loss": 0.6173,
+      "step": 63290
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00029018684210526313,
+      "loss": 0.613,
+      "step": 63300
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002901078947368421,
+      "loss": 0.6297,
+      "step": 63310
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000290028947368421,
+      "loss": 0.6177,
+      "step": 63320
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00028995,
+      "loss": 0.6246,
+      "step": 63330
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002898710526315789,
+      "loss": 0.6265,
+      "step": 63340
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00028979210526315787,
+      "loss": 0.635,
+      "step": 63350
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002897131578947368,
+      "loss": 0.6188,
+      "step": 63360
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00028963421052631577,
+      "loss": 0.618,
+      "step": 63370
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002895552631578947,
+      "loss": 0.6167,
+      "step": 63380
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00028947631578947366,
+      "loss": 0.6207,
+      "step": 63390
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002893973684210526,
+      "loss": 0.6235,
+      "step": 63400
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00028931842105263156,
+      "loss": 0.6006,
+      "step": 63410
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002892394736842105,
+      "loss": 0.6155,
+      "step": 63420
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00028916052631578946,
+      "loss": 0.6181,
+      "step": 63430
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002890815789473684,
+      "loss": 0.6073,
+      "step": 63440
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00028900263157894735,
+      "loss": 0.6012,
+      "step": 63450
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002889236842105263,
+      "loss": 0.6,
+      "step": 63460
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00028884473684210525,
+      "loss": 0.6143,
+      "step": 63470
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002887657894736842,
+      "loss": 0.6202,
+      "step": 63480
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002886868421052631,
+      "loss": 0.6072,
+      "step": 63490
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002886078947368421,
+      "loss": 0.6031,
+      "step": 63500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028852894736842104,
+      "loss": 0.623,
+      "step": 63510
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028845,
+      "loss": 0.6119,
+      "step": 63520
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028837105263157894,
+      "loss": 0.6141,
+      "step": 63530
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002882921052631579,
+      "loss": 0.607,
+      "step": 63540
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028821315789473684,
+      "loss": 0.6221,
+      "step": 63550
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028813421052631573,
+      "loss": 0.623,
+      "step": 63560
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002880552631578947,
+      "loss": 0.6202,
+      "step": 63570
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002879763157894736,
+      "loss": 0.6273,
+      "step": 63580
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028789736842105263,
+      "loss": 0.6149,
+      "step": 63590
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002878184210526316,
+      "loss": 0.6185,
+      "step": 63600
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002877394736842105,
+      "loss": 0.5978,
+      "step": 63610
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002876605263157895,
+      "loss": 0.614,
+      "step": 63620
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002875815789473684,
+      "loss": 0.6211,
+      "step": 63630
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002875026315789473,
+      "loss": 0.6186,
+      "step": 63640
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028742368421052626,
+      "loss": 0.6133,
+      "step": 63650
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002873447368421052,
+      "loss": 0.6148,
+      "step": 63660
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028726578947368416,
+      "loss": 0.6075,
+      "step": 63670
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028718684210526316,
+      "loss": 0.6113,
+      "step": 63680
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002871078947368421,
+      "loss": 0.6121,
+      "step": 63690
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028702894736842106,
+      "loss": 0.6223,
+      "step": 63700
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028694999999999995,
+      "loss": 0.6098,
+      "step": 63710
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002868710526315789,
+      "loss": 0.6202,
+      "step": 63720
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028679210526315785,
+      "loss": 0.6097,
+      "step": 63730
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002867131578947368,
+      "loss": 0.6047,
+      "step": 63740
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028663421052631575,
+      "loss": 0.609,
+      "step": 63750
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002865552631578947,
+      "loss": 0.6166,
+      "step": 63760
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028647631578947364,
+      "loss": 0.6015,
+      "step": 63770
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002863973684210526,
+      "loss": 0.6228,
+      "step": 63780
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028631842105263154,
+      "loss": 0.6104,
+      "step": 63790
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002862394736842105,
+      "loss": 0.6093,
+      "step": 63800
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028616052631578944,
+      "loss": 0.6078,
+      "step": 63810
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002860815789473684,
+      "loss": 0.6127,
+      "step": 63820
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028600263157894733,
+      "loss": 0.6215,
+      "step": 63830
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002859236842105263,
+      "loss": 0.6187,
+      "step": 63840
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028584473684210523,
+      "loss": 0.6207,
+      "step": 63850
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002857657894736842,
+      "loss": 0.6077,
+      "step": 63860
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028568684210526313,
+      "loss": 0.6114,
+      "step": 63870
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002856078947368421,
+      "loss": 0.6312,
+      "step": 63880
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000285528947368421,
+      "loss": 0.6231,
+      "step": 63890
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028544999999999997,
+      "loss": 0.6297,
+      "step": 63900
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002853710526315789,
+      "loss": 0.6151,
+      "step": 63910
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028529210526315787,
+      "loss": 0.612,
+      "step": 63920
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002852131578947368,
+      "loss": 0.6184,
+      "step": 63930
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028513421052631577,
+      "loss": 0.6172,
+      "step": 63940
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002850552631578947,
+      "loss": 0.6244,
+      "step": 63950
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028497631578947366,
+      "loss": 0.6144,
+      "step": 63960
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002848973684210526,
+      "loss": 0.6161,
+      "step": 63970
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028481842105263156,
+      "loss": 0.6147,
+      "step": 63980
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002847394736842105,
+      "loss": 0.6128,
+      "step": 63990
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028466052631578946,
+      "loss": 0.6253,
+      "step": 64000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002845815789473684,
+      "loss": 0.6227,
+      "step": 64010
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028450263157894735,
+      "loss": 0.6294,
+      "step": 64020
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002844236842105263,
+      "loss": 0.6138,
+      "step": 64030
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028434473684210525,
+      "loss": 0.6185,
+      "step": 64040
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028426578947368414,
+      "loss": 0.6142,
+      "step": 64050
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028418684210526315,
+      "loss": 0.6274,
+      "step": 64060
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002841078947368421,
+      "loss": 0.6248,
+      "step": 64070
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028402894736842104,
+      "loss": 0.617,
+      "step": 64080
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028395,
+      "loss": 0.6273,
+      "step": 64090
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028387105263157894,
+      "loss": 0.6357,
+      "step": 64100
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002837921052631579,
+      "loss": 0.6099,
+      "step": 64110
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028371315789473684,
+      "loss": 0.6208,
+      "step": 64120
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028363421052631573,
+      "loss": 0.6179,
+      "step": 64130
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002835552631578947,
+      "loss": 0.6301,
+      "step": 64140
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002834763157894737,
+      "loss": 0.6158,
+      "step": 64150
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028339736842105263,
+      "loss": 0.627,
+      "step": 64160
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002833184210526316,
+      "loss": 0.6273,
+      "step": 64170
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002832394736842105,
+      "loss": 0.6132,
+      "step": 64180
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002831605263157895,
+      "loss": 0.6196,
+      "step": 64190
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028308157894736837,
+      "loss": 0.6202,
+      "step": 64200
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002830026315789473,
+      "loss": 0.6246,
+      "step": 64210
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028292368421052626,
+      "loss": 0.623,
+      "step": 64220
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002828447368421052,
+      "loss": 0.6155,
+      "step": 64230
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002827657894736842,
+      "loss": 0.6124,
+      "step": 64240
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028268684210526316,
+      "loss": 0.6046,
+      "step": 64250
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002826078947368421,
+      "loss": 0.6129,
+      "step": 64260
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028252894736842106,
+      "loss": 0.6272,
+      "step": 64270
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028244999999999995,
+      "loss": 0.6276,
+      "step": 64280
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002823710526315789,
+      "loss": 0.6191,
+      "step": 64290
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028229210526315785,
+      "loss": 0.5869,
+      "step": 64300
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002822131578947368,
+      "loss": 0.6192,
+      "step": 64310
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028213421052631575,
+      "loss": 0.6201,
+      "step": 64320
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028205526315789475,
+      "loss": 0.603,
+      "step": 64330
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002819763157894737,
+      "loss": 0.586,
+      "step": 64340
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002818973684210526,
+      "loss": 0.6202,
+      "step": 64350
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028181842105263154,
+      "loss": 0.6076,
+      "step": 64360
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002817394736842105,
+      "loss": 0.6187,
+      "step": 64370
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028166052631578944,
+      "loss": 0.6214,
+      "step": 64380
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002815815789473684,
+      "loss": 0.6157,
+      "step": 64390
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028150263157894733,
+      "loss": 0.6171,
+      "step": 64400
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002814236842105263,
+      "loss": 0.6168,
+      "step": 64410
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028134473684210523,
+      "loss": 0.6155,
+      "step": 64420
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002812657894736842,
+      "loss": 0.6266,
+      "step": 64430
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002811868421052631,
+      "loss": 0.6188,
+      "step": 64440
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002811078947368421,
+      "loss": 0.6228,
+      "step": 64450
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000281028947368421,
+      "loss": 0.6205,
+      "step": 64460
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028094999999999997,
+      "loss": 0.6278,
+      "step": 64470
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002808710526315789,
+      "loss": 0.6114,
+      "step": 64480
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028079210526315787,
+      "loss": 0.6115,
+      "step": 64490
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002807131578947368,
+      "loss": 0.6262,
+      "step": 64500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028063421052631576,
+      "loss": 0.6117,
+      "step": 64510
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002805552631578947,
+      "loss": 0.6196,
+      "step": 64520
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028047631578947366,
+      "loss": 0.6172,
+      "step": 64530
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002803973684210526,
+      "loss": 0.6063,
+      "step": 64540
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028031842105263156,
+      "loss": 0.6109,
+      "step": 64550
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002802394736842105,
+      "loss": 0.6074,
+      "step": 64560
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028016052631578945,
+      "loss": 0.6105,
+      "step": 64570
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002800815789473684,
+      "loss": 0.6199,
+      "step": 64580
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028000263157894735,
+      "loss": 0.6157,
+      "step": 64590
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002799236842105263,
+      "loss": 0.6174,
+      "step": 64600
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00027984473684210525,
+      "loss": 0.6083,
+      "step": 64610
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002797657894736842,
+      "loss": 0.6119,
+      "step": 64620
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00027968684210526314,
+      "loss": 0.6272,
+      "step": 64630
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002796078947368421,
+      "loss": 0.6078,
+      "step": 64640
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00027952894736842104,
+      "loss": 0.6267,
+      "step": 64650
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00027945,
+      "loss": 0.622,
+      "step": 64660
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00027937105263157894,
+      "loss": 0.6205,
+      "step": 64670
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002792921052631579,
+      "loss": 0.6193,
+      "step": 64680
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002792131578947368,
+      "loss": 0.6246,
+      "step": 64690
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00027913421052631573,
+      "loss": 0.627,
+      "step": 64700
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00027905526315789473,
+      "loss": 0.627,
+      "step": 64710
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002789763157894737,
+      "loss": 0.6091,
+      "step": 64720
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00027889736842105263,
+      "loss": 0.6127,
+      "step": 64730
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002788184210526316,
+      "loss": 0.6177,
+      "step": 64740
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002787394736842105,
+      "loss": 0.6183,
+      "step": 64750
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00027866052631578947,
+      "loss": 0.6197,
+      "step": 64760
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00027858157894736837,
+      "loss": 0.6205,
+      "step": 64770
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002785026315789473,
+      "loss": 0.6174,
+      "step": 64780
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00027842368421052626,
+      "loss": 0.621,
+      "step": 64790
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00027834473684210527,
+      "loss": 0.6157,
+      "step": 64800
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002782657894736842,
+      "loss": 0.6121,
+      "step": 64810
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00027818684210526316,
+      "loss": 0.6165,
+      "step": 64820
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002781078947368421,
+      "loss": 0.6229,
+      "step": 64830
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000278028947368421,
+      "loss": 0.6208,
+      "step": 64840
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00027794999999999995,
+      "loss": 0.6162,
+      "step": 64850
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002778710526315789,
+      "loss": 0.6046,
+      "step": 64860
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00027779210526315785,
+      "loss": 0.6047,
+      "step": 64870
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002777131578947368,
+      "loss": 0.6118,
+      "step": 64880
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002776342105263158,
+      "loss": 0.6132,
+      "step": 64890
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00027755526315789475,
+      "loss": 0.6199,
+      "step": 64900
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00027747631578947364,
+      "loss": 0.6303,
+      "step": 64910
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002773973684210526,
+      "loss": 0.6183,
+      "step": 64920
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00027731842105263154,
+      "loss": 0.612,
+      "step": 64930
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002772394736842105,
+      "loss": 0.6118,
+      "step": 64940
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00027716052631578944,
+      "loss": 0.6252,
+      "step": 64950
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002770815789473684,
+      "loss": 0.6116,
+      "step": 64960
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00027700263157894733,
+      "loss": 0.6268,
+      "step": 64970
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002769236842105263,
+      "loss": 0.6126,
+      "step": 64980
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00027685263157894736,
+      "loss": 0.6084,
+      "step": 64990
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002767736842105263,
+      "loss": 0.6029,
+      "step": 65000
+    },
+    {
+      "epoch": 0.65,
+      "eval_accuracy": 0.8712975328502192,
+      "eval_loss": 0.59521484375,
+      "eval_runtime": 97.298,
+      "eval_samples_per_second": 822.217,
+      "eval_steps_per_second": 1.614,
+      "step": 65000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00027669473684210525,
+      "loss": 0.6275,
+      "step": 65010
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002766157894736842,
+      "loss": 0.6172,
+      "step": 65020
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00027653684210526315,
+      "loss": 0.637,
+      "step": 65030
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00027645789473684204,
+      "loss": 0.6185,
+      "step": 65040
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00027637894736842105,
+      "loss": 0.6182,
+      "step": 65050
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002763,
+      "loss": 0.6237,
+      "step": 65060
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00027622105263157894,
+      "loss": 0.6106,
+      "step": 65070
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002761421052631579,
+      "loss": 0.6254,
+      "step": 65080
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00027606315789473684,
+      "loss": 0.6164,
+      "step": 65090
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002759842105263158,
+      "loss": 0.6195,
+      "step": 65100
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002759131578947368,
+      "loss": 0.6106,
+      "step": 65110
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00027583421052631576,
+      "loss": 0.6329,
+      "step": 65120
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002757552631578947,
+      "loss": 0.6166,
+      "step": 65130
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00027567631578947365,
+      "loss": 0.6229,
+      "step": 65140
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002755973684210526,
+      "loss": 0.6177,
+      "step": 65150
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00027551842105263155,
+      "loss": 0.629,
+      "step": 65160
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002754394736842105,
+      "loss": 0.6317,
+      "step": 65170
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00027536052631578945,
+      "loss": 0.6117,
+      "step": 65180
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002752815789473684,
+      "loss": 0.6122,
+      "step": 65190
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00027520263157894734,
+      "loss": 0.6091,
+      "step": 65200
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002751236842105263,
+      "loss": 0.6138,
+      "step": 65210
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00027504473684210524,
+      "loss": 0.6181,
+      "step": 65220
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002749657894736842,
+      "loss": 0.6213,
+      "step": 65230
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00027488684210526314,
+      "loss": 0.6174,
+      "step": 65240
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002748078947368421,
+      "loss": 0.6162,
+      "step": 65250
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00027472894736842103,
+      "loss": 0.6247,
+      "step": 65260
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00027465,
+      "loss": 0.6168,
+      "step": 65270
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00027457105263157893,
+      "loss": 0.6246,
+      "step": 65280
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002744921052631579,
+      "loss": 0.6223,
+      "step": 65290
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002744131578947368,
+      "loss": 0.6196,
+      "step": 65300
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002743342105263158,
+      "loss": 0.6196,
+      "step": 65310
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002742552631578947,
+      "loss": 0.6128,
+      "step": 65320
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00027417631578947367,
+      "loss": 0.6262,
+      "step": 65330
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002740973684210526,
+      "loss": 0.6173,
+      "step": 65340
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00027401842105263157,
+      "loss": 0.6169,
+      "step": 65350
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002739394736842105,
+      "loss": 0.6241,
+      "step": 65360
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00027386052631578946,
+      "loss": 0.6195,
+      "step": 65370
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00027378157894736836,
+      "loss": 0.6215,
+      "step": 65380
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002737026315789473,
+      "loss": 0.62,
+      "step": 65390
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002736236842105263,
+      "loss": 0.6165,
+      "step": 65400
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00027354473684210526,
+      "loss": 0.6089,
+      "step": 65410
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002734657894736842,
+      "loss": 0.6126,
+      "step": 65420
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00027338684210526315,
+      "loss": 0.6049,
+      "step": 65430
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002733078947368421,
+      "loss": 0.6083,
+      "step": 65440
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00027322894736842105,
+      "loss": 0.6204,
+      "step": 65450
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00027314999999999994,
+      "loss": 0.6178,
+      "step": 65460
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002730710526315789,
+      "loss": 0.6159,
+      "step": 65470
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00027299210526315784,
+      "loss": 0.6279,
+      "step": 65480
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00027291315789473684,
+      "loss": 0.6057,
+      "step": 65490
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002728342105263158,
+      "loss": 0.6107,
+      "step": 65500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027275526315789474,
+      "loss": 0.6107,
+      "step": 65510
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002726763157894737,
+      "loss": 0.6217,
+      "step": 65520
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002725973684210526,
+      "loss": 0.6232,
+      "step": 65530
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027251842105263153,
+      "loss": 0.6323,
+      "step": 65540
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002724394736842105,
+      "loss": 0.6193,
+      "step": 65550
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027236052631578943,
+      "loss": 0.6173,
+      "step": 65560
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002722815789473684,
+      "loss": 0.6192,
+      "step": 65570
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002722026315789474,
+      "loss": 0.6125,
+      "step": 65580
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002721236842105263,
+      "loss": 0.6193,
+      "step": 65590
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002720447368421053,
+      "loss": 0.6226,
+      "step": 65600
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027196578947368417,
+      "loss": 0.6216,
+      "step": 65610
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002718868421052631,
+      "loss": 0.614,
+      "step": 65620
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027180789473684207,
+      "loss": 0.6163,
+      "step": 65630
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000271728947368421,
+      "loss": 0.622,
+      "step": 65640
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027164999999999996,
+      "loss": 0.6128,
+      "step": 65650
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002715710526315789,
+      "loss": 0.6167,
+      "step": 65660
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027149210526315786,
+      "loss": 0.6153,
+      "step": 65670
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002714131578947368,
+      "loss": 0.6106,
+      "step": 65680
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027133421052631576,
+      "loss": 0.6077,
+      "step": 65690
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002712552631578947,
+      "loss": 0.618,
+      "step": 65700
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027117631578947365,
+      "loss": 0.6081,
+      "step": 65710
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002710973684210526,
+      "loss": 0.627,
+      "step": 65720
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027101842105263155,
+      "loss": 0.6084,
+      "step": 65730
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002709394736842105,
+      "loss": 0.6128,
+      "step": 65740
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027086052631578945,
+      "loss": 0.6034,
+      "step": 65750
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002707815789473684,
+      "loss": 0.6198,
+      "step": 65760
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027070263157894734,
+      "loss": 0.6161,
+      "step": 65770
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002706236842105263,
+      "loss": 0.6204,
+      "step": 65780
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027054473684210524,
+      "loss": 0.6147,
+      "step": 65790
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002704657894736842,
+      "loss": 0.6048,
+      "step": 65800
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027038684210526314,
+      "loss": 0.6058,
+      "step": 65810
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002703078947368421,
+      "loss": 0.6189,
+      "step": 65820
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027022894736842103,
+      "loss": 0.6136,
+      "step": 65830
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027015,
+      "loss": 0.6153,
+      "step": 65840
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027007105263157893,
+      "loss": 0.6257,
+      "step": 65850
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002699921052631579,
+      "loss": 0.6279,
+      "step": 65860
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002699131578947368,
+      "loss": 0.6163,
+      "step": 65870
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002698342105263158,
+      "loss": 0.6127,
+      "step": 65880
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002697552631578947,
+      "loss": 0.6032,
+      "step": 65890
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026967631578947367,
+      "loss": 0.6238,
+      "step": 65900
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002695973684210526,
+      "loss": 0.6083,
+      "step": 65910
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026951842105263157,
+      "loss": 0.6041,
+      "step": 65920
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002694394736842105,
+      "loss": 0.6112,
+      "step": 65930
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026936052631578946,
+      "loss": 0.6085,
+      "step": 65940
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026928157894736836,
+      "loss": 0.6115,
+      "step": 65950
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026920263157894736,
+      "loss": 0.6172,
+      "step": 65960
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002691236842105263,
+      "loss": 0.6136,
+      "step": 65970
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026904473684210526,
+      "loss": 0.6114,
+      "step": 65980
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002689657894736842,
+      "loss": 0.606,
+      "step": 65990
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026888684210526315,
+      "loss": 0.6049,
+      "step": 66000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002688078947368421,
+      "loss": 0.6122,
+      "step": 66010
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000268728947368421,
+      "loss": 0.613,
+      "step": 66020
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026864999999999994,
+      "loss": 0.6181,
+      "step": 66030
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002685710526315789,
+      "loss": 0.6076,
+      "step": 66040
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002684921052631579,
+      "loss": 0.6094,
+      "step": 66050
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026841315789473684,
+      "loss": 0.6061,
+      "step": 66060
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002683342105263158,
+      "loss": 0.6089,
+      "step": 66070
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026825526315789474,
+      "loss": 0.6199,
+      "step": 66080
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002681763157894737,
+      "loss": 0.6147,
+      "step": 66090
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002680973684210526,
+      "loss": 0.6188,
+      "step": 66100
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026801842105263153,
+      "loss": 0.6109,
+      "step": 66110
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002679394736842105,
+      "loss": 0.6093,
+      "step": 66120
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002678605263157894,
+      "loss": 0.617,
+      "step": 66130
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026778157894736843,
+      "loss": 0.6152,
+      "step": 66140
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002677026315789474,
+      "loss": 0.6077,
+      "step": 66150
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002676236842105263,
+      "loss": 0.6152,
+      "step": 66160
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002675447368421052,
+      "loss": 0.6165,
+      "step": 66170
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026746578947368417,
+      "loss": 0.6205,
+      "step": 66180
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002673868421052631,
+      "loss": 0.6121,
+      "step": 66190
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026730789473684206,
+      "loss": 0.6138,
+      "step": 66200
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000267228947368421,
+      "loss": 0.6088,
+      "step": 66210
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026714999999999996,
+      "loss": 0.6114,
+      "step": 66220
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002670710526315789,
+      "loss": 0.6249,
+      "step": 66230
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026699210526315786,
+      "loss": 0.6134,
+      "step": 66240
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002669131578947368,
+      "loss": 0.6137,
+      "step": 66250
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026683421052631575,
+      "loss": 0.6173,
+      "step": 66260
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002667552631578947,
+      "loss": 0.6057,
+      "step": 66270
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026667631578947365,
+      "loss": 0.6091,
+      "step": 66280
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002665973684210526,
+      "loss": 0.6195,
+      "step": 66290
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026651842105263155,
+      "loss": 0.6172,
+      "step": 66300
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002664394736842105,
+      "loss": 0.6113,
+      "step": 66310
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026636052631578944,
+      "loss": 0.6148,
+      "step": 66320
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002662815789473684,
+      "loss": 0.605,
+      "step": 66330
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026620263157894734,
+      "loss": 0.6161,
+      "step": 66340
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002661236842105263,
+      "loss": 0.6172,
+      "step": 66350
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026604473684210524,
+      "loss": 0.6171,
+      "step": 66360
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002659657894736842,
+      "loss": 0.609,
+      "step": 66370
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026588684210526313,
+      "loss": 0.6061,
+      "step": 66380
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002658078947368421,
+      "loss": 0.5975,
+      "step": 66390
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026572894736842103,
+      "loss": 0.6149,
+      "step": 66400
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026565,
+      "loss": 0.623,
+      "step": 66410
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026557105263157893,
+      "loss": 0.607,
+      "step": 66420
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002654921052631579,
+      "loss": 0.6125,
+      "step": 66430
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002654131578947368,
+      "loss": 0.6125,
+      "step": 66440
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026533421052631577,
+      "loss": 0.5879,
+      "step": 66450
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002652552631578947,
+      "loss": 0.6151,
+      "step": 66460
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026517631578947367,
+      "loss": 0.6237,
+      "step": 66470
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002650973684210526,
+      "loss": 0.6168,
+      "step": 66480
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026501842105263157,
+      "loss": 0.6005,
+      "step": 66490
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002649394736842105,
+      "loss": 0.6086,
+      "step": 66500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002648605263157894,
+      "loss": 0.6099,
+      "step": 66510
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002647815789473684,
+      "loss": 0.6189,
+      "step": 66520
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026470263157894736,
+      "loss": 0.6127,
+      "step": 66530
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002646236842105263,
+      "loss": 0.6078,
+      "step": 66540
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026454473684210526,
+      "loss": 0.6057,
+      "step": 66550
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002644657894736842,
+      "loss": 0.6142,
+      "step": 66560
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026438684210526315,
+      "loss": 0.6087,
+      "step": 66570
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002643078947368421,
+      "loss": 0.6079,
+      "step": 66580
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000264228947368421,
+      "loss": 0.5868,
+      "step": 66590
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026414999999999994,
+      "loss": 0.5994,
+      "step": 66600
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026407105263157895,
+      "loss": 0.5836,
+      "step": 66610
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002639921052631579,
+      "loss": 0.6064,
+      "step": 66620
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002639210526315789,
+      "loss": 0.6007,
+      "step": 66630
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026384210526315786,
+      "loss": 0.5975,
+      "step": 66640
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002637631578947368,
+      "loss": 0.5916,
+      "step": 66650
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026368421052631576,
+      "loss": 0.6027,
+      "step": 66660
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002636052631578947,
+      "loss": 0.6034,
+      "step": 66670
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026352631578947366,
+      "loss": 0.6089,
+      "step": 66680
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002634473684210526,
+      "loss": 0.5924,
+      "step": 66690
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026336842105263155,
+      "loss": 0.5966,
+      "step": 66700
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002632894736842105,
+      "loss": 0.59,
+      "step": 66710
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026321052631578945,
+      "loss": 0.5982,
+      "step": 66720
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002631315789473684,
+      "loss": 0.5976,
+      "step": 66730
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026305263157894735,
+      "loss": 0.608,
+      "step": 66740
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002629736842105263,
+      "loss": 0.6017,
+      "step": 66750
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026289473684210524,
+      "loss": 0.5987,
+      "step": 66760
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002628157894736842,
+      "loss": 0.5861,
+      "step": 66770
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026273684210526314,
+      "loss": 0.5971,
+      "step": 66780
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002626578947368421,
+      "loss": 0.5958,
+      "step": 66790
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026257894736842104,
+      "loss": 0.596,
+      "step": 66800
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002625,
+      "loss": 0.5842,
+      "step": 66810
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026242105263157893,
+      "loss": 0.5935,
+      "step": 66820
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002623421052631579,
+      "loss": 0.5913,
+      "step": 66830
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026226315789473683,
+      "loss": 0.6111,
+      "step": 66840
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002621842105263158,
+      "loss": 0.6131,
+      "step": 66850
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002621052631578947,
+      "loss": 0.6299,
+      "step": 66860
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002620263157894737,
+      "loss": 0.6226,
+      "step": 66870
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002619473684210526,
+      "loss": 0.5992,
+      "step": 66880
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026186842105263157,
+      "loss": 0.6188,
+      "step": 66890
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002617894736842105,
+      "loss": 0.6228,
+      "step": 66900
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026171052631578947,
+      "loss": 0.6228,
+      "step": 66910
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002616315789473684,
+      "loss": 0.622,
+      "step": 66920
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002615526315789473,
+      "loss": 0.6124,
+      "step": 66930
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026147368421052626,
+      "loss": 0.6236,
+      "step": 66940
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026139473684210526,
+      "loss": 0.6076,
+      "step": 66950
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002613157894736842,
+      "loss": 0.6119,
+      "step": 66960
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026123684210526316,
+      "loss": 0.6109,
+      "step": 66970
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002611578947368421,
+      "loss": 0.618,
+      "step": 66980
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026107894736842105,
+      "loss": 0.6172,
+      "step": 66990
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000261,
+      "loss": 0.6186,
+      "step": 67000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002609210526315789,
+      "loss": 0.6134,
+      "step": 67010
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026084210526315784,
+      "loss": 0.6014,
+      "step": 67020
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002607631578947368,
+      "loss": 0.6121,
+      "step": 67030
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026068421052631574,
+      "loss": 0.6131,
+      "step": 67040
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026060526315789474,
+      "loss": 0.6212,
+      "step": 67050
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002605263157894737,
+      "loss": 0.6212,
+      "step": 67060
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026044736842105264,
+      "loss": 0.6098,
+      "step": 67070
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026036842105263153,
+      "loss": 0.6035,
+      "step": 67080
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002602894736842105,
+      "loss": 0.5951,
+      "step": 67090
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026021052631578943,
+      "loss": 0.5969,
+      "step": 67100
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002601315789473684,
+      "loss": 0.5876,
+      "step": 67110
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002600526315789473,
+      "loss": 0.6134,
+      "step": 67120
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002599736842105263,
+      "loss": 0.6154,
+      "step": 67130
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002598947368421053,
+      "loss": 0.6118,
+      "step": 67140
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002598157894736842,
+      "loss": 0.6091,
+      "step": 67150
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002597368421052631,
+      "loss": 0.6192,
+      "step": 67160
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025965789473684207,
+      "loss": 0.6088,
+      "step": 67170
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000259578947368421,
+      "loss": 0.6051,
+      "step": 67180
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025949999999999997,
+      "loss": 0.6081,
+      "step": 67190
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002594210526315789,
+      "loss": 0.6192,
+      "step": 67200
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025934210526315786,
+      "loss": 0.6138,
+      "step": 67210
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002592631578947368,
+      "loss": 0.6073,
+      "step": 67220
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025918421052631576,
+      "loss": 0.6092,
+      "step": 67230
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002591052631578947,
+      "loss": 0.6033,
+      "step": 67240
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025902631578947365,
+      "loss": 0.5994,
+      "step": 67250
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002589473684210526,
+      "loss": 0.6172,
+      "step": 67260
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025886842105263155,
+      "loss": 0.6103,
+      "step": 67270
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002587894736842105,
+      "loss": 0.6128,
+      "step": 67280
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025871052631578945,
+      "loss": 0.6112,
+      "step": 67290
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002586315789473684,
+      "loss": 0.6225,
+      "step": 67300
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025855263157894734,
+      "loss": 0.6106,
+      "step": 67310
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002584736842105263,
+      "loss": 0.6076,
+      "step": 67320
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025839473684210524,
+      "loss": 0.61,
+      "step": 67330
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002583157894736842,
+      "loss": 0.6145,
+      "step": 67340
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025823684210526314,
+      "loss": 0.5988,
+      "step": 67350
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002581578947368421,
+      "loss": 0.6034,
+      "step": 67360
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025807894736842103,
+      "loss": 0.6168,
+      "step": 67370
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000258,
+      "loss": 0.6039,
+      "step": 67380
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025792105263157893,
+      "loss": 0.6013,
+      "step": 67390
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002578421052631579,
+      "loss": 0.6092,
+      "step": 67400
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025776315789473683,
+      "loss": 0.6197,
+      "step": 67410
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002576842105263158,
+      "loss": 0.6197,
+      "step": 67420
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002576052631578947,
+      "loss": 0.6105,
+      "step": 67430
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025752631578947367,
+      "loss": 0.6155,
+      "step": 67440
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002574473684210526,
+      "loss": 0.6026,
+      "step": 67450
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025736842105263157,
+      "loss": 0.6058,
+      "step": 67460
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002572894736842105,
+      "loss": 0.6007,
+      "step": 67470
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025721052631578947,
+      "loss": 0.6056,
+      "step": 67480
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002571315789473684,
+      "loss": 0.6256,
+      "step": 67490
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002570526315789473,
+      "loss": 0.615,
+      "step": 67500
+    },
+    {
+      "epoch": 0.68,
+      "eval_accuracy": 0.8707361782979045,
+      "eval_loss": 0.5947265625,
+      "eval_runtime": 97.3705,
+      "eval_samples_per_second": 821.604,
+      "eval_steps_per_second": 1.612,
+      "step": 67500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002569736842105263,
+      "loss": 0.6038,
+      "step": 67510
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025689473684210526,
+      "loss": 0.6058,
+      "step": 67520
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002568157894736842,
+      "loss": 0.6052,
+      "step": 67530
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025673684210526316,
+      "loss": 0.6114,
+      "step": 67540
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002566578947368421,
+      "loss": 0.606,
+      "step": 67550
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025657894736842105,
+      "loss": 0.6142,
+      "step": 67560
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025649999999999995,
+      "loss": 0.6073,
+      "step": 67570
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002564210526315789,
+      "loss": 0.6219,
+      "step": 67580
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025634210526315784,
+      "loss": 0.6162,
+      "step": 67590
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002562631578947368,
+      "loss": 0.6199,
+      "step": 67600
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002561842105263158,
+      "loss": 0.6224,
+      "step": 67610
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025610526315789474,
+      "loss": 0.6154,
+      "step": 67620
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002560263157894737,
+      "loss": 0.6133,
+      "step": 67630
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025594736842105264,
+      "loss": 0.6166,
+      "step": 67640
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025586842105263153,
+      "loss": 0.6177,
+      "step": 67650
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002557894736842105,
+      "loss": 0.607,
+      "step": 67660
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025571052631578943,
+      "loss": 0.6188,
+      "step": 67670
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002556315789473684,
+      "loss": 0.6005,
+      "step": 67680
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002555526315789473,
+      "loss": 0.6192,
+      "step": 67690
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025547368421052633,
+      "loss": 0.614,
+      "step": 67700
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002553947368421053,
+      "loss": 0.6191,
+      "step": 67710
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025531578947368417,
+      "loss": 0.6169,
+      "step": 67720
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002552368421052631,
+      "loss": 0.6028,
+      "step": 67730
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025515789473684207,
+      "loss": 0.6043,
+      "step": 67740
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000255078947368421,
+      "loss": 0.6143,
+      "step": 67750
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025499999999999996,
+      "loss": 0.6207,
+      "step": 67760
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002549210526315789,
+      "loss": 0.6236,
+      "step": 67770
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025484210526315786,
+      "loss": 0.6054,
+      "step": 67780
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002547631578947368,
+      "loss": 0.5998,
+      "step": 67790
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025468421052631576,
+      "loss": 0.6129,
+      "step": 67800
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002546052631578947,
+      "loss": 0.6191,
+      "step": 67810
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025452631578947365,
+      "loss": 0.627,
+      "step": 67820
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002544473684210526,
+      "loss": 0.6209,
+      "step": 67830
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025436842105263155,
+      "loss": 0.6149,
+      "step": 67840
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002542894736842105,
+      "loss": 0.6252,
+      "step": 67850
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025421052631578945,
+      "loss": 0.6138,
+      "step": 67860
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002541315789473684,
+      "loss": 0.6198,
+      "step": 67870
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025405263157894734,
+      "loss": 0.6133,
+      "step": 67880
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002539736842105263,
+      "loss": 0.62,
+      "step": 67890
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025389473684210524,
+      "loss": 0.6147,
+      "step": 67900
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002538157894736842,
+      "loss": 0.6097,
+      "step": 67910
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025373684210526314,
+      "loss": 0.6146,
+      "step": 67920
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025366578947368416,
+      "loss": 0.6189,
+      "step": 67930
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002535868421052631,
+      "loss": 0.6189,
+      "step": 67940
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002535078947368421,
+      "loss": 0.6292,
+      "step": 67950
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025342894736842106,
+      "loss": 0.6081,
+      "step": 67960
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025335,
+      "loss": 0.6154,
+      "step": 67970
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025327105263157895,
+      "loss": 0.6009,
+      "step": 67980
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025319210526315785,
+      "loss": 0.6056,
+      "step": 67990
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002531131578947368,
+      "loss": 0.6165,
+      "step": 68000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025303421052631574,
+      "loss": 0.6294,
+      "step": 68010
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002529552631578947,
+      "loss": 0.6219,
+      "step": 68020
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025287631578947364,
+      "loss": 0.6237,
+      "step": 68030
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002527973684210526,
+      "loss": 0.6049,
+      "step": 68040
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002527184210526316,
+      "loss": 0.6195,
+      "step": 68050
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025263947368421054,
+      "loss": 0.6178,
+      "step": 68060
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025256052631578943,
+      "loss": 0.6155,
+      "step": 68070
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002524815789473684,
+      "loss": 0.6085,
+      "step": 68080
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025240263157894733,
+      "loss": 0.6197,
+      "step": 68090
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002523236842105263,
+      "loss": 0.61,
+      "step": 68100
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025224473684210523,
+      "loss": 0.6132,
+      "step": 68110
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002521657894736842,
+      "loss": 0.6193,
+      "step": 68120
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002520868421052631,
+      "loss": 0.604,
+      "step": 68130
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025200789473684207,
+      "loss": 0.6106,
+      "step": 68140
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000251928947368421,
+      "loss": 0.6148,
+      "step": 68150
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025184999999999997,
+      "loss": 0.5906,
+      "step": 68160
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002517710526315789,
+      "loss": 0.6127,
+      "step": 68170
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025169210526315787,
+      "loss": 0.6092,
+      "step": 68180
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002516131578947368,
+      "loss": 0.6262,
+      "step": 68190
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025153421052631576,
+      "loss": 0.618,
+      "step": 68200
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002514552631578947,
+      "loss": 0.6237,
+      "step": 68210
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025137631578947366,
+      "loss": 0.6087,
+      "step": 68220
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002512973684210526,
+      "loss": 0.6156,
+      "step": 68230
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025121842105263156,
+      "loss": 0.6084,
+      "step": 68240
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002511394736842105,
+      "loss": 0.6167,
+      "step": 68250
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025106052631578945,
+      "loss": 0.616,
+      "step": 68260
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002509815789473684,
+      "loss": 0.6084,
+      "step": 68270
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025090263157894735,
+      "loss": 0.6154,
+      "step": 68280
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002508236842105263,
+      "loss": 0.6127,
+      "step": 68290
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025074473684210525,
+      "loss": 0.6065,
+      "step": 68300
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002506657894736842,
+      "loss": 0.6116,
+      "step": 68310
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025058684210526314,
+      "loss": 0.6053,
+      "step": 68320
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002505078947368421,
+      "loss": 0.6185,
+      "step": 68330
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025042894736842104,
+      "loss": 0.6165,
+      "step": 68340
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025035,
+      "loss": 0.6125,
+      "step": 68350
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025027105263157893,
+      "loss": 0.6134,
+      "step": 68360
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002501921052631579,
+      "loss": 0.6174,
+      "step": 68370
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025011315789473683,
+      "loss": 0.6202,
+      "step": 68380
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002500342105263158,
+      "loss": 0.6204,
+      "step": 68390
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024995526315789473,
+      "loss": 0.6053,
+      "step": 68400
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002498763157894736,
+      "loss": 0.6128,
+      "step": 68410
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002497973684210526,
+      "loss": 0.6161,
+      "step": 68420
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002497184210526316,
+      "loss": 0.6068,
+      "step": 68430
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002496394736842105,
+      "loss": 0.6023,
+      "step": 68440
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024956052631578947,
+      "loss": 0.6008,
+      "step": 68450
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002494815789473684,
+      "loss": 0.6074,
+      "step": 68460
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024940263157894737,
+      "loss": 0.5798,
+      "step": 68470
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024932368421052626,
+      "loss": 0.603,
+      "step": 68480
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002492447368421052,
+      "loss": 0.6085,
+      "step": 68490
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00024916578947368416,
+      "loss": 0.6058,
+      "step": 68500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00024908684210526316,
+      "loss": 0.6086,
+      "step": 68510
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002490078947368421,
+      "loss": 0.6189,
+      "step": 68520
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00024892894736842106,
+      "loss": 0.609,
+      "step": 68530
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00024885,
+      "loss": 0.6023,
+      "step": 68540
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00024877105263157895,
+      "loss": 0.6105,
+      "step": 68550
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00024869210526315785,
+      "loss": 0.6157,
+      "step": 68560
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002486131578947368,
+      "loss": 0.6098,
+      "step": 68570
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00024853421052631574,
+      "loss": 0.6157,
+      "step": 68580
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002484552631578947,
+      "loss": 0.6039,
+      "step": 68590
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002483763157894737,
+      "loss": 0.6003,
+      "step": 68600
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00024829736842105264,
+      "loss": 0.597,
+      "step": 68610
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002482184210526316,
+      "loss": 0.6065,
+      "step": 68620
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002481394736842105,
+      "loss": 0.6053,
+      "step": 68630
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00024806052631578943,
+      "loss": 0.612,
+      "step": 68640
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002479815789473684,
+      "loss": 0.6069,
+      "step": 68650
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00024790263157894733,
+      "loss": 0.6088,
+      "step": 68660
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002478236842105263,
+      "loss": 0.6081,
+      "step": 68670
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002477447368421052,
+      "loss": 0.6058,
+      "step": 68680
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002476657894736842,
+      "loss": 0.6156,
+      "step": 68690
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002475868421052632,
+      "loss": 0.6117,
+      "step": 68700
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00024750789473684207,
+      "loss": 0.6123,
+      "step": 68710
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.000247428947368421,
+      "loss": 0.611,
+      "step": 68720
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00024734999999999997,
+      "loss": 0.6127,
+      "step": 68730
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002472710526315789,
+      "loss": 0.6079,
+      "step": 68740
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00024719210526315786,
+      "loss": 0.6086,
+      "step": 68750
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002471131578947368,
+      "loss": 0.6038,
+      "step": 68760
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00024703421052631576,
+      "loss": 0.6027,
+      "step": 68770
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002469552631578947,
+      "loss": 0.604,
+      "step": 68780
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00024687631578947366,
+      "loss": 0.6101,
+      "step": 68790
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002467973684210526,
+      "loss": 0.5935,
+      "step": 68800
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00024671842105263155,
+      "loss": 0.613,
+      "step": 68810
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002466394736842105,
+      "loss": 0.6102,
+      "step": 68820
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00024656052631578945,
+      "loss": 0.6134,
+      "step": 68830
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002464815789473684,
+      "loss": 0.6165,
+      "step": 68840
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00024640263157894735,
+      "loss": 0.6066,
+      "step": 68850
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002463236842105263,
+      "loss": 0.5881,
+      "step": 68860
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00024624473684210524,
+      "loss": 0.6132,
+      "step": 68870
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002461657894736842,
+      "loss": 0.6242,
+      "step": 68880
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00024608684210526314,
+      "loss": 0.6087,
+      "step": 68890
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002460078947368421,
+      "loss": 0.5976,
+      "step": 68900
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00024592894736842104,
+      "loss": 0.5987,
+      "step": 68910
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00024585,
+      "loss": 0.6161,
+      "step": 68920
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00024577105263157893,
+      "loss": 0.6109,
+      "step": 68930
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002456921052631579,
+      "loss": 0.6189,
+      "step": 68940
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00024561315789473683,
+      "loss": 0.6157,
+      "step": 68950
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002455342105263158,
+      "loss": 0.6059,
+      "step": 68960
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002454552631578947,
+      "loss": 0.6033,
+      "step": 68970
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002453763157894737,
+      "loss": 0.6102,
+      "step": 68980
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002452973684210526,
+      "loss": 0.6003,
+      "step": 68990
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00024521842105263157,
+      "loss": 0.6109,
+      "step": 69000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002451394736842105,
+      "loss": 0.604,
+      "step": 69010
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00024506052631578947,
+      "loss": 0.6056,
+      "step": 69020
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002449815789473684,
+      "loss": 0.6088,
+      "step": 69030
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00024490263157894737,
+      "loss": 0.5902,
+      "step": 69040
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00024482368421052626,
+      "loss": 0.6112,
+      "step": 69050
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002447447368421052,
+      "loss": 0.611,
+      "step": 69060
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002446657894736842,
+      "loss": 0.6094,
+      "step": 69070
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00024458684210526316,
+      "loss": 0.6069,
+      "step": 69080
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002445078947368421,
+      "loss": 0.6074,
+      "step": 69090
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00024442894736842106,
+      "loss": 0.6131,
+      "step": 69100
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00024435,
+      "loss": 0.5869,
+      "step": 69110
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002442710526315789,
+      "loss": 0.6104,
+      "step": 69120
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00024419210526315785,
+      "loss": 0.6049,
+      "step": 69130
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00024411315789473682,
+      "loss": 0.6167,
+      "step": 69140
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00024403421052631574,
+      "loss": 0.6033,
+      "step": 69150
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00024395526315789474,
+      "loss": 0.6039,
+      "step": 69160
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00024387631578947367,
+      "loss": 0.5829,
+      "step": 69170
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00024379736842105261,
+      "loss": 0.5959,
+      "step": 69180
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00024371842105263156,
+      "loss": 0.6027,
+      "step": 69190
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002436394736842105,
+      "loss": 0.6057,
+      "step": 69200
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00024356052631578946,
+      "loss": 0.5997,
+      "step": 69210
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002434815789473684,
+      "loss": 0.6002,
+      "step": 69220
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00024340263157894733,
+      "loss": 0.5976,
+      "step": 69230
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00024332368421052628,
+      "loss": 0.607,
+      "step": 69240
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00024324473684210523,
+      "loss": 0.6073,
+      "step": 69250
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002431657894736842,
+      "loss": 0.5946,
+      "step": 69260
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00024308684210526315,
+      "loss": 0.6051,
+      "step": 69270
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002430078947368421,
+      "loss": 0.598,
+      "step": 69280
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00024292894736842105,
+      "loss": 0.602,
+      "step": 69290
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00024284999999999997,
+      "loss": 0.5844,
+      "step": 69300
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00024277105263157892,
+      "loss": 0.6098,
+      "step": 69310
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00024269210526315786,
+      "loss": 0.6077,
+      "step": 69320
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002426131578947368,
+      "loss": 0.6136,
+      "step": 69330
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00024253421052631576,
+      "loss": 0.6044,
+      "step": 69340
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00024245526315789474,
+      "loss": 0.6037,
+      "step": 69350
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00024237631578947368,
+      "loss": 0.6013,
+      "step": 69360
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002422973684210526,
+      "loss": 0.5935,
+      "step": 69370
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00024221842105263155,
+      "loss": 0.6034,
+      "step": 69380
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002421394736842105,
+      "loss": 0.6045,
+      "step": 69390
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00024206052631578945,
+      "loss": 0.5987,
+      "step": 69400
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002419815789473684,
+      "loss": 0.6052,
+      "step": 69410
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00024190263157894735,
+      "loss": 0.6117,
+      "step": 69420
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00024182368421052627,
+      "loss": 0.5942,
+      "step": 69430
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00024174473684210527,
+      "loss": 0.5926,
+      "step": 69440
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002416657894736842,
+      "loss": 0.605,
+      "step": 69450
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00024158684210526314,
+      "loss": 0.5998,
+      "step": 69460
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002415078947368421,
+      "loss": 0.6041,
+      "step": 69470
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00024142894736842104,
+      "loss": 0.5895,
+      "step": 69480
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00024134999999999998,
+      "loss": 0.5974,
+      "step": 69490
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00024127105263157893,
+      "loss": 0.5932,
+      "step": 69500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00024119210526315785,
+      "loss": 0.6043,
+      "step": 69510
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002411131578947368,
+      "loss": 0.605,
+      "step": 69520
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00024103421052631575,
+      "loss": 0.6021,
+      "step": 69530
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00024095526315789473,
+      "loss": 0.5918,
+      "step": 69540
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00024087631578947367,
+      "loss": 0.5874,
+      "step": 69550
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00024079736842105262,
+      "loss": 0.5916,
+      "step": 69560
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00024071842105263157,
+      "loss": 0.6072,
+      "step": 69570
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002406394736842105,
+      "loss": 0.5992,
+      "step": 69580
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00024056052631578944,
+      "loss": 0.5925,
+      "step": 69590
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002404815789473684,
+      "loss": 0.5935,
+      "step": 69600
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00024040263157894734,
+      "loss": 0.5945,
+      "step": 69610
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00024032368421052629,
+      "loss": 0.6028,
+      "step": 69620
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00024024473684210526,
+      "loss": 0.5978,
+      "step": 69630
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002401657894736842,
+      "loss": 0.6066,
+      "step": 69640
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00024008684210526316,
+      "loss": 0.5957,
+      "step": 69650
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00024000789473684208,
+      "loss": 0.5925,
+      "step": 69660
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023992894736842103,
+      "loss": 0.592,
+      "step": 69670
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023984999999999998,
+      "loss": 0.5931,
+      "step": 69680
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023977105263157892,
+      "loss": 0.607,
+      "step": 69690
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023969210526315787,
+      "loss": 0.5953,
+      "step": 69700
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023961315789473682,
+      "loss": 0.5925,
+      "step": 69710
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002395342105263158,
+      "loss": 0.6155,
+      "step": 69720
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023945526315789472,
+      "loss": 0.6088,
+      "step": 69730
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023937631578947367,
+      "loss": 0.5937,
+      "step": 69740
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002392973684210526,
+      "loss": 0.6061,
+      "step": 69750
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023921842105263156,
+      "loss": 0.6011,
+      "step": 69760
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002391394736842105,
+      "loss": 0.6094,
+      "step": 69770
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023906052631578946,
+      "loss": 0.6065,
+      "step": 69780
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023898157894736838,
+      "loss": 0.5998,
+      "step": 69790
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023890263157894733,
+      "loss": 0.6013,
+      "step": 69800
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023882368421052628,
+      "loss": 0.6027,
+      "step": 69810
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023874473684210525,
+      "loss": 0.6078,
+      "step": 69820
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002386657894736842,
+      "loss": 0.6176,
+      "step": 69830
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023858684210526315,
+      "loss": 0.6084,
+      "step": 69840
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002385078947368421,
+      "loss": 0.605,
+      "step": 69850
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023842894736842102,
+      "loss": 0.6058,
+      "step": 69860
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023834999999999997,
+      "loss": 0.5942,
+      "step": 69870
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023827105263157891,
+      "loss": 0.5972,
+      "step": 69880
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023819210526315786,
+      "loss": 0.5919,
+      "step": 69890
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002381131578947368,
+      "loss": 0.5953,
+      "step": 69900
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023803421052631579,
+      "loss": 0.5834,
+      "step": 69910
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023795526315789473,
+      "loss": 0.5894,
+      "step": 69920
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023787631578947368,
+      "loss": 0.5811,
+      "step": 69930
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002378052631578947,
+      "loss": 0.6117,
+      "step": 69940
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023772631578947365,
+      "loss": 0.6008,
+      "step": 69950
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002376473684210526,
+      "loss": 0.6081,
+      "step": 69960
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023756842105263158,
+      "loss": 0.6041,
+      "step": 69970
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023748947368421052,
+      "loss": 0.6165,
+      "step": 69980
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023741052631578947,
+      "loss": 0.5945,
+      "step": 69990
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002373315789473684,
+      "loss": 0.6078,
+      "step": 70000
+    },
+    {
+      "epoch": 0.7,
+      "eval_accuracy": 0.8734435544718635,
+      "eval_loss": 0.583984375,
+      "eval_runtime": 98.2973,
+      "eval_samples_per_second": 813.857,
+      "eval_steps_per_second": 1.597,
+      "step": 70000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023725263157894734,
+      "loss": 0.5939,
+      "step": 70010
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002371736842105263,
+      "loss": 0.6039,
+      "step": 70020
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023709473684210524,
+      "loss": 0.5968,
+      "step": 70030
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002370157894736842,
+      "loss": 0.5757,
+      "step": 70040
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023693684210526314,
+      "loss": 0.5833,
+      "step": 70050
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023685789473684206,
+      "loss": 0.5848,
+      "step": 70060
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023677894736842106,
+      "loss": 0.6055,
+      "step": 70070
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023669999999999998,
+      "loss": 0.6161,
+      "step": 70080
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023662105263157893,
+      "loss": 0.5991,
+      "step": 70090
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023654210526315788,
+      "loss": 0.6033,
+      "step": 70100
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023646315789473682,
+      "loss": 0.6065,
+      "step": 70110
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023638421052631577,
+      "loss": 0.6057,
+      "step": 70120
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023630526315789472,
+      "loss": 0.6078,
+      "step": 70130
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023622631578947364,
+      "loss": 0.6102,
+      "step": 70140
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002361473684210526,
+      "loss": 0.6025,
+      "step": 70150
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023606842105263157,
+      "loss": 0.5977,
+      "step": 70160
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023598947368421051,
+      "loss": 0.6034,
+      "step": 70170
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023591052631578946,
+      "loss": 0.6022,
+      "step": 70180
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002358315789473684,
+      "loss": 0.6029,
+      "step": 70190
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023575263157894736,
+      "loss": 0.5929,
+      "step": 70200
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023567368421052628,
+      "loss": 0.6114,
+      "step": 70210
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023559473684210523,
+      "loss": 0.6118,
+      "step": 70220
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023551578947368418,
+      "loss": 0.6014,
+      "step": 70230
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023543684210526313,
+      "loss": 0.595,
+      "step": 70240
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002353578947368421,
+      "loss": 0.5955,
+      "step": 70250
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023527894736842105,
+      "loss": 0.6039,
+      "step": 70260
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002352,
+      "loss": 0.6257,
+      "step": 70270
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023512105263157892,
+      "loss": 0.6124,
+      "step": 70280
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023504210526315787,
+      "loss": 0.6083,
+      "step": 70290
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023496315789473682,
+      "loss": 0.6003,
+      "step": 70300
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023488421052631576,
+      "loss": 0.6006,
+      "step": 70310
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002348052631578947,
+      "loss": 0.5982,
+      "step": 70320
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023472631578947366,
+      "loss": 0.6025,
+      "step": 70330
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023464736842105258,
+      "loss": 0.6044,
+      "step": 70340
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023456842105263158,
+      "loss": 0.5941,
+      "step": 70350
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002344894736842105,
+      "loss": 0.597,
+      "step": 70360
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023441052631578945,
+      "loss": 0.5882,
+      "step": 70370
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002343315789473684,
+      "loss": 0.5973,
+      "step": 70380
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023425263157894735,
+      "loss": 0.5996,
+      "step": 70390
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002341736842105263,
+      "loss": 0.6067,
+      "step": 70400
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023409473684210525,
+      "loss": 0.6058,
+      "step": 70410
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023401578947368417,
+      "loss": 0.6062,
+      "step": 70420
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023393684210526312,
+      "loss": 0.6052,
+      "step": 70430
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002338578947368421,
+      "loss": 0.5923,
+      "step": 70440
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023377894736842104,
+      "loss": 0.6033,
+      "step": 70450
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002337,
+      "loss": 0.6068,
+      "step": 70460
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023362105263157894,
+      "loss": 0.5843,
+      "step": 70470
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023354210526315789,
+      "loss": 0.5965,
+      "step": 70480
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002334631578947368,
+      "loss": 0.5903,
+      "step": 70490
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023338421052631575,
+      "loss": 0.6136,
+      "step": 70500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002333052631578947,
+      "loss": 0.6037,
+      "step": 70510
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00023322631578947365,
+      "loss": 0.613,
+      "step": 70520
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00023314736842105263,
+      "loss": 0.6103,
+      "step": 70530
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00023306842105263157,
+      "loss": 0.5961,
+      "step": 70540
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00023298947368421052,
+      "loss": 0.6017,
+      "step": 70550
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00023291052631578947,
+      "loss": 0.6015,
+      "step": 70560
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002328315789473684,
+      "loss": 0.6025,
+      "step": 70570
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00023275263157894734,
+      "loss": 0.5949,
+      "step": 70580
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002326736842105263,
+      "loss": 0.6025,
+      "step": 70590
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00023259473684210524,
+      "loss": 0.6016,
+      "step": 70600
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00023251578947368419,
+      "loss": 0.5956,
+      "step": 70610
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00023243684210526313,
+      "loss": 0.597,
+      "step": 70620
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002323578947368421,
+      "loss": 0.6034,
+      "step": 70630
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00023227894736842103,
+      "loss": 0.5926,
+      "step": 70640
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00023219999999999998,
+      "loss": 0.5916,
+      "step": 70650
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00023212105263157893,
+      "loss": 0.5967,
+      "step": 70660
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00023204210526315788,
+      "loss": 0.6008,
+      "step": 70670
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00023196315789473682,
+      "loss": 0.5852,
+      "step": 70680
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00023188421052631577,
+      "loss": 0.5923,
+      "step": 70690
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002318052631578947,
+      "loss": 0.5933,
+      "step": 70700
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00023172631578947364,
+      "loss": 0.6051,
+      "step": 70710
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00023164736842105262,
+      "loss": 0.5979,
+      "step": 70720
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00023156842105263157,
+      "loss": 0.5987,
+      "step": 70730
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00023148947368421051,
+      "loss": 0.5959,
+      "step": 70740
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00023141052631578946,
+      "loss": 0.5904,
+      "step": 70750
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002313315789473684,
+      "loss": 0.6042,
+      "step": 70760
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00023125263157894733,
+      "loss": 0.6005,
+      "step": 70770
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00023117368421052628,
+      "loss": 0.6,
+      "step": 70780
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00023109473684210523,
+      "loss": 0.5935,
+      "step": 70790
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00023101578947368418,
+      "loss": 0.608,
+      "step": 70800
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00023093684210526315,
+      "loss": 0.6092,
+      "step": 70810
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002308578947368421,
+      "loss": 0.5978,
+      "step": 70820
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00023077894736842105,
+      "loss": 0.6018,
+      "step": 70830
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002307,
+      "loss": 0.6036,
+      "step": 70840
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00023062105263157892,
+      "loss": 0.6035,
+      "step": 70850
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00023054210526315787,
+      "loss": 0.5986,
+      "step": 70860
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00023046315789473681,
+      "loss": 0.6031,
+      "step": 70870
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00023038421052631576,
+      "loss": 0.5977,
+      "step": 70880
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002303052631578947,
+      "loss": 0.6025,
+      "step": 70890
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00023022631578947366,
+      "loss": 0.594,
+      "step": 70900
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00023014736842105263,
+      "loss": 0.614,
+      "step": 70910
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00023006842105263156,
+      "loss": 0.5989,
+      "step": 70920
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002299894736842105,
+      "loss": 0.595,
+      "step": 70930
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00022991842105263155,
+      "loss": 0.5981,
+      "step": 70940
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002298394736842105,
+      "loss": 0.5895,
+      "step": 70950
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00022976052631578945,
+      "loss": 0.6078,
+      "step": 70960
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00022968157894736842,
+      "loss": 0.6083,
+      "step": 70970
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00022960263157894737,
+      "loss": 0.6098,
+      "step": 70980
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002295236842105263,
+      "loss": 0.5887,
+      "step": 70990
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00022944473684210524,
+      "loss": 0.5875,
+      "step": 71000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002293657894736842,
+      "loss": 0.6078,
+      "step": 71010
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00022928684210526314,
+      "loss": 0.5973,
+      "step": 71020
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002292078947368421,
+      "loss": 0.6166,
+      "step": 71030
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00022912894736842104,
+      "loss": 0.6056,
+      "step": 71040
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00022904999999999996,
+      "loss": 0.5935,
+      "step": 71050
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002289710526315789,
+      "loss": 0.5915,
+      "step": 71060
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00022889210526315788,
+      "loss": 0.6014,
+      "step": 71070
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00022881315789473683,
+      "loss": 0.601,
+      "step": 71080
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00022873421052631578,
+      "loss": 0.6017,
+      "step": 71090
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00022865526315789473,
+      "loss": 0.6092,
+      "step": 71100
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00022857631578947367,
+      "loss": 0.6009,
+      "step": 71110
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002284973684210526,
+      "loss": 0.5998,
+      "step": 71120
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00022841842105263154,
+      "loss": 0.5892,
+      "step": 71130
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002283394736842105,
+      "loss": 0.5822,
+      "step": 71140
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00022826052631578944,
+      "loss": 0.5975,
+      "step": 71150
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00022818157894736842,
+      "loss": 0.617,
+      "step": 71160
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00022810263157894736,
+      "loss": 0.5986,
+      "step": 71170
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002280236842105263,
+      "loss": 0.5952,
+      "step": 71180
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00022794473684210526,
+      "loss": 0.6105,
+      "step": 71190
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00022786578947368418,
+      "loss": 0.6043,
+      "step": 71200
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00022778684210526313,
+      "loss": 0.6116,
+      "step": 71210
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00022770789473684208,
+      "loss": 0.6148,
+      "step": 71220
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00022762894736842103,
+      "loss": 0.611,
+      "step": 71230
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00022754999999999997,
+      "loss": 0.6016,
+      "step": 71240
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00022747105263157895,
+      "loss": 0.6035,
+      "step": 71250
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002273921052631579,
+      "loss": 0.5876,
+      "step": 71260
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00022731315789473682,
+      "loss": 0.5937,
+      "step": 71270
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00022723421052631577,
+      "loss": 0.5963,
+      "step": 71280
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00022715526315789472,
+      "loss": 0.5965,
+      "step": 71290
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00022707631578947366,
+      "loss": 0.5937,
+      "step": 71300
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002269973684210526,
+      "loss": 0.5938,
+      "step": 71310
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00022691842105263156,
+      "loss": 0.5797,
+      "step": 71320
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00022683947368421048,
+      "loss": 0.5936,
+      "step": 71330
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00022676052631578946,
+      "loss": 0.5935,
+      "step": 71340
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002266815789473684,
+      "loss": 0.5929,
+      "step": 71350
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00022660263157894735,
+      "loss": 0.5987,
+      "step": 71360
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002265236842105263,
+      "loss": 0.5836,
+      "step": 71370
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00022644473684210525,
+      "loss": 0.576,
+      "step": 71380
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002263657894736842,
+      "loss": 0.5896,
+      "step": 71390
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00022628684210526312,
+      "loss": 0.5827,
+      "step": 71400
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00022620789473684207,
+      "loss": 0.593,
+      "step": 71410
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00022612894736842102,
+      "loss": 0.5809,
+      "step": 71420
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00022604999999999997,
+      "loss": 0.5905,
+      "step": 71430
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00022597105263157894,
+      "loss": 0.5885,
+      "step": 71440
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002258921052631579,
+      "loss": 0.5924,
+      "step": 71450
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00022581315789473684,
+      "loss": 0.5871,
+      "step": 71460
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00022573421052631579,
+      "loss": 0.5978,
+      "step": 71470
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002256552631578947,
+      "loss": 0.5868,
+      "step": 71480
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00022557631578947365,
+      "loss": 0.596,
+      "step": 71490
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002254973684210526,
+      "loss": 0.5952,
+      "step": 71500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022541842105263155,
+      "loss": 0.5963,
+      "step": 71510
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002253394736842105,
+      "loss": 0.6128,
+      "step": 71520
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022526052631578948,
+      "loss": 0.6038,
+      "step": 71530
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022518157894736842,
+      "loss": 0.6014,
+      "step": 71540
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022510263157894734,
+      "loss": 0.6037,
+      "step": 71550
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002250236842105263,
+      "loss": 0.6062,
+      "step": 71560
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022494473684210524,
+      "loss": 0.5995,
+      "step": 71570
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002248657894736842,
+      "loss": 0.6024,
+      "step": 71580
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022478684210526314,
+      "loss": 0.599,
+      "step": 71590
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022470789473684209,
+      "loss": 0.6094,
+      "step": 71600
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.000224628947368421,
+      "loss": 0.6053,
+      "step": 71610
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022455,
+      "loss": 0.5928,
+      "step": 71620
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022447105263157893,
+      "loss": 0.6068,
+      "step": 71630
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022439210526315788,
+      "loss": 0.602,
+      "step": 71640
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022431315789473683,
+      "loss": 0.6049,
+      "step": 71650
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022423421052631578,
+      "loss": 0.6079,
+      "step": 71660
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022415526315789472,
+      "loss": 0.5928,
+      "step": 71670
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022407631578947367,
+      "loss": 0.5994,
+      "step": 71680
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002239973684210526,
+      "loss": 0.597,
+      "step": 71690
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022391842105263154,
+      "loss": 0.6096,
+      "step": 71700
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002238394736842105,
+      "loss": 0.6095,
+      "step": 71710
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022376052631578947,
+      "loss": 0.6015,
+      "step": 71720
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022368157894736841,
+      "loss": 0.6118,
+      "step": 71730
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022360263157894736,
+      "loss": 0.6141,
+      "step": 71740
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002235236842105263,
+      "loss": 0.6042,
+      "step": 71750
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022344473684210523,
+      "loss": 0.608,
+      "step": 71760
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022336578947368418,
+      "loss": 0.6134,
+      "step": 71770
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022328684210526313,
+      "loss": 0.6184,
+      "step": 71780
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022320789473684208,
+      "loss": 0.6086,
+      "step": 71790
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022312894736842103,
+      "loss": 0.6171,
+      "step": 71800
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022305,
+      "loss": 0.6145,
+      "step": 71810
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022297105263157895,
+      "loss": 0.6011,
+      "step": 71820
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022289210526315787,
+      "loss": 0.6098,
+      "step": 71830
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022281315789473682,
+      "loss": 0.6053,
+      "step": 71840
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022273421052631577,
+      "loss": 0.6126,
+      "step": 71850
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022265526315789472,
+      "loss": 0.5937,
+      "step": 71860
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022257631578947366,
+      "loss": 0.6048,
+      "step": 71870
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002224973684210526,
+      "loss": 0.6046,
+      "step": 71880
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022241842105263153,
+      "loss": 0.6002,
+      "step": 71890
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022233947368421054,
+      "loss": 0.5936,
+      "step": 71900
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022226052631578946,
+      "loss": 0.6053,
+      "step": 71910
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002221815789473684,
+      "loss": 0.5948,
+      "step": 71920
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022210263157894735,
+      "loss": 0.5856,
+      "step": 71930
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002220315789473684,
+      "loss": 0.5845,
+      "step": 71940
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022195263157894735,
+      "loss": 0.5855,
+      "step": 71950
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022187368421052627,
+      "loss": 0.5879,
+      "step": 71960
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022179473684210525,
+      "loss": 0.582,
+      "step": 71970
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002217157894736842,
+      "loss": 0.5874,
+      "step": 71980
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022163684210526314,
+      "loss": 0.5876,
+      "step": 71990
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002215578947368421,
+      "loss": 0.5806,
+      "step": 72000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022147894736842104,
+      "loss": 0.5831,
+      "step": 72010
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002214,
+      "loss": 0.58,
+      "step": 72020
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002213210526315789,
+      "loss": 0.5838,
+      "step": 72030
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022124210526315786,
+      "loss": 0.5866,
+      "step": 72040
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002211631578947368,
+      "loss": 0.5803,
+      "step": 72050
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022108421052631578,
+      "loss": 0.5848,
+      "step": 72060
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022100526315789473,
+      "loss": 0.5816,
+      "step": 72070
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022092631578947368,
+      "loss": 0.5852,
+      "step": 72080
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022084736842105263,
+      "loss": 0.5868,
+      "step": 72090
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022076842105263157,
+      "loss": 0.5966,
+      "step": 72100
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002206894736842105,
+      "loss": 0.5947,
+      "step": 72110
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022061052631578944,
+      "loss": 0.5872,
+      "step": 72120
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002205315789473684,
+      "loss": 0.5674,
+      "step": 72130
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022045263157894734,
+      "loss": 0.5797,
+      "step": 72140
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002203736842105263,
+      "loss": 0.6056,
+      "step": 72150
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022029473684210526,
+      "loss": 0.5937,
+      "step": 72160
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002202157894736842,
+      "loss": 0.5895,
+      "step": 72170
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022013684210526313,
+      "loss": 0.5959,
+      "step": 72180
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022005789473684208,
+      "loss": 0.5931,
+      "step": 72190
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00021997894736842103,
+      "loss": 0.5929,
+      "step": 72200
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00021989999999999998,
+      "loss": 0.586,
+      "step": 72210
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00021982105263157893,
+      "loss": 0.5836,
+      "step": 72220
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00021974210526315787,
+      "loss": 0.6077,
+      "step": 72230
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002196631578947368,
+      "loss": 0.5922,
+      "step": 72240
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00021958421052631577,
+      "loss": 0.597,
+      "step": 72250
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00021950526315789472,
+      "loss": 0.5902,
+      "step": 72260
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00021942631578947367,
+      "loss": 0.5983,
+      "step": 72270
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00021934736842105262,
+      "loss": 0.5861,
+      "step": 72280
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00021926842105263156,
+      "loss": 0.6096,
+      "step": 72290
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002191894736842105,
+      "loss": 0.6019,
+      "step": 72300
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00021911052631578943,
+      "loss": 0.6009,
+      "step": 72310
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00021903157894736838,
+      "loss": 0.5955,
+      "step": 72320
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00021895263157894733,
+      "loss": 0.5925,
+      "step": 72330
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002188736842105263,
+      "loss": 0.6106,
+      "step": 72340
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00021879473684210525,
+      "loss": 0.6014,
+      "step": 72350
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002187157894736842,
+      "loss": 0.5975,
+      "step": 72360
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00021863684210526315,
+      "loss": 0.595,
+      "step": 72370
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002185578947368421,
+      "loss": 0.5984,
+      "step": 72380
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00021847894736842102,
+      "loss": 0.5999,
+      "step": 72390
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00021839999999999997,
+      "loss": 0.5956,
+      "step": 72400
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00021832105263157892,
+      "loss": 0.5924,
+      "step": 72410
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00021824210526315787,
+      "loss": 0.5907,
+      "step": 72420
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00021816315789473681,
+      "loss": 0.5842,
+      "step": 72430
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002180842105263158,
+      "loss": 0.5978,
+      "step": 72440
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00021800526315789474,
+      "loss": 0.5884,
+      "step": 72450
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00021792631578947366,
+      "loss": 0.5898,
+      "step": 72460
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002178473684210526,
+      "loss": 0.5941,
+      "step": 72470
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00021776842105263156,
+      "loss": 0.6033,
+      "step": 72480
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002176894736842105,
+      "loss": 0.6142,
+      "step": 72490
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00021761052631578945,
+      "loss": 0.5965,
+      "step": 72500
+    },
+    {
+      "epoch": 0.72,
+      "eval_accuracy": 0.8749469840503773,
+      "eval_loss": 0.576171875,
+      "eval_runtime": 99.6694,
+      "eval_samples_per_second": 802.653,
+      "eval_steps_per_second": 1.575,
+      "step": 72500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002175315789473684,
+      "loss": 0.5961,
+      "step": 72510
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021745263157894732,
+      "loss": 0.6001,
+      "step": 72520
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021737368421052632,
+      "loss": 0.6151,
+      "step": 72530
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021729473684210525,
+      "loss": 0.6152,
+      "step": 72540
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002172157894736842,
+      "loss": 0.5974,
+      "step": 72550
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021713684210526314,
+      "loss": 0.6027,
+      "step": 72560
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002170578947368421,
+      "loss": 0.6109,
+      "step": 72570
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021697894736842104,
+      "loss": 0.5968,
+      "step": 72580
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002169,
+      "loss": 0.6037,
+      "step": 72590
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002168210526315789,
+      "loss": 0.5886,
+      "step": 72600
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021674210526315786,
+      "loss": 0.5981,
+      "step": 72610
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021666315789473683,
+      "loss": 0.6046,
+      "step": 72620
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021658421052631578,
+      "loss": 0.5896,
+      "step": 72630
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021650526315789473,
+      "loss": 0.5975,
+      "step": 72640
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021642631578947368,
+      "loss": 0.5955,
+      "step": 72650
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021634736842105262,
+      "loss": 0.592,
+      "step": 72660
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021626842105263155,
+      "loss": 0.6062,
+      "step": 72670
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002161894736842105,
+      "loss": 0.596,
+      "step": 72680
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021611052631578944,
+      "loss": 0.6005,
+      "step": 72690
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002160315789473684,
+      "loss": 0.6064,
+      "step": 72700
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021595263157894737,
+      "loss": 0.5907,
+      "step": 72710
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021587368421052631,
+      "loss": 0.5921,
+      "step": 72720
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021579473684210526,
+      "loss": 0.6098,
+      "step": 72730
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021571578947368418,
+      "loss": 0.5913,
+      "step": 72740
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021563684210526313,
+      "loss": 0.5979,
+      "step": 72750
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021555789473684208,
+      "loss": 0.6038,
+      "step": 72760
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021547894736842103,
+      "loss": 0.6031,
+      "step": 72770
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021539999999999998,
+      "loss": 0.6046,
+      "step": 72780
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021532105263157893,
+      "loss": 0.6048,
+      "step": 72790
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021524210526315785,
+      "loss": 0.5917,
+      "step": 72800
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021516315789473685,
+      "loss": 0.5912,
+      "step": 72810
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021508421052631577,
+      "loss": 0.6047,
+      "step": 72820
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021500526315789472,
+      "loss": 0.5936,
+      "step": 72830
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021492631578947367,
+      "loss": 0.5931,
+      "step": 72840
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021484736842105262,
+      "loss": 0.5977,
+      "step": 72850
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021476842105263156,
+      "loss": 0.6,
+      "step": 72860
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002146894736842105,
+      "loss": 0.5975,
+      "step": 72870
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021461052631578943,
+      "loss": 0.5967,
+      "step": 72880
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021453157894736838,
+      "loss": 0.6096,
+      "step": 72890
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021445263157894736,
+      "loss": 0.5931,
+      "step": 72900
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002143736842105263,
+      "loss": 0.5879,
+      "step": 72910
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021429473684210525,
+      "loss": 0.5894,
+      "step": 72920
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002142157894736842,
+      "loss": 0.5986,
+      "step": 72930
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021414473684210522,
+      "loss": 0.585,
+      "step": 72940
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021406578947368417,
+      "loss": 0.5892,
+      "step": 72950
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021398684210526312,
+      "loss": 0.5951,
+      "step": 72960
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002139078947368421,
+      "loss": 0.596,
+      "step": 72970
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021382894736842104,
+      "loss": 0.596,
+      "step": 72980
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021375,
+      "loss": 0.5977,
+      "step": 72990
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021367105263157894,
+      "loss": 0.593,
+      "step": 73000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002135921052631579,
+      "loss": 0.5901,
+      "step": 73010
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002135131578947368,
+      "loss": 0.5919,
+      "step": 73020
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021343421052631576,
+      "loss": 0.5984,
+      "step": 73030
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002133552631578947,
+      "loss": 0.597,
+      "step": 73040
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021327631578947365,
+      "loss": 0.6008,
+      "step": 73050
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021319736842105263,
+      "loss": 0.5879,
+      "step": 73060
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021311842105263158,
+      "loss": 0.5941,
+      "step": 73070
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021303947368421053,
+      "loss": 0.588,
+      "step": 73080
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021296052631578945,
+      "loss": 0.6014,
+      "step": 73090
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002128815789473684,
+      "loss": 0.6001,
+      "step": 73100
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021280263157894734,
+      "loss": 0.5957,
+      "step": 73110
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002127236842105263,
+      "loss": 0.5876,
+      "step": 73120
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021264473684210524,
+      "loss": 0.59,
+      "step": 73130
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002125657894736842,
+      "loss": 0.6049,
+      "step": 73140
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021248684210526316,
+      "loss": 0.597,
+      "step": 73150
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002124078947368421,
+      "loss": 0.604,
+      "step": 73160
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021232894736842103,
+      "loss": 0.6002,
+      "step": 73170
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021224999999999998,
+      "loss": 0.5974,
+      "step": 73180
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021217105263157893,
+      "loss": 0.5976,
+      "step": 73190
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021209210526315788,
+      "loss": 0.5933,
+      "step": 73200
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021201315789473683,
+      "loss": 0.5826,
+      "step": 73210
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021193421052631578,
+      "loss": 0.5848,
+      "step": 73220
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002118552631578947,
+      "loss": 0.5892,
+      "step": 73230
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021177631578947364,
+      "loss": 0.5816,
+      "step": 73240
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021169736842105262,
+      "loss": 0.5683,
+      "step": 73250
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021161842105263157,
+      "loss": 0.5969,
+      "step": 73260
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021154736842105262,
+      "loss": 0.5983,
+      "step": 73270
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021146842105263156,
+      "loss": 0.5935,
+      "step": 73280
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021138947368421049,
+      "loss": 0.6056,
+      "step": 73290
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021131052631578943,
+      "loss": 0.5983,
+      "step": 73300
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002112315789473684,
+      "loss": 0.6023,
+      "step": 73310
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021115263157894736,
+      "loss": 0.591,
+      "step": 73320
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002110736842105263,
+      "loss": 0.5914,
+      "step": 73330
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021099473684210525,
+      "loss": 0.5944,
+      "step": 73340
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002109157894736842,
+      "loss": 0.5988,
+      "step": 73350
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021083684210526312,
+      "loss": 0.587,
+      "step": 73360
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021075789473684207,
+      "loss": 0.5785,
+      "step": 73370
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021067894736842102,
+      "loss": 0.6024,
+      "step": 73380
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021059999999999997,
+      "loss": 0.5973,
+      "step": 73390
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021052105263157892,
+      "loss": 0.5994,
+      "step": 73400
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002104421052631579,
+      "loss": 0.5964,
+      "step": 73410
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021036315789473684,
+      "loss": 0.5854,
+      "step": 73420
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002102842105263158,
+      "loss": 0.5947,
+      "step": 73430
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002102052631578947,
+      "loss": 0.5925,
+      "step": 73440
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021012631578947366,
+      "loss": 0.5833,
+      "step": 73450
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002100473684210526,
+      "loss": 0.5844,
+      "step": 73460
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00020996842105263156,
+      "loss": 0.5915,
+      "step": 73470
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002098894736842105,
+      "loss": 0.5919,
+      "step": 73480
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00020981052631578945,
+      "loss": 0.5949,
+      "step": 73490
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00020973157894736843,
+      "loss": 0.5937,
+      "step": 73500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020965263157894735,
+      "loss": 0.5894,
+      "step": 73510
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002095736842105263,
+      "loss": 0.5986,
+      "step": 73520
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020949473684210525,
+      "loss": 0.5838,
+      "step": 73530
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002094157894736842,
+      "loss": 0.5831,
+      "step": 73540
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020933684210526314,
+      "loss": 0.5878,
+      "step": 73550
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002092578947368421,
+      "loss": 0.5914,
+      "step": 73560
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000209178947368421,
+      "loss": 0.5783,
+      "step": 73570
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020909999999999996,
+      "loss": 0.599,
+      "step": 73580
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020902105263157894,
+      "loss": 0.59,
+      "step": 73590
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020894210526315788,
+      "loss": 0.5932,
+      "step": 73600
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020886315789473683,
+      "loss": 0.5887,
+      "step": 73610
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020878421052631578,
+      "loss": 0.5971,
+      "step": 73620
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020870526315789473,
+      "loss": 0.5933,
+      "step": 73630
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020862631578947368,
+      "loss": 0.5931,
+      "step": 73640
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002085473684210526,
+      "loss": 0.5976,
+      "step": 73650
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020846842105263155,
+      "loss": 0.5878,
+      "step": 73660
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002083894736842105,
+      "loss": 0.5901,
+      "step": 73670
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020831052631578944,
+      "loss": 0.5919,
+      "step": 73680
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020823157894736842,
+      "loss": 0.6052,
+      "step": 73690
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020815263157894737,
+      "loss": 0.5924,
+      "step": 73700
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020807368421052631,
+      "loss": 0.5885,
+      "step": 73710
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020799473684210524,
+      "loss": 0.596,
+      "step": 73720
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020791578947368418,
+      "loss": 0.5957,
+      "step": 73730
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020783684210526313,
+      "loss": 0.5959,
+      "step": 73740
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020775789473684208,
+      "loss": 0.5915,
+      "step": 73750
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020767894736842103,
+      "loss": 0.5849,
+      "step": 73760
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020759999999999998,
+      "loss": 0.6014,
+      "step": 73770
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020752105263157895,
+      "loss": 0.5915,
+      "step": 73780
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020744210526315787,
+      "loss": 0.5988,
+      "step": 73790
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020736315789473682,
+      "loss": 0.5941,
+      "step": 73800
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020728421052631577,
+      "loss": 0.5902,
+      "step": 73810
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020720526315789472,
+      "loss": 0.5847,
+      "step": 73820
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020712631578947367,
+      "loss": 0.5897,
+      "step": 73830
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020704736842105262,
+      "loss": 0.5903,
+      "step": 73840
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020696842105263154,
+      "loss": 0.592,
+      "step": 73850
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020688947368421049,
+      "loss": 0.5881,
+      "step": 73860
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020681052631578946,
+      "loss": 0.5812,
+      "step": 73870
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002067315789473684,
+      "loss": 0.5876,
+      "step": 73880
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020665263157894736,
+      "loss": 0.5995,
+      "step": 73890
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002065736842105263,
+      "loss": 0.5869,
+      "step": 73900
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020649473684210525,
+      "loss": 0.5843,
+      "step": 73910
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002064157894736842,
+      "loss": 0.5949,
+      "step": 73920
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020633684210526312,
+      "loss": 0.6077,
+      "step": 73930
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020625789473684207,
+      "loss": 0.5957,
+      "step": 73940
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020617894736842102,
+      "loss": 0.589,
+      "step": 73950
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002061,
+      "loss": 0.6004,
+      "step": 73960
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020602105263157894,
+      "loss": 0.596,
+      "step": 73970
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002059421052631579,
+      "loss": 0.5931,
+      "step": 73980
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020586315789473684,
+      "loss": 0.5855,
+      "step": 73990
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020578421052631576,
+      "loss": 0.5928,
+      "step": 74000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002057052631578947,
+      "loss": 0.5871,
+      "step": 74010
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020562631578947366,
+      "loss": 0.597,
+      "step": 74020
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002055473684210526,
+      "loss": 0.5806,
+      "step": 74030
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020546842105263155,
+      "loss": 0.5913,
+      "step": 74040
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002053894736842105,
+      "loss": 0.587,
+      "step": 74050
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020531052631578948,
+      "loss": 0.5958,
+      "step": 74060
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020523157894736843,
+      "loss": 0.5976,
+      "step": 74070
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020515263157894735,
+      "loss": 0.5926,
+      "step": 74080
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002050736842105263,
+      "loss": 0.5866,
+      "step": 74090
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020499473684210524,
+      "loss": 0.5865,
+      "step": 74100
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002049157894736842,
+      "loss": 0.596,
+      "step": 74110
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020483684210526314,
+      "loss": 0.5871,
+      "step": 74120
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002047578947368421,
+      "loss": 0.5887,
+      "step": 74130
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000204678947368421,
+      "loss": 0.58,
+      "step": 74140
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020459999999999999,
+      "loss": 0.5958,
+      "step": 74150
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020452105263157893,
+      "loss": 0.5887,
+      "step": 74160
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020444210526315788,
+      "loss": 0.6021,
+      "step": 74170
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020436315789473683,
+      "loss": 0.5973,
+      "step": 74180
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020428421052631578,
+      "loss": 0.5917,
+      "step": 74190
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020420526315789473,
+      "loss": 0.5885,
+      "step": 74200
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020412631578947365,
+      "loss": 0.5923,
+      "step": 74210
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002040473684210526,
+      "loss": 0.5991,
+      "step": 74220
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020396842105263155,
+      "loss": 0.594,
+      "step": 74230
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020388947368421052,
+      "loss": 0.5846,
+      "step": 74240
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020381052631578947,
+      "loss": 0.5927,
+      "step": 74250
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020373157894736842,
+      "loss": 0.5902,
+      "step": 74260
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020365263157894737,
+      "loss": 0.5959,
+      "step": 74270
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002035736842105263,
+      "loss": 0.595,
+      "step": 74280
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020349473684210524,
+      "loss": 0.5999,
+      "step": 74290
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020341578947368418,
+      "loss": 0.5981,
+      "step": 74300
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020333684210526313,
+      "loss": 0.5802,
+      "step": 74310
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020325789473684208,
+      "loss": 0.5898,
+      "step": 74320
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020317894736842103,
+      "loss": 0.5907,
+      "step": 74330
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002031,
+      "loss": 0.5945,
+      "step": 74340
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020302105263157895,
+      "loss": 0.5891,
+      "step": 74350
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020294210526315787,
+      "loss": 0.5939,
+      "step": 74360
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020286315789473682,
+      "loss": 0.5917,
+      "step": 74370
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020278421052631577,
+      "loss": 0.6046,
+      "step": 74380
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020270526315789472,
+      "loss": 0.5886,
+      "step": 74390
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020262631578947367,
+      "loss": 0.5927,
+      "step": 74400
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020254736842105261,
+      "loss": 0.5799,
+      "step": 74410
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020246842105263154,
+      "loss": 0.5988,
+      "step": 74420
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002023894736842105,
+      "loss": 0.5747,
+      "step": 74430
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020231052631578946,
+      "loss": 0.6025,
+      "step": 74440
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002022315789473684,
+      "loss": 0.5961,
+      "step": 74450
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020215263157894736,
+      "loss": 0.6067,
+      "step": 74460
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002020736842105263,
+      "loss": 0.5855,
+      "step": 74470
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020199473684210525,
+      "loss": 0.603,
+      "step": 74480
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020191578947368417,
+      "loss": 0.5925,
+      "step": 74490
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020183684210526312,
+      "loss": 0.5881,
+      "step": 74500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00020175789473684207,
+      "loss": 0.5895,
+      "step": 74510
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00020167894736842105,
+      "loss": 0.5814,
+      "step": 74520
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002016,
+      "loss": 0.5903,
+      "step": 74530
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00020152105263157894,
+      "loss": 0.5922,
+      "step": 74540
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002014421052631579,
+      "loss": 0.5976,
+      "step": 74550
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00020136315789473684,
+      "loss": 0.5997,
+      "step": 74560
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00020128421052631576,
+      "loss": 0.5949,
+      "step": 74570
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002012052631578947,
+      "loss": 0.5946,
+      "step": 74580
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00020112631578947366,
+      "loss": 0.5893,
+      "step": 74590
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002010473684210526,
+      "loss": 0.5897,
+      "step": 74600
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00020096842105263155,
+      "loss": 0.5852,
+      "step": 74610
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00020088947368421053,
+      "loss": 0.5922,
+      "step": 74620
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00020081052631578948,
+      "loss": 0.5941,
+      "step": 74630
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002007315789473684,
+      "loss": 0.5982,
+      "step": 74640
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00020065263157894735,
+      "loss": 0.5922,
+      "step": 74650
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002005736842105263,
+      "loss": 0.5886,
+      "step": 74660
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00020049473684210524,
+      "loss": 0.5966,
+      "step": 74670
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002004157894736842,
+      "loss": 0.5947,
+      "step": 74680
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00020033684210526314,
+      "loss": 0.601,
+      "step": 74690
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00020025789473684206,
+      "loss": 0.5969,
+      "step": 74700
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00020017894736842104,
+      "loss": 0.593,
+      "step": 74710
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00020009999999999998,
+      "loss": 0.5953,
+      "step": 74720
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00020002105263157893,
+      "loss": 0.5841,
+      "step": 74730
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019994210526315788,
+      "loss": 0.5884,
+      "step": 74740
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019986315789473683,
+      "loss": 0.5938,
+      "step": 74750
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019978421052631578,
+      "loss": 0.588,
+      "step": 74760
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001997052631578947,
+      "loss": 0.597,
+      "step": 74770
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019962631578947365,
+      "loss": 0.602,
+      "step": 74780
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001995473684210526,
+      "loss": 0.6009,
+      "step": 74790
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019946842105263157,
+      "loss": 0.5925,
+      "step": 74800
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019938947368421052,
+      "loss": 0.6012,
+      "step": 74810
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019931052631578947,
+      "loss": 0.6001,
+      "step": 74820
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019923157894736842,
+      "loss": 0.5994,
+      "step": 74830
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019915263157894736,
+      "loss": 0.5961,
+      "step": 74840
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019907368421052629,
+      "loss": 0.6031,
+      "step": 74850
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019899473684210523,
+      "loss": 0.5944,
+      "step": 74860
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019891578947368418,
+      "loss": 0.5957,
+      "step": 74870
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019883684210526313,
+      "loss": 0.5954,
+      "step": 74880
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019875789473684208,
+      "loss": 0.5925,
+      "step": 74890
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019867894736842105,
+      "loss": 0.5936,
+      "step": 74900
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001986,
+      "loss": 0.6004,
+      "step": 74910
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019852105263157892,
+      "loss": 0.5987,
+      "step": 74920
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019844210526315787,
+      "loss": 0.5971,
+      "step": 74930
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019836315789473682,
+      "loss": 0.5983,
+      "step": 74940
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019828421052631577,
+      "loss": 0.5889,
+      "step": 74950
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019820526315789472,
+      "loss": 0.5969,
+      "step": 74960
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019812631578947367,
+      "loss": 0.6017,
+      "step": 74970
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001980473684210526,
+      "loss": 0.5875,
+      "step": 74980
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001979684210526316,
+      "loss": 0.5982,
+      "step": 74990
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001978894736842105,
+      "loss": 0.5837,
+      "step": 75000
+    },
+    {
+      "epoch": 0.75,
+      "eval_accuracy": 0.8770046977807257,
+      "eval_loss": 0.56396484375,
+      "eval_runtime": 98.5907,
+      "eval_samples_per_second": 811.436,
+      "eval_steps_per_second": 1.592,
+      "step": 75000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019781052631578946,
+      "loss": 0.5881,
+      "step": 75010
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001977315789473684,
+      "loss": 0.5946,
+      "step": 75020
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019765263157894736,
+      "loss": 0.588,
+      "step": 75030
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001975736842105263,
+      "loss": 0.5886,
+      "step": 75040
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019749473684210525,
+      "loss": 0.5792,
+      "step": 75050
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019741578947368417,
+      "loss": 0.5923,
+      "step": 75060
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019733684210526312,
+      "loss": 0.6011,
+      "step": 75070
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001972578947368421,
+      "loss": 0.5931,
+      "step": 75080
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019717894736842105,
+      "loss": 0.5948,
+      "step": 75090
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001971,
+      "loss": 0.5909,
+      "step": 75100
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019702105263157894,
+      "loss": 0.5982,
+      "step": 75110
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001969421052631579,
+      "loss": 0.5944,
+      "step": 75120
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001968631578947368,
+      "loss": 0.5918,
+      "step": 75130
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019678421052631576,
+      "loss": 0.5985,
+      "step": 75140
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001967052631578947,
+      "loss": 0.5968,
+      "step": 75150
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019662631578947366,
+      "loss": 0.5939,
+      "step": 75160
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001965473684210526,
+      "loss": 0.5976,
+      "step": 75170
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019646842105263158,
+      "loss": 0.5988,
+      "step": 75180
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019638947368421053,
+      "loss": 0.6072,
+      "step": 75190
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019631052631578945,
+      "loss": 0.5909,
+      "step": 75200
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001962315789473684,
+      "loss": 0.5811,
+      "step": 75210
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019615263157894735,
+      "loss": 0.5932,
+      "step": 75220
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001960736842105263,
+      "loss": 0.5838,
+      "step": 75230
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019599473684210524,
+      "loss": 0.5789,
+      "step": 75240
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001959157894736842,
+      "loss": 0.5852,
+      "step": 75250
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001958368421052631,
+      "loss": 0.5922,
+      "step": 75260
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019575789473684211,
+      "loss": 0.5971,
+      "step": 75270
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019568684210526314,
+      "loss": 0.59,
+      "step": 75280
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019560789473684208,
+      "loss": 0.5934,
+      "step": 75290
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019552894736842103,
+      "loss": 0.5961,
+      "step": 75300
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019544999999999998,
+      "loss": 0.5928,
+      "step": 75310
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019537105263157893,
+      "loss": 0.5838,
+      "step": 75320
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001952921052631579,
+      "loss": 0.5978,
+      "step": 75330
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019521315789473683,
+      "loss": 0.5994,
+      "step": 75340
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019513421052631577,
+      "loss": 0.5816,
+      "step": 75350
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019505526315789472,
+      "loss": 0.6051,
+      "step": 75360
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019497631578947367,
+      "loss": 0.5852,
+      "step": 75370
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019489736842105262,
+      "loss": 0.5833,
+      "step": 75380
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019481842105263157,
+      "loss": 0.5901,
+      "step": 75390
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001947394736842105,
+      "loss": 0.5961,
+      "step": 75400
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019466052631578944,
+      "loss": 0.5979,
+      "step": 75410
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019458157894736838,
+      "loss": 0.5892,
+      "step": 75420
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019450263157894736,
+      "loss": 0.6031,
+      "step": 75430
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001944236842105263,
+      "loss": 0.592,
+      "step": 75440
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019434473684210526,
+      "loss": 0.6006,
+      "step": 75450
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001942657894736842,
+      "loss": 0.5834,
+      "step": 75460
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019418684210526315,
+      "loss": 0.5959,
+      "step": 75470
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019410789473684207,
+      "loss": 0.5937,
+      "step": 75480
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019402894736842102,
+      "loss": 0.5794,
+      "step": 75490
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019394999999999997,
+      "loss": 0.589,
+      "step": 75500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019387105263157892,
+      "loss": 0.5825,
+      "step": 75510
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001937921052631579,
+      "loss": 0.5892,
+      "step": 75520
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019371315789473684,
+      "loss": 0.5842,
+      "step": 75530
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001936342105263158,
+      "loss": 0.5992,
+      "step": 75540
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001935552631578947,
+      "loss": 0.5851,
+      "step": 75550
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019347631578947366,
+      "loss": 0.5842,
+      "step": 75560
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001933973684210526,
+      "loss": 0.5936,
+      "step": 75570
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019331842105263156,
+      "loss": 0.58,
+      "step": 75580
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001932394736842105,
+      "loss": 0.5877,
+      "step": 75590
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019316052631578945,
+      "loss": 0.5899,
+      "step": 75600
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019308157894736843,
+      "loss": 0.5859,
+      "step": 75610
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019300263157894738,
+      "loss": 0.5906,
+      "step": 75620
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001929236842105263,
+      "loss": 0.6013,
+      "step": 75630
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019284473684210525,
+      "loss": 0.5922,
+      "step": 75640
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001927657894736842,
+      "loss": 0.593,
+      "step": 75650
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019268684210526314,
+      "loss": 0.5846,
+      "step": 75660
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001926078947368421,
+      "loss": 0.6047,
+      "step": 75670
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019252894736842104,
+      "loss": 0.5999,
+      "step": 75680
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019244999999999996,
+      "loss": 0.5929,
+      "step": 75690
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001923710526315789,
+      "loss": 0.5971,
+      "step": 75700
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019229210526315789,
+      "loss": 0.5924,
+      "step": 75710
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019221315789473683,
+      "loss": 0.5932,
+      "step": 75720
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019213421052631578,
+      "loss": 0.5953,
+      "step": 75730
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019205526315789473,
+      "loss": 0.5933,
+      "step": 75740
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019197631578947368,
+      "loss": 0.5838,
+      "step": 75750
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001918973684210526,
+      "loss": 0.5949,
+      "step": 75760
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019181842105263155,
+      "loss": 0.583,
+      "step": 75770
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001917394736842105,
+      "loss": 0.5799,
+      "step": 75780
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019166052631578944,
+      "loss": 0.5821,
+      "step": 75790
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019158157894736842,
+      "loss": 0.582,
+      "step": 75800
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019150263157894737,
+      "loss": 0.5926,
+      "step": 75810
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019142368421052632,
+      "loss": 0.5814,
+      "step": 75820
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019134473684210524,
+      "loss": 0.5775,
+      "step": 75830
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019126578947368419,
+      "loss": 0.589,
+      "step": 75840
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019118684210526313,
+      "loss": 0.5776,
+      "step": 75850
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019110789473684208,
+      "loss": 0.5853,
+      "step": 75860
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019102894736842103,
+      "loss": 0.5883,
+      "step": 75870
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019094999999999998,
+      "loss": 0.5909,
+      "step": 75880
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019087105263157895,
+      "loss": 0.592,
+      "step": 75890
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001907921052631579,
+      "loss": 0.5967,
+      "step": 75900
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019071315789473682,
+      "loss": 0.5933,
+      "step": 75910
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019063421052631577,
+      "loss": 0.5856,
+      "step": 75920
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019055526315789472,
+      "loss": 0.5865,
+      "step": 75930
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019047631578947367,
+      "loss": 0.5849,
+      "step": 75940
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019039736842105262,
+      "loss": 0.5971,
+      "step": 75950
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019031842105263157,
+      "loss": 0.5906,
+      "step": 75960
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001902394736842105,
+      "loss": 0.5767,
+      "step": 75970
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019016052631578944,
+      "loss": 0.5784,
+      "step": 75980
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001900815789473684,
+      "loss": 0.5832,
+      "step": 75990
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019000263157894736,
+      "loss": 0.5943,
+      "step": 76000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001899236842105263,
+      "loss": 0.5917,
+      "step": 76010
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018984473684210526,
+      "loss": 0.596,
+      "step": 76020
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001897657894736842,
+      "loss": 0.588,
+      "step": 76030
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018968684210526313,
+      "loss": 0.6016,
+      "step": 76040
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018960789473684207,
+      "loss": 0.5879,
+      "step": 76050
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018952894736842102,
+      "loss": 0.5827,
+      "step": 76060
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018944999999999997,
+      "loss": 0.5805,
+      "step": 76070
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018937105263157895,
+      "loss": 0.5955,
+      "step": 76080
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001892921052631579,
+      "loss": 0.5977,
+      "step": 76090
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018921315789473684,
+      "loss": 0.5988,
+      "step": 76100
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001891342105263158,
+      "loss": 0.5972,
+      "step": 76110
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001890552631578947,
+      "loss": 0.5906,
+      "step": 76120
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018897631578947366,
+      "loss": 0.5982,
+      "step": 76130
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001888973684210526,
+      "loss": 0.5872,
+      "step": 76140
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018881842105263156,
+      "loss": 0.5906,
+      "step": 76150
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001887394736842105,
+      "loss": 0.6073,
+      "step": 76160
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018866052631578948,
+      "loss": 0.6051,
+      "step": 76170
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018858157894736843,
+      "loss": 0.5927,
+      "step": 76180
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018850263157894735,
+      "loss": 0.5896,
+      "step": 76190
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001884236842105263,
+      "loss": 0.586,
+      "step": 76200
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018834473684210525,
+      "loss": 0.5846,
+      "step": 76210
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001882657894736842,
+      "loss": 0.5976,
+      "step": 76220
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018818684210526314,
+      "loss": 0.5955,
+      "step": 76230
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001881078947368421,
+      "loss": 0.5932,
+      "step": 76240
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.000188028947368421,
+      "loss": 0.593,
+      "step": 76250
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018794999999999996,
+      "loss": 0.5816,
+      "step": 76260
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018787105263157894,
+      "loss": 0.5801,
+      "step": 76270
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018779999999999998,
+      "loss": 0.5974,
+      "step": 76280
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018772105263157893,
+      "loss": 0.5982,
+      "step": 76290
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018764210526315788,
+      "loss": 0.5994,
+      "step": 76300
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001875631578947368,
+      "loss": 0.5968,
+      "step": 76310
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018748421052631575,
+      "loss": 0.6025,
+      "step": 76320
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018740526315789473,
+      "loss": 0.5866,
+      "step": 76330
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018732631578947367,
+      "loss": 0.5938,
+      "step": 76340
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018724736842105262,
+      "loss": 0.5912,
+      "step": 76350
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018716842105263157,
+      "loss": 0.5797,
+      "step": 76360
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018708947368421052,
+      "loss": 0.6013,
+      "step": 76370
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018701052631578947,
+      "loss": 0.5921,
+      "step": 76380
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001869315789473684,
+      "loss": 0.5994,
+      "step": 76390
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018685263157894734,
+      "loss": 0.5708,
+      "step": 76400
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018677368421052628,
+      "loss": 0.5767,
+      "step": 76410
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018669473684210523,
+      "loss": 0.5678,
+      "step": 76420
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001866157894736842,
+      "loss": 0.5777,
+      "step": 76430
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018653684210526316,
+      "loss": 0.5798,
+      "step": 76440
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001864578947368421,
+      "loss": 0.5749,
+      "step": 76450
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018637894736842103,
+      "loss": 0.5691,
+      "step": 76460
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018629999999999997,
+      "loss": 0.5552,
+      "step": 76470
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018622105263157892,
+      "loss": 0.5727,
+      "step": 76480
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018614210526315787,
+      "loss": 0.5828,
+      "step": 76490
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018606315789473682,
+      "loss": 0.5758,
+      "step": 76500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018598421052631577,
+      "loss": 0.5768,
+      "step": 76510
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018590526315789474,
+      "loss": 0.5644,
+      "step": 76520
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001858263157894737,
+      "loss": 0.5651,
+      "step": 76530
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001857473684210526,
+      "loss": 0.5793,
+      "step": 76540
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018566842105263156,
+      "loss": 0.5897,
+      "step": 76550
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001855894736842105,
+      "loss": 0.5769,
+      "step": 76560
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018551052631578946,
+      "loss": 0.5866,
+      "step": 76570
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001854315789473684,
+      "loss": 0.5854,
+      "step": 76580
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018535263157894735,
+      "loss": 0.5828,
+      "step": 76590
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018527368421052628,
+      "loss": 0.581,
+      "step": 76600
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018519473684210525,
+      "loss": 0.57,
+      "step": 76610
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001851157894736842,
+      "loss": 0.5655,
+      "step": 76620
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018503684210526315,
+      "loss": 0.5806,
+      "step": 76630
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001849578947368421,
+      "loss": 0.5757,
+      "step": 76640
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018488684210526314,
+      "loss": 0.5872,
+      "step": 76650
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018480789473684207,
+      "loss": 0.5703,
+      "step": 76660
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.000184728947368421,
+      "loss": 0.5723,
+      "step": 76670
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018465,
+      "loss": 0.5638,
+      "step": 76680
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018457105263157894,
+      "loss": 0.5819,
+      "step": 76690
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018449210526315789,
+      "loss": 0.5739,
+      "step": 76700
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018441315789473683,
+      "loss": 0.5862,
+      "step": 76710
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018433421052631578,
+      "loss": 0.5625,
+      "step": 76720
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018425526315789473,
+      "loss": 0.5555,
+      "step": 76730
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018417631578947365,
+      "loss": 0.5683,
+      "step": 76740
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001840973684210526,
+      "loss": 0.5669,
+      "step": 76750
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018401842105263155,
+      "loss": 0.5604,
+      "step": 76760
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018393947368421052,
+      "loss": 0.5773,
+      "step": 76770
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018386052631578947,
+      "loss": 0.5661,
+      "step": 76780
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018378157894736842,
+      "loss": 0.5645,
+      "step": 76790
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018370263157894737,
+      "loss": 0.5782,
+      "step": 76800
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001836236842105263,
+      "loss": 0.5892,
+      "step": 76810
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018354473684210524,
+      "loss": 0.5789,
+      "step": 76820
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018346578947368419,
+      "loss": 0.5755,
+      "step": 76830
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018338684210526313,
+      "loss": 0.5721,
+      "step": 76840
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018330789473684208,
+      "loss": 0.5785,
+      "step": 76850
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018322894736842106,
+      "loss": 0.5665,
+      "step": 76860
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018315,
+      "loss": 0.567,
+      "step": 76870
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018307105263157893,
+      "loss": 0.5833,
+      "step": 76880
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018299210526315788,
+      "loss": 0.596,
+      "step": 76890
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018291315789473682,
+      "loss": 0.5872,
+      "step": 76900
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018283421052631577,
+      "loss": 0.5898,
+      "step": 76910
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018275526315789472,
+      "loss": 0.5931,
+      "step": 76920
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018267631578947367,
+      "loss": 0.592,
+      "step": 76930
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001825973684210526,
+      "loss": 0.5917,
+      "step": 76940
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018251842105263154,
+      "loss": 0.5895,
+      "step": 76950
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018243947368421051,
+      "loss": 0.5849,
+      "step": 76960
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018236052631578946,
+      "loss": 0.5931,
+      "step": 76970
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001822815789473684,
+      "loss": 0.5846,
+      "step": 76980
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018220263157894736,
+      "loss": 0.5949,
+      "step": 76990
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001821236842105263,
+      "loss": 0.5936,
+      "step": 77000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018204473684210526,
+      "loss": 0.5891,
+      "step": 77010
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018196578947368418,
+      "loss": 0.5882,
+      "step": 77020
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018188684210526313,
+      "loss": 0.5777,
+      "step": 77030
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018180789473684207,
+      "loss": 0.5873,
+      "step": 77040
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018172894736842105,
+      "loss": 0.5893,
+      "step": 77050
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018165,
+      "loss": 0.5855,
+      "step": 77060
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018157105263157895,
+      "loss": 0.5977,
+      "step": 77070
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001814921052631579,
+      "loss": 0.6028,
+      "step": 77080
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018141315789473682,
+      "loss": 0.5925,
+      "step": 77090
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018133421052631576,
+      "loss": 0.5958,
+      "step": 77100
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001812552631578947,
+      "loss": 0.5835,
+      "step": 77110
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018117631578947366,
+      "loss": 0.5854,
+      "step": 77120
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001810973684210526,
+      "loss": 0.5894,
+      "step": 77130
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018101842105263158,
+      "loss": 0.5928,
+      "step": 77140
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018093947368421053,
+      "loss": 0.5896,
+      "step": 77150
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018086052631578948,
+      "loss": 0.5805,
+      "step": 77160
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001807815789473684,
+      "loss": 0.5839,
+      "step": 77170
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018070263157894735,
+      "loss": 0.5863,
+      "step": 77180
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001806236842105263,
+      "loss": 0.5859,
+      "step": 77190
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018054473684210525,
+      "loss": 0.5914,
+      "step": 77200
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001804657894736842,
+      "loss": 0.5905,
+      "step": 77210
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018038684210526314,
+      "loss": 0.5948,
+      "step": 77220
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018030789473684206,
+      "loss": 0.5882,
+      "step": 77230
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018022894736842104,
+      "loss": 0.59,
+      "step": 77240
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018015,
+      "loss": 0.5804,
+      "step": 77250
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018007105263157894,
+      "loss": 0.5853,
+      "step": 77260
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00017999210526315788,
+      "loss": 0.5858,
+      "step": 77270
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00017991315789473683,
+      "loss": 0.6019,
+      "step": 77280
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00017983421052631578,
+      "loss": 0.599,
+      "step": 77290
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001797552631578947,
+      "loss": 0.5872,
+      "step": 77300
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00017967631578947365,
+      "loss": 0.5993,
+      "step": 77310
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001795973684210526,
+      "loss": 0.6082,
+      "step": 77320
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00017951842105263157,
+      "loss": 0.5967,
+      "step": 77330
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00017943947368421052,
+      "loss": 0.5995,
+      "step": 77340
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00017936052631578947,
+      "loss": 0.5943,
+      "step": 77350
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00017928157894736842,
+      "loss": 0.5915,
+      "step": 77360
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00017920263157894734,
+      "loss": 0.5712,
+      "step": 77370
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001791236842105263,
+      "loss": 0.5805,
+      "step": 77380
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00017904473684210524,
+      "loss": 0.5757,
+      "step": 77390
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00017896578947368419,
+      "loss": 0.5926,
+      "step": 77400
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00017888684210526313,
+      "loss": 0.5889,
+      "step": 77410
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001788078947368421,
+      "loss": 0.5812,
+      "step": 77420
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00017872894736842106,
+      "loss": 0.5989,
+      "step": 77430
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00017865,
+      "loss": 0.5755,
+      "step": 77440
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00017857105263157893,
+      "loss": 0.5941,
+      "step": 77450
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00017849210526315788,
+      "loss": 0.5814,
+      "step": 77460
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00017841315789473682,
+      "loss": 0.5839,
+      "step": 77470
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00017833421052631577,
+      "loss": 0.5837,
+      "step": 77480
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00017825526315789472,
+      "loss": 0.5764,
+      "step": 77490
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017817631578947367,
+      "loss": 0.5866,
+      "step": 77500
+    },
+    {
+      "epoch": 0.78,
+      "eval_accuracy": 0.8773601876392698,
+      "eval_loss": 0.5615234375,
+      "eval_runtime": 98.3407,
+      "eval_samples_per_second": 813.499,
+      "eval_steps_per_second": 1.596,
+      "step": 77500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001780973684210526,
+      "loss": 0.5687,
+      "step": 77510
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017801842105263156,
+      "loss": 0.5813,
+      "step": 77520
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001779394736842105,
+      "loss": 0.5739,
+      "step": 77530
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017786052631578946,
+      "loss": 0.5825,
+      "step": 77540
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001777815789473684,
+      "loss": 0.5699,
+      "step": 77550
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017770263157894736,
+      "loss": 0.5796,
+      "step": 77560
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001776236842105263,
+      "loss": 0.571,
+      "step": 77570
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017754473684210523,
+      "loss": 0.5849,
+      "step": 77580
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017746578947368418,
+      "loss": 0.5826,
+      "step": 77590
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017738684210526312,
+      "loss": 0.6047,
+      "step": 77600
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001773078947368421,
+      "loss": 0.5951,
+      "step": 77610
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017722894736842105,
+      "loss": 0.5876,
+      "step": 77620
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017715,
+      "loss": 0.5915,
+      "step": 77630
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017707105263157894,
+      "loss": 0.5894,
+      "step": 77640
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001769921052631579,
+      "loss": 0.5891,
+      "step": 77650
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017691315789473681,
+      "loss": 0.5863,
+      "step": 77660
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017683421052631576,
+      "loss": 0.5891,
+      "step": 77670
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001767552631578947,
+      "loss": 0.5976,
+      "step": 77680
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017667631578947366,
+      "loss": 0.5871,
+      "step": 77690
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017659736842105263,
+      "loss": 0.596,
+      "step": 77700
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017651842105263158,
+      "loss": 0.592,
+      "step": 77710
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017643947368421053,
+      "loss": 0.6012,
+      "step": 77720
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017636052631578945,
+      "loss": 0.5855,
+      "step": 77730
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001762815789473684,
+      "loss": 0.5898,
+      "step": 77740
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017620263157894735,
+      "loss": 0.6017,
+      "step": 77750
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001761236842105263,
+      "loss": 0.5926,
+      "step": 77760
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017604473684210525,
+      "loss": 0.591,
+      "step": 77770
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001759657894736842,
+      "loss": 0.596,
+      "step": 77780
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017588684210526311,
+      "loss": 0.5855,
+      "step": 77790
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001758078947368421,
+      "loss": 0.5889,
+      "step": 77800
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017572894736842104,
+      "loss": 0.5921,
+      "step": 77810
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017565,
+      "loss": 0.5874,
+      "step": 77820
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017557105263157894,
+      "loss": 0.5985,
+      "step": 77830
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017549210526315788,
+      "loss": 0.5736,
+      "step": 77840
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017541315789473683,
+      "loss": 0.5707,
+      "step": 77850
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017533421052631575,
+      "loss": 0.5861,
+      "step": 77860
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001752552631578947,
+      "loss": 0.5843,
+      "step": 77870
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017517631578947365,
+      "loss": 0.5825,
+      "step": 77880
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017509736842105263,
+      "loss": 0.5816,
+      "step": 77890
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017501842105263157,
+      "loss": 0.5691,
+      "step": 77900
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017493947368421052,
+      "loss": 0.5769,
+      "step": 77910
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017486052631578947,
+      "loss": 0.5875,
+      "step": 77920
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017478157894736842,
+      "loss": 0.5881,
+      "step": 77930
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017470263157894734,
+      "loss": 0.5793,
+      "step": 77940
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001746236842105263,
+      "loss": 0.5803,
+      "step": 77950
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017454473684210524,
+      "loss": 0.5789,
+      "step": 77960
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017446578947368418,
+      "loss": 0.5912,
+      "step": 77970
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017438684210526316,
+      "loss": 0.5939,
+      "step": 77980
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001743078947368421,
+      "loss": 0.5964,
+      "step": 77990
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017422894736842106,
+      "loss": 0.5815,
+      "step": 78000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017414999999999998,
+      "loss": 0.5832,
+      "step": 78010
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017407105263157893,
+      "loss": 0.5843,
+      "step": 78020
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017399210526315787,
+      "loss": 0.5763,
+      "step": 78030
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017391315789473682,
+      "loss": 0.5813,
+      "step": 78040
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017383421052631577,
+      "loss": 0.5774,
+      "step": 78050
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017375526315789472,
+      "loss": 0.5784,
+      "step": 78060
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001736763157894737,
+      "loss": 0.5669,
+      "step": 78070
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017359736842105264,
+      "loss": 0.5702,
+      "step": 78080
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017351842105263156,
+      "loss": 0.5785,
+      "step": 78090
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001734394736842105,
+      "loss": 0.5729,
+      "step": 78100
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017336052631578946,
+      "loss": 0.5699,
+      "step": 78110
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001732815789473684,
+      "loss": 0.578,
+      "step": 78120
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017320263157894736,
+      "loss": 0.5723,
+      "step": 78130
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001731236842105263,
+      "loss": 0.5908,
+      "step": 78140
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017304473684210523,
+      "loss": 0.5829,
+      "step": 78150
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017296578947368417,
+      "loss": 0.59,
+      "step": 78160
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017288684210526315,
+      "loss": 0.5888,
+      "step": 78170
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001728078947368421,
+      "loss": 0.5903,
+      "step": 78180
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017272894736842105,
+      "loss": 0.5908,
+      "step": 78190
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017265,
+      "loss": 0.5988,
+      "step": 78200
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017257105263157894,
+      "loss": 0.5943,
+      "step": 78210
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017249210526315786,
+      "loss": 0.5881,
+      "step": 78220
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001724131578947368,
+      "loss": 0.5965,
+      "step": 78230
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017233421052631576,
+      "loss": 0.598,
+      "step": 78240
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001722552631578947,
+      "loss": 0.5954,
+      "step": 78250
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017217631578947369,
+      "loss": 0.6063,
+      "step": 78260
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017209736842105263,
+      "loss": 0.607,
+      "step": 78270
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017201842105263158,
+      "loss": 0.5848,
+      "step": 78280
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001719394736842105,
+      "loss": 0.5862,
+      "step": 78290
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017186052631578945,
+      "loss": 0.5925,
+      "step": 78300
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001717815789473684,
+      "loss": 0.5859,
+      "step": 78310
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017170263157894735,
+      "loss": 0.5928,
+      "step": 78320
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001716236842105263,
+      "loss": 0.5863,
+      "step": 78330
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017154473684210524,
+      "loss": 0.5849,
+      "step": 78340
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017146578947368422,
+      "loss": 0.5821,
+      "step": 78350
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017138684210526317,
+      "loss": 0.5849,
+      "step": 78360
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001713078947368421,
+      "loss": 0.5799,
+      "step": 78370
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017122894736842104,
+      "loss": 0.5936,
+      "step": 78380
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017114999999999999,
+      "loss": 0.5874,
+      "step": 78390
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017107105263157893,
+      "loss": 0.5931,
+      "step": 78400
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017099210526315788,
+      "loss": 0.5942,
+      "step": 78410
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017091315789473683,
+      "loss": 0.5835,
+      "step": 78420
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017083421052631575,
+      "loss": 0.585,
+      "step": 78430
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001707552631578947,
+      "loss": 0.5985,
+      "step": 78440
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017067631578947368,
+      "loss": 0.5995,
+      "step": 78450
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017059736842105262,
+      "loss": 0.5992,
+      "step": 78460
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017051842105263157,
+      "loss": 0.5808,
+      "step": 78470
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017043947368421052,
+      "loss": 0.5961,
+      "step": 78480
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017036052631578947,
+      "loss": 0.5751,
+      "step": 78490
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001702815789473684,
+      "loss": 0.5802,
+      "step": 78500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00017020263157894734,
+      "loss": 0.5793,
+      "step": 78510
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001701236842105263,
+      "loss": 0.5904,
+      "step": 78520
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00017004473684210524,
+      "loss": 0.5702,
+      "step": 78530
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001699657894736842,
+      "loss": 0.5715,
+      "step": 78540
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016988684210526316,
+      "loss": 0.5723,
+      "step": 78550
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001698078947368421,
+      "loss": 0.5778,
+      "step": 78560
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016973684210526313,
+      "loss": 0.5969,
+      "step": 78570
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016965789473684208,
+      "loss": 0.6002,
+      "step": 78580
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016957894736842102,
+      "loss": 0.5946,
+      "step": 78590
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016949999999999997,
+      "loss": 0.5937,
+      "step": 78600
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016942105263157895,
+      "loss": 0.5941,
+      "step": 78610
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001693421052631579,
+      "loss": 0.611,
+      "step": 78620
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016926315789473684,
+      "loss": 0.5932,
+      "step": 78630
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016918421052631577,
+      "loss": 0.6037,
+      "step": 78640
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016910526315789471,
+      "loss": 0.596,
+      "step": 78650
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016902631578947366,
+      "loss": 0.5977,
+      "step": 78660
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001689473684210526,
+      "loss": 0.6065,
+      "step": 78670
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016886842105263156,
+      "loss": 0.6014,
+      "step": 78680
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001687894736842105,
+      "loss": 0.5962,
+      "step": 78690
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016871052631578948,
+      "loss": 0.5967,
+      "step": 78700
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001686315789473684,
+      "loss": 0.5962,
+      "step": 78710
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016855263157894735,
+      "loss": 0.5877,
+      "step": 78720
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001684736842105263,
+      "loss": 0.5971,
+      "step": 78730
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016839473684210525,
+      "loss": 0.5903,
+      "step": 78740
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001683157894736842,
+      "loss": 0.5888,
+      "step": 78750
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016823684210526315,
+      "loss": 0.5827,
+      "step": 78760
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016815789473684207,
+      "loss": 0.581,
+      "step": 78770
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016807894736842102,
+      "loss": 0.574,
+      "step": 78780
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.000168,
+      "loss": 0.5862,
+      "step": 78790
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016792105263157894,
+      "loss": 0.5846,
+      "step": 78800
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001678421052631579,
+      "loss": 0.595,
+      "step": 78810
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016776315789473684,
+      "loss": 0.5791,
+      "step": 78820
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016768421052631578,
+      "loss": 0.5823,
+      "step": 78830
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016760526315789473,
+      "loss": 0.5918,
+      "step": 78840
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016752631578947365,
+      "loss": 0.5902,
+      "step": 78850
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001674473684210526,
+      "loss": 0.5868,
+      "step": 78860
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016736842105263155,
+      "loss": 0.5845,
+      "step": 78870
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001672894736842105,
+      "loss": 0.5821,
+      "step": 78880
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016721052631578947,
+      "loss": 0.5929,
+      "step": 78890
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016713157894736842,
+      "loss": 0.5958,
+      "step": 78900
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016705263157894737,
+      "loss": 0.5871,
+      "step": 78910
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001669736842105263,
+      "loss": 0.5838,
+      "step": 78920
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016689473684210524,
+      "loss": 0.5899,
+      "step": 78930
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001668157894736842,
+      "loss": 0.5921,
+      "step": 78940
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016673684210526314,
+      "loss": 0.5951,
+      "step": 78950
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016665789473684208,
+      "loss": 0.5961,
+      "step": 78960
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016657894736842103,
+      "loss": 0.5973,
+      "step": 78970
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001665,
+      "loss": 0.5763,
+      "step": 78980
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016642105263157896,
+      "loss": 0.5862,
+      "step": 78990
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016634210526315788,
+      "loss": 0.592,
+      "step": 79000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016626315789473683,
+      "loss": 0.5803,
+      "step": 79010
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016618421052631577,
+      "loss": 0.5817,
+      "step": 79020
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016610526315789472,
+      "loss": 0.5859,
+      "step": 79030
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016602631578947367,
+      "loss": 0.5815,
+      "step": 79040
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016594736842105262,
+      "loss": 0.5921,
+      "step": 79050
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016586842105263154,
+      "loss": 0.5914,
+      "step": 79060
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016578947368421052,
+      "loss": 0.58,
+      "step": 79070
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016571052631578946,
+      "loss": 0.5861,
+      "step": 79080
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001656315789473684,
+      "loss": 0.5866,
+      "step": 79090
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016555263157894736,
+      "loss": 0.5887,
+      "step": 79100
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001654736842105263,
+      "loss": 0.5906,
+      "step": 79110
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016539473684210526,
+      "loss": 0.5837,
+      "step": 79120
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016531578947368418,
+      "loss": 0.585,
+      "step": 79130
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016523684210526313,
+      "loss": 0.5758,
+      "step": 79140
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016515789473684208,
+      "loss": 0.5805,
+      "step": 79150
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016507894736842102,
+      "loss": 0.5879,
+      "step": 79160
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.000165,
+      "loss": 0.5835,
+      "step": 79170
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016492105263157895,
+      "loss": 0.5816,
+      "step": 79180
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001648421052631579,
+      "loss": 0.5812,
+      "step": 79190
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016476315789473682,
+      "loss": 0.5868,
+      "step": 79200
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016468421052631577,
+      "loss": 0.5846,
+      "step": 79210
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001646052631578947,
+      "loss": 0.6016,
+      "step": 79220
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016452631578947366,
+      "loss": 0.5815,
+      "step": 79230
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001644473684210526,
+      "loss": 0.5742,
+      "step": 79240
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016436842105263156,
+      "loss": 0.5834,
+      "step": 79250
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016428947368421053,
+      "loss": 0.5736,
+      "step": 79260
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016421052631578948,
+      "loss": 0.5911,
+      "step": 79270
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001641315789473684,
+      "loss": 0.5774,
+      "step": 79280
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016405263157894735,
+      "loss": 0.5825,
+      "step": 79290
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001639736842105263,
+      "loss": 0.5693,
+      "step": 79300
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016389473684210525,
+      "loss": 0.561,
+      "step": 79310
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001638157894736842,
+      "loss": 0.573,
+      "step": 79320
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016373684210526314,
+      "loss": 0.5787,
+      "step": 79330
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016365789473684207,
+      "loss": 0.5856,
+      "step": 79340
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016357894736842104,
+      "loss": 0.5764,
+      "step": 79350
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001635,
+      "loss": 0.569,
+      "step": 79360
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016342105263157894,
+      "loss": 0.5721,
+      "step": 79370
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016334210526315789,
+      "loss": 0.5951,
+      "step": 79380
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016326315789473683,
+      "loss": 0.5956,
+      "step": 79390
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016318421052631578,
+      "loss": 0.5996,
+      "step": 79400
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001631052631578947,
+      "loss": 0.5883,
+      "step": 79410
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016302631578947365,
+      "loss": 0.5969,
+      "step": 79420
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001629473684210526,
+      "loss": 0.5874,
+      "step": 79430
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016286842105263155,
+      "loss": 0.589,
+      "step": 79440
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016278947368421052,
+      "loss": 0.588,
+      "step": 79450
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016271052631578947,
+      "loss": 0.5926,
+      "step": 79460
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016263157894736842,
+      "loss": 0.5965,
+      "step": 79470
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016255263157894737,
+      "loss": 0.5897,
+      "step": 79480
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001624736842105263,
+      "loss": 0.5836,
+      "step": 79490
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016239473684210524,
+      "loss": 0.58,
+      "step": 79500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001623157894736842,
+      "loss": 0.581,
+      "step": 79510
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016223684210526314,
+      "loss": 0.5848,
+      "step": 79520
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016215789473684208,
+      "loss": 0.5796,
+      "step": 79530
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016207894736842106,
+      "loss": 0.5901,
+      "step": 79540
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.000162,
+      "loss": 0.5849,
+      "step": 79550
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016192105263157893,
+      "loss": 0.5809,
+      "step": 79560
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016184210526315788,
+      "loss": 0.5734,
+      "step": 79570
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016176315789473683,
+      "loss": 0.5786,
+      "step": 79580
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016168421052631577,
+      "loss": 0.5781,
+      "step": 79590
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016160526315789472,
+      "loss": 0.5922,
+      "step": 79600
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016152631578947367,
+      "loss": 0.5848,
+      "step": 79610
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001614473684210526,
+      "loss": 0.5804,
+      "step": 79620
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016136842105263157,
+      "loss": 0.5825,
+      "step": 79630
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016128947368421052,
+      "loss": 0.5732,
+      "step": 79640
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016121052631578946,
+      "loss": 0.5813,
+      "step": 79650
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001611315789473684,
+      "loss": 0.5909,
+      "step": 79660
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016105263157894736,
+      "loss": 0.5927,
+      "step": 79670
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001609736842105263,
+      "loss": 0.5889,
+      "step": 79680
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016089473684210523,
+      "loss": 0.5828,
+      "step": 79690
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016081578947368418,
+      "loss": 0.595,
+      "step": 79700
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016073684210526313,
+      "loss": 0.5848,
+      "step": 79710
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001606578947368421,
+      "loss": 0.5866,
+      "step": 79720
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016057894736842105,
+      "loss": 0.5735,
+      "step": 79730
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001605,
+      "loss": 0.5765,
+      "step": 79740
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016042105263157895,
+      "loss": 0.5838,
+      "step": 79750
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001603421052631579,
+      "loss": 0.5802,
+      "step": 79760
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016026315789473682,
+      "loss": 0.5899,
+      "step": 79770
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016018421052631576,
+      "loss": 0.5889,
+      "step": 79780
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001601052631578947,
+      "loss": 0.5812,
+      "step": 79790
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016002631578947366,
+      "loss": 0.5699,
+      "step": 79800
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001599473684210526,
+      "loss": 0.5774,
+      "step": 79810
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00015986842105263158,
+      "loss": 0.5809,
+      "step": 79820
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00015978947368421053,
+      "loss": 0.5825,
+      "step": 79830
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00015971052631578945,
+      "loss": 0.5816,
+      "step": 79840
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001596315789473684,
+      "loss": 0.5878,
+      "step": 79850
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00015955263157894735,
+      "loss": 0.586,
+      "step": 79860
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001594736842105263,
+      "loss": 0.5965,
+      "step": 79870
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00015939473684210525,
+      "loss": 0.5867,
+      "step": 79880
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001593157894736842,
+      "loss": 0.5934,
+      "step": 79890
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00015923684210526312,
+      "loss": 0.5853,
+      "step": 79900
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00015915789473684212,
+      "loss": 0.5937,
+      "step": 79910
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00015907894736842104,
+      "loss": 0.5861,
+      "step": 79920
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.000159,
+      "loss": 0.5847,
+      "step": 79930
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00015892105263157894,
+      "loss": 0.5819,
+      "step": 79940
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00015884210526315789,
+      "loss": 0.5847,
+      "step": 79950
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00015876315789473683,
+      "loss": 0.5844,
+      "step": 79960
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00015868421052631578,
+      "loss": 0.5868,
+      "step": 79970
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001586052631578947,
+      "loss": 0.5931,
+      "step": 79980
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00015852631578947365,
+      "loss": 0.5832,
+      "step": 79990
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00015844736842105263,
+      "loss": 0.5835,
+      "step": 80000
+    },
+    {
+      "epoch": 0.8,
+      "eval_accuracy": 0.8784877003135946,
+      "eval_loss": 0.55615234375,
+      "eval_runtime": 99.7857,
+      "eval_samples_per_second": 801.718,
+      "eval_steps_per_second": 1.573,
+      "step": 80000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00015836842105263158,
+      "loss": 0.5931,
+      "step": 80010
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00015828947368421052,
+      "loss": 0.5987,
+      "step": 80020
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00015821052631578947,
+      "loss": 0.5891,
+      "step": 80030
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00015813157894736842,
+      "loss": 0.5803,
+      "step": 80040
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00015805263157894734,
+      "loss": 0.5794,
+      "step": 80050
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001579736842105263,
+      "loss": 0.5788,
+      "step": 80060
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00015789473684210524,
+      "loss": 0.5733,
+      "step": 80070
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00015781578947368419,
+      "loss": 0.5641,
+      "step": 80080
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00015773684210526313,
+      "loss": 0.5627,
+      "step": 80090
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001576578947368421,
+      "loss": 0.5655,
+      "step": 80100
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00015757894736842106,
+      "loss": 0.5681,
+      "step": 80110
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00015749999999999998,
+      "loss": 0.5754,
+      "step": 80120
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00015742105263157893,
+      "loss": 0.5619,
+      "step": 80130
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00015734210526315788,
+      "loss": 0.5702,
+      "step": 80140
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00015726315789473682,
+      "loss": 0.574,
+      "step": 80150
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00015718421052631577,
+      "loss": 0.5703,
+      "step": 80160
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00015710526315789472,
+      "loss": 0.5702,
+      "step": 80170
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00015702631578947364,
+      "loss": 0.5722,
+      "step": 80180
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00015694736842105264,
+      "loss": 0.5736,
+      "step": 80190
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00015686842105263157,
+      "loss": 0.5659,
+      "step": 80200
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00015678947368421051,
+      "loss": 0.5753,
+      "step": 80210
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00015671052631578946,
+      "loss": 0.5607,
+      "step": 80220
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001566315789473684,
+      "loss": 0.5664,
+      "step": 80230
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00015655263157894736,
+      "loss": 0.5535,
+      "step": 80240
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001564736842105263,
+      "loss": 0.5607,
+      "step": 80250
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00015639473684210523,
+      "loss": 0.5691,
+      "step": 80260
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00015631578947368418,
+      "loss": 0.563,
+      "step": 80270
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00015623684210526315,
+      "loss": 0.56,
+      "step": 80280
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001561578947368421,
+      "loss": 0.5626,
+      "step": 80290
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00015607894736842105,
+      "loss": 0.5638,
+      "step": 80300
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.000156,
+      "loss": 0.5645,
+      "step": 80310
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00015592105263157895,
+      "loss": 0.5762,
+      "step": 80320
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00015584210526315787,
+      "loss": 0.5436,
+      "step": 80330
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00015576315789473681,
+      "loss": 0.5765,
+      "step": 80340
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00015568421052631576,
+      "loss": 0.5858,
+      "step": 80350
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001556052631578947,
+      "loss": 0.5911,
+      "step": 80360
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00015552631578947366,
+      "loss": 0.5866,
+      "step": 80370
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00015544736842105264,
+      "loss": 0.5814,
+      "step": 80380
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00015536842105263158,
+      "loss": 0.583,
+      "step": 80390
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00015528947368421053,
+      "loss": 0.5753,
+      "step": 80400
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00015521052631578945,
+      "loss": 0.595,
+      "step": 80410
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001551315789473684,
+      "loss": 0.5942,
+      "step": 80420
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00015505263157894735,
+      "loss": 0.5923,
+      "step": 80430
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001549736842105263,
+      "loss": 0.5812,
+      "step": 80440
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00015489473684210525,
+      "loss": 0.5761,
+      "step": 80450
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001548157894736842,
+      "loss": 0.5892,
+      "step": 80460
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00015473684210526317,
+      "loss": 0.583,
+      "step": 80470
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001546578947368421,
+      "loss": 0.5818,
+      "step": 80480
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00015457894736842104,
+      "loss": 0.5782,
+      "step": 80490
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001545,
+      "loss": 0.5796,
+      "step": 80500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00015442105263157894,
+      "loss": 0.5846,
+      "step": 80510
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00015434210526315788,
+      "loss": 0.5836,
+      "step": 80520
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00015426315789473683,
+      "loss": 0.5852,
+      "step": 80530
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00015418421052631575,
+      "loss": 0.5989,
+      "step": 80540
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001541052631578947,
+      "loss": 0.5874,
+      "step": 80550
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00015402631578947368,
+      "loss": 0.5817,
+      "step": 80560
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00015394736842105263,
+      "loss": 0.5643,
+      "step": 80570
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00015387631578947367,
+      "loss": 0.5793,
+      "step": 80580
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00015379736842105262,
+      "loss": 0.5642,
+      "step": 80590
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00015371842105263157,
+      "loss": 0.5814,
+      "step": 80600
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001536394736842105,
+      "loss": 0.5886,
+      "step": 80610
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00015356052631578944,
+      "loss": 0.5893,
+      "step": 80620
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00015348157894736842,
+      "loss": 0.5857,
+      "step": 80630
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00015340263157894736,
+      "loss": 0.5738,
+      "step": 80640
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001533236842105263,
+      "loss": 0.5745,
+      "step": 80650
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00015324473684210526,
+      "loss": 0.5835,
+      "step": 80660
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001531657894736842,
+      "loss": 0.5711,
+      "step": 80670
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00015308684210526313,
+      "loss": 0.5938,
+      "step": 80680
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00015300789473684208,
+      "loss": 0.5784,
+      "step": 80690
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00015292894736842103,
+      "loss": 0.5649,
+      "step": 80700
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00015284999999999997,
+      "loss": 0.5775,
+      "step": 80710
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00015277105263157895,
+      "loss": 0.5788,
+      "step": 80720
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001526921052631579,
+      "loss": 0.5825,
+      "step": 80730
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00015261315789473685,
+      "loss": 0.5768,
+      "step": 80740
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00015253421052631577,
+      "loss": 0.5765,
+      "step": 80750
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00015245526315789472,
+      "loss": 0.5839,
+      "step": 80760
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00015237631578947366,
+      "loss": 0.5694,
+      "step": 80770
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001522973684210526,
+      "loss": 0.5798,
+      "step": 80780
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00015221842105263156,
+      "loss": 0.5916,
+      "step": 80790
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001521394736842105,
+      "loss": 0.5865,
+      "step": 80800
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00015206052631578943,
+      "loss": 0.585,
+      "step": 80810
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00015198157894736843,
+      "loss": 0.5772,
+      "step": 80820
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00015190263157894735,
+      "loss": 0.5854,
+      "step": 80830
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001518236842105263,
+      "loss": 0.5884,
+      "step": 80840
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00015174473684210525,
+      "loss": 0.5866,
+      "step": 80850
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001516657894736842,
+      "loss": 0.5776,
+      "step": 80860
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00015158684210526315,
+      "loss": 0.5877,
+      "step": 80870
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001515078947368421,
+      "loss": 0.5977,
+      "step": 80880
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00015142894736842102,
+      "loss": 0.5751,
+      "step": 80890
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00015134999999999997,
+      "loss": 0.5772,
+      "step": 80900
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00015127105263157894,
+      "loss": 0.5785,
+      "step": 80910
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001511921052631579,
+      "loss": 0.5769,
+      "step": 80920
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00015111315789473684,
+      "loss": 0.5816,
+      "step": 80930
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00015103421052631579,
+      "loss": 0.5778,
+      "step": 80940
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00015095526315789473,
+      "loss": 0.5827,
+      "step": 80950
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00015087631578947366,
+      "loss": 0.5702,
+      "step": 80960
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001507973684210526,
+      "loss": 0.5725,
+      "step": 80970
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00015071842105263155,
+      "loss": 0.5662,
+      "step": 80980
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001506394736842105,
+      "loss": 0.5715,
+      "step": 80990
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00015056052631578948,
+      "loss": 0.571,
+      "step": 81000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00015048157894736842,
+      "loss": 0.5752,
+      "step": 81010
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00015040263157894737,
+      "loss": 0.5633,
+      "step": 81020
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00015032368421052632,
+      "loss": 0.5713,
+      "step": 81030
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00015024473684210524,
+      "loss": 0.5765,
+      "step": 81040
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001501657894736842,
+      "loss": 0.5673,
+      "step": 81050
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00015008684210526314,
+      "loss": 0.5729,
+      "step": 81060
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001500078947368421,
+      "loss": 0.5759,
+      "step": 81070
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014992894736842103,
+      "loss": 0.5741,
+      "step": 81080
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014984999999999998,
+      "loss": 0.5722,
+      "step": 81090
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014977105263157893,
+      "loss": 0.5785,
+      "step": 81100
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014969210526315788,
+      "loss": 0.5802,
+      "step": 81110
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014961315789473683,
+      "loss": 0.5773,
+      "step": 81120
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014953421052631578,
+      "loss": 0.5748,
+      "step": 81130
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014945526315789472,
+      "loss": 0.5515,
+      "step": 81140
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014937631578947367,
+      "loss": 0.5761,
+      "step": 81150
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014929736842105262,
+      "loss": 0.5603,
+      "step": 81160
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014921842105263157,
+      "loss": 0.5695,
+      "step": 81170
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014913947368421052,
+      "loss": 0.5781,
+      "step": 81180
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014906052631578947,
+      "loss": 0.5679,
+      "step": 81190
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014898157894736841,
+      "loss": 0.5697,
+      "step": 81200
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014890263157894736,
+      "loss": 0.5576,
+      "step": 81210
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001488236842105263,
+      "loss": 0.5711,
+      "step": 81220
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014874473684210526,
+      "loss": 0.5682,
+      "step": 81230
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014866578947368418,
+      "loss": 0.5714,
+      "step": 81240
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014858684210526313,
+      "loss": 0.5631,
+      "step": 81250
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001485078947368421,
+      "loss": 0.5921,
+      "step": 81260
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014842894736842105,
+      "loss": 0.5783,
+      "step": 81270
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014834999999999997,
+      "loss": 0.5733,
+      "step": 81280
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014827105263157892,
+      "loss": 0.5809,
+      "step": 81290
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001481921052631579,
+      "loss": 0.5854,
+      "step": 81300
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014811315789473685,
+      "loss": 0.5747,
+      "step": 81310
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014803421052631577,
+      "loss": 0.5947,
+      "step": 81320
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014795526315789472,
+      "loss": 0.5814,
+      "step": 81330
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014787631578947366,
+      "loss": 0.5882,
+      "step": 81340
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001477973684210526,
+      "loss": 0.574,
+      "step": 81350
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014771842105263156,
+      "loss": 0.5781,
+      "step": 81360
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001476394736842105,
+      "loss": 0.5737,
+      "step": 81370
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014756052631578946,
+      "loss": 0.577,
+      "step": 81380
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001474815789473684,
+      "loss": 0.5697,
+      "step": 81390
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014740263157894735,
+      "loss": 0.5641,
+      "step": 81400
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001473236842105263,
+      "loss": 0.5671,
+      "step": 81410
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014724473684210525,
+      "loss": 0.5739,
+      "step": 81420
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001471657894736842,
+      "loss": 0.5815,
+      "step": 81430
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014708684210526315,
+      "loss": 0.5678,
+      "step": 81440
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001470078947368421,
+      "loss": 0.5733,
+      "step": 81450
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014692894736842104,
+      "loss": 0.5691,
+      "step": 81460
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014685,
+      "loss": 0.5652,
+      "step": 81470
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014677105263157894,
+      "loss": 0.578,
+      "step": 81480
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001466921052631579,
+      "loss": 0.571,
+      "step": 81490
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014661315789473684,
+      "loss": 0.5736,
+      "step": 81500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014653421052631578,
+      "loss": 0.5844,
+      "step": 81510
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014645526315789473,
+      "loss": 0.5876,
+      "step": 81520
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014637631578947365,
+      "loss": 0.5787,
+      "step": 81530
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014629736842105263,
+      "loss": 0.5773,
+      "step": 81540
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014621842105263158,
+      "loss": 0.5797,
+      "step": 81550
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001461394736842105,
+      "loss": 0.5792,
+      "step": 81560
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014606052631578945,
+      "loss": 0.5893,
+      "step": 81570
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014598947368421052,
+      "loss": 0.5873,
+      "step": 81580
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014591052631578944,
+      "loss": 0.5808,
+      "step": 81590
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014583157894736842,
+      "loss": 0.5913,
+      "step": 81600
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014575263157894737,
+      "loss": 0.5545,
+      "step": 81610
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014567368421052632,
+      "loss": 0.557,
+      "step": 81620
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014559473684210524,
+      "loss": 0.5605,
+      "step": 81630
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014551578947368419,
+      "loss": 0.5748,
+      "step": 81640
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014543684210526316,
+      "loss": 0.5755,
+      "step": 81650
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014535789473684208,
+      "loss": 0.582,
+      "step": 81660
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014527894736842103,
+      "loss": 0.5609,
+      "step": 81670
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014519999999999998,
+      "loss": 0.5772,
+      "step": 81680
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014512105263157893,
+      "loss": 0.5669,
+      "step": 81690
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014504210526315788,
+      "loss": 0.5783,
+      "step": 81700
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014496315789473682,
+      "loss": 0.5776,
+      "step": 81710
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014488421052631577,
+      "loss": 0.5697,
+      "step": 81720
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014480526315789472,
+      "loss": 0.5707,
+      "step": 81730
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014472631578947367,
+      "loss": 0.5693,
+      "step": 81740
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014464736842105262,
+      "loss": 0.5816,
+      "step": 81750
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014456842105263157,
+      "loss": 0.5844,
+      "step": 81760
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001444894736842105,
+      "loss": 0.574,
+      "step": 81770
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014441052631578946,
+      "loss": 0.5768,
+      "step": 81780
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001443315789473684,
+      "loss": 0.5779,
+      "step": 81790
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014425263157894736,
+      "loss": 0.5867,
+      "step": 81800
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001441736842105263,
+      "loss": 0.5844,
+      "step": 81810
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014409473684210525,
+      "loss": 0.5804,
+      "step": 81820
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001440157894736842,
+      "loss": 0.5758,
+      "step": 81830
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014393684210526315,
+      "loss": 0.5703,
+      "step": 81840
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001438578947368421,
+      "loss": 0.5617,
+      "step": 81850
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014377894736842105,
+      "loss": 0.5782,
+      "step": 81860
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014369999999999997,
+      "loss": 0.5749,
+      "step": 81870
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014362105263157894,
+      "loss": 0.5667,
+      "step": 81880
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001435421052631579,
+      "loss": 0.5723,
+      "step": 81890
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014346315789473684,
+      "loss": 0.5713,
+      "step": 81900
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014338421052631576,
+      "loss": 0.5775,
+      "step": 81910
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001433052631578947,
+      "loss": 0.5775,
+      "step": 81920
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014322631578947369,
+      "loss": 0.5758,
+      "step": 81930
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014314736842105263,
+      "loss": 0.5675,
+      "step": 81940
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014306842105263156,
+      "loss": 0.5707,
+      "step": 81950
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001429894736842105,
+      "loss": 0.5756,
+      "step": 81960
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014291052631578945,
+      "loss": 0.5799,
+      "step": 81970
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001428315789473684,
+      "loss": 0.5749,
+      "step": 81980
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014275263157894735,
+      "loss": 0.5787,
+      "step": 81990
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001426736842105263,
+      "loss": 0.5879,
+      "step": 82000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014259473684210525,
+      "loss": 0.5665,
+      "step": 82010
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001425157894736842,
+      "loss": 0.5831,
+      "step": 82020
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014243684210526314,
+      "loss": 0.5762,
+      "step": 82030
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001423578947368421,
+      "loss": 0.5659,
+      "step": 82040
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014227894736842104,
+      "loss": 0.5767,
+      "step": 82050
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001422,
+      "loss": 0.5791,
+      "step": 82060
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014212105263157894,
+      "loss": 0.5824,
+      "step": 82070
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014204210526315788,
+      "loss": 0.5663,
+      "step": 82080
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014196315789473683,
+      "loss": 0.5722,
+      "step": 82090
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014188421052631578,
+      "loss": 0.5732,
+      "step": 82100
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014180526315789473,
+      "loss": 0.5631,
+      "step": 82110
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014172631578947368,
+      "loss": 0.5808,
+      "step": 82120
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014164736842105263,
+      "loss": 0.5758,
+      "step": 82130
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014156842105263157,
+      "loss": 0.5772,
+      "step": 82140
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001414894736842105,
+      "loss": 0.5568,
+      "step": 82150
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014141052631578947,
+      "loss": 0.5579,
+      "step": 82160
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014133157894736842,
+      "loss": 0.557,
+      "step": 82170
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014125263157894737,
+      "loss": 0.5525,
+      "step": 82180
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001411736842105263,
+      "loss": 0.5627,
+      "step": 82190
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014109473684210524,
+      "loss": 0.5533,
+      "step": 82200
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001410157894736842,
+      "loss": 0.5594,
+      "step": 82210
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014093684210526316,
+      "loss": 0.5646,
+      "step": 82220
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014085789473684208,
+      "loss": 0.5733,
+      "step": 82230
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014077894736842103,
+      "loss": 0.5598,
+      "step": 82240
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014069999999999998,
+      "loss": 0.5533,
+      "step": 82250
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014062105263157895,
+      "loss": 0.5643,
+      "step": 82260
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014054210526315787,
+      "loss": 0.5852,
+      "step": 82270
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014046315789473682,
+      "loss": 0.5951,
+      "step": 82280
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014038421052631577,
+      "loss": 0.5833,
+      "step": 82290
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014030526315789472,
+      "loss": 0.5765,
+      "step": 82300
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014022631578947367,
+      "loss": 0.569,
+      "step": 82310
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014014736842105262,
+      "loss": 0.5765,
+      "step": 82320
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014006842105263156,
+      "loss": 0.5693,
+      "step": 82330
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001399894736842105,
+      "loss": 0.5802,
+      "step": 82340
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013991052631578946,
+      "loss": 0.5775,
+      "step": 82350
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001398315789473684,
+      "loss": 0.5754,
+      "step": 82360
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013975263157894736,
+      "loss": 0.5753,
+      "step": 82370
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001396736842105263,
+      "loss": 0.5813,
+      "step": 82380
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013959473684210525,
+      "loss": 0.5823,
+      "step": 82390
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001395157894736842,
+      "loss": 0.5783,
+      "step": 82400
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013943684210526315,
+      "loss": 0.5688,
+      "step": 82410
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001393578947368421,
+      "loss": 0.584,
+      "step": 82420
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013927894736842105,
+      "loss": 0.5766,
+      "step": 82430
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001392,
+      "loss": 0.5804,
+      "step": 82440
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013912105263157894,
+      "loss": 0.587,
+      "step": 82450
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001390421052631579,
+      "loss": 0.58,
+      "step": 82460
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001389631578947368,
+      "loss": 0.5762,
+      "step": 82470
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013888421052631576,
+      "loss": 0.5972,
+      "step": 82480
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013880526315789474,
+      "loss": 0.5743,
+      "step": 82490
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013872631578947369,
+      "loss": 0.5889,
+      "step": 82500
+    },
+    {
+      "epoch": 0.82,
+      "eval_accuracy": 0.8794982706220447,
+      "eval_loss": 0.5517578125,
+      "eval_runtime": 100.2401,
+      "eval_samples_per_second": 798.084,
+      "eval_steps_per_second": 1.566,
+      "step": 82500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001386473684210526,
+      "loss": 0.5757,
+      "step": 82510
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013856842105263155,
+      "loss": 0.5756,
+      "step": 82520
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013848947368421053,
+      "loss": 0.5738,
+      "step": 82530
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013841052631578948,
+      "loss": 0.5746,
+      "step": 82540
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013833947368421053,
+      "loss": 0.5783,
+      "step": 82550
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013826052631578947,
+      "loss": 0.5797,
+      "step": 82560
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013818157894736842,
+      "loss": 0.5635,
+      "step": 82570
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013810263157894734,
+      "loss": 0.5661,
+      "step": 82580
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001380236842105263,
+      "loss": 0.5643,
+      "step": 82590
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013794473684210527,
+      "loss": 0.5672,
+      "step": 82600
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001378657894736842,
+      "loss": 0.5687,
+      "step": 82610
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013778684210526314,
+      "loss": 0.5608,
+      "step": 82620
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013770789473684209,
+      "loss": 0.5723,
+      "step": 82630
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013762894736842103,
+      "loss": 0.5653,
+      "step": 82640
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013754999999999998,
+      "loss": 0.5741,
+      "step": 82650
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013747105263157893,
+      "loss": 0.5806,
+      "step": 82660
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013739210526315788,
+      "loss": 0.5871,
+      "step": 82670
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013731315789473683,
+      "loss": 0.5728,
+      "step": 82680
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013723421052631578,
+      "loss": 0.5813,
+      "step": 82690
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013715526315789472,
+      "loss": 0.589,
+      "step": 82700
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013707631578947367,
+      "loss": 0.5815,
+      "step": 82710
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013699736842105262,
+      "loss": 0.5947,
+      "step": 82720
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013691842105263157,
+      "loss": 0.575,
+      "step": 82730
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013683947368421052,
+      "loss": 0.5756,
+      "step": 82740
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013676052631578947,
+      "loss": 0.5771,
+      "step": 82750
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013668157894736841,
+      "loss": 0.5708,
+      "step": 82760
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013660263157894736,
+      "loss": 0.5732,
+      "step": 82770
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013652368421052628,
+      "loss": 0.5662,
+      "step": 82780
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013644473684210526,
+      "loss": 0.5629,
+      "step": 82790
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001363657894736842,
+      "loss": 0.5633,
+      "step": 82800
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013628684210526316,
+      "loss": 0.5677,
+      "step": 82810
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013620789473684208,
+      "loss": 0.5756,
+      "step": 82820
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013612894736842105,
+      "loss": 0.5683,
+      "step": 82830
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013605,
+      "loss": 0.5652,
+      "step": 82840
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013597105263157895,
+      "loss": 0.5726,
+      "step": 82850
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013589210526315787,
+      "loss": 0.5673,
+      "step": 82860
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013581315789473682,
+      "loss": 0.5751,
+      "step": 82870
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001357342105263158,
+      "loss": 0.5849,
+      "step": 82880
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013565526315789471,
+      "loss": 0.5781,
+      "step": 82890
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013557631578947366,
+      "loss": 0.5906,
+      "step": 82900
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001354973684210526,
+      "loss": 0.5825,
+      "step": 82910
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013541842105263156,
+      "loss": 0.5814,
+      "step": 82920
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001353394736842105,
+      "loss": 0.5805,
+      "step": 82930
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013526052631578946,
+      "loss": 0.59,
+      "step": 82940
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001351815789473684,
+      "loss": 0.5801,
+      "step": 82950
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013510263157894735,
+      "loss": 0.5804,
+      "step": 82960
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001350236842105263,
+      "loss": 0.5688,
+      "step": 82970
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013494473684210525,
+      "loss": 0.5784,
+      "step": 82980
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001348657894736842,
+      "loss": 0.5782,
+      "step": 82990
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013478684210526315,
+      "loss": 0.5714,
+      "step": 83000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001347078947368421,
+      "loss": 0.5778,
+      "step": 83010
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013462894736842104,
+      "loss": 0.5644,
+      "step": 83020
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013455,
+      "loss": 0.5734,
+      "step": 83030
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013447105263157894,
+      "loss": 0.5741,
+      "step": 83040
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001343921052631579,
+      "loss": 0.571,
+      "step": 83050
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013431315789473684,
+      "loss": 0.57,
+      "step": 83060
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013423421052631578,
+      "loss": 0.5812,
+      "step": 83070
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013415526315789473,
+      "loss": 0.5815,
+      "step": 83080
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013407631578947368,
+      "loss": 0.5782,
+      "step": 83090
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001339973684210526,
+      "loss": 0.5726,
+      "step": 83100
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013391842105263158,
+      "loss": 0.5714,
+      "step": 83110
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013383947368421053,
+      "loss": 0.5716,
+      "step": 83120
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013376052631578947,
+      "loss": 0.5585,
+      "step": 83130
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001336815789473684,
+      "loss": 0.5718,
+      "step": 83140
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013360263157894734,
+      "loss": 0.5602,
+      "step": 83150
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013352368421052632,
+      "loss": 0.5593,
+      "step": 83160
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013344473684210527,
+      "loss": 0.5625,
+      "step": 83170
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001333657894736842,
+      "loss": 0.572,
+      "step": 83180
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013328684210526314,
+      "loss": 0.5735,
+      "step": 83190
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013320789473684208,
+      "loss": 0.5645,
+      "step": 83200
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013312894736842103,
+      "loss": 0.5663,
+      "step": 83210
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013304999999999998,
+      "loss": 0.5636,
+      "step": 83220
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013297105263157893,
+      "loss": 0.5556,
+      "step": 83230
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013289210526315788,
+      "loss": 0.5663,
+      "step": 83240
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013281315789473683,
+      "loss": 0.5591,
+      "step": 83250
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013273421052631577,
+      "loss": 0.5604,
+      "step": 83260
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013265526315789472,
+      "loss": 0.5677,
+      "step": 83270
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013257631578947367,
+      "loss": 0.5581,
+      "step": 83280
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013249736842105262,
+      "loss": 0.5548,
+      "step": 83290
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013241842105263157,
+      "loss": 0.5691,
+      "step": 83300
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013233947368421052,
+      "loss": 0.5538,
+      "step": 83310
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013226052631578946,
+      "loss": 0.563,
+      "step": 83320
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001321815789473684,
+      "loss": 0.5574,
+      "step": 83330
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013210263157894736,
+      "loss": 0.5488,
+      "step": 83340
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001320236842105263,
+      "loss": 0.5604,
+      "step": 83350
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013194473684210526,
+      "loss": 0.562,
+      "step": 83360
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001318657894736842,
+      "loss": 0.5502,
+      "step": 83370
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013178684210526313,
+      "loss": 0.5579,
+      "step": 83380
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001317078947368421,
+      "loss": 0.5547,
+      "step": 83390
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013162894736842105,
+      "loss": 0.5577,
+      "step": 83400
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013155,
+      "loss": 0.5585,
+      "step": 83410
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013147105263157892,
+      "loss": 0.5655,
+      "step": 83420
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013139210526315787,
+      "loss": 0.5482,
+      "step": 83430
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013131315789473684,
+      "loss": 0.5582,
+      "step": 83440
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001312342105263158,
+      "loss": 0.5532,
+      "step": 83450
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013115526315789471,
+      "loss": 0.5531,
+      "step": 83460
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013107631578947366,
+      "loss": 0.5593,
+      "step": 83470
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001309973684210526,
+      "loss": 0.5784,
+      "step": 83480
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013091842105263159,
+      "loss": 0.5901,
+      "step": 83490
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001308394736842105,
+      "loss": 0.5839,
+      "step": 83500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013076052631578946,
+      "loss": 0.5809,
+      "step": 83510
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001306815789473684,
+      "loss": 0.5645,
+      "step": 83520
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013060263157894735,
+      "loss": 0.557,
+      "step": 83530
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001305236842105263,
+      "loss": 0.5608,
+      "step": 83540
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013044473684210525,
+      "loss": 0.5725,
+      "step": 83550
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001303657894736842,
+      "loss": 0.5722,
+      "step": 83560
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013028684210526314,
+      "loss": 0.5826,
+      "step": 83570
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001302078947368421,
+      "loss": 0.5803,
+      "step": 83580
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013012894736842104,
+      "loss": 0.5769,
+      "step": 83590
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013005,
+      "loss": 0.5756,
+      "step": 83600
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012997105263157894,
+      "loss": 0.5828,
+      "step": 83610
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012989210526315789,
+      "loss": 0.5682,
+      "step": 83620
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012981315789473683,
+      "loss": 0.5652,
+      "step": 83630
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012973421052631578,
+      "loss": 0.5668,
+      "step": 83640
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012965526315789473,
+      "loss": 0.5774,
+      "step": 83650
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012957631578947368,
+      "loss": 0.5789,
+      "step": 83660
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012949736842105263,
+      "loss": 0.5753,
+      "step": 83670
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012941842105263158,
+      "loss": 0.5741,
+      "step": 83680
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012933947368421052,
+      "loss": 0.577,
+      "step": 83690
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012926052631578945,
+      "loss": 0.5776,
+      "step": 83700
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001291815789473684,
+      "loss": 0.581,
+      "step": 83710
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012910263157894737,
+      "loss": 0.5839,
+      "step": 83720
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012902368421052632,
+      "loss": 0.5858,
+      "step": 83730
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012894473684210524,
+      "loss": 0.5822,
+      "step": 83740
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001288657894736842,
+      "loss": 0.5686,
+      "step": 83750
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012878684210526314,
+      "loss": 0.5732,
+      "step": 83760
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001287078947368421,
+      "loss": 0.5713,
+      "step": 83770
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012862894736842103,
+      "loss": 0.5744,
+      "step": 83780
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012854999999999998,
+      "loss": 0.5802,
+      "step": 83790
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012847105263157893,
+      "loss": 0.5658,
+      "step": 83800
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012839210526315788,
+      "loss": 0.5678,
+      "step": 83810
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012831315789473683,
+      "loss": 0.5731,
+      "step": 83820
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012823421052631577,
+      "loss": 0.567,
+      "step": 83830
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012815526315789472,
+      "loss": 0.5827,
+      "step": 83840
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012807631578947367,
+      "loss": 0.5555,
+      "step": 83850
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012799736842105262,
+      "loss": 0.5572,
+      "step": 83860
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012791842105263157,
+      "loss": 0.5613,
+      "step": 83870
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012783947368421052,
+      "loss": 0.5544,
+      "step": 83880
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012776052631578946,
+      "loss": 0.5537,
+      "step": 83890
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001276815789473684,
+      "loss": 0.5513,
+      "step": 83900
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012760263157894736,
+      "loss": 0.5503,
+      "step": 83910
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001275236842105263,
+      "loss": 0.5586,
+      "step": 83920
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012744473684210526,
+      "loss": 0.5536,
+      "step": 83930
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001273657894736842,
+      "loss": 0.559,
+      "step": 83940
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012728684210526315,
+      "loss": 0.5508,
+      "step": 83950
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001272078947368421,
+      "loss": 0.5663,
+      "step": 83960
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012712894736842105,
+      "loss": 0.5546,
+      "step": 83970
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012705,
+      "loss": 0.5634,
+      "step": 83980
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012697105263157892,
+      "loss": 0.551,
+      "step": 83990
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001268921052631579,
+      "loss": 0.563,
+      "step": 84000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012681315789473684,
+      "loss": 0.5495,
+      "step": 84010
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012673421052631576,
+      "loss": 0.5614,
+      "step": 84020
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001266552631578947,
+      "loss": 0.5684,
+      "step": 84030
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001265763157894737,
+      "loss": 0.5754,
+      "step": 84040
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012649736842105264,
+      "loss": 0.5728,
+      "step": 84050
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012641842105263156,
+      "loss": 0.577,
+      "step": 84060
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001263394736842105,
+      "loss": 0.5716,
+      "step": 84070
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012626052631578945,
+      "loss": 0.5776,
+      "step": 84080
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012618157894736843,
+      "loss": 0.5699,
+      "step": 84090
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012610263157894735,
+      "loss": 0.5577,
+      "step": 84100
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001260236842105263,
+      "loss": 0.5633,
+      "step": 84110
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012594473684210525,
+      "loss": 0.5601,
+      "step": 84120
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001258657894736842,
+      "loss": 0.5728,
+      "step": 84130
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012578684210526314,
+      "loss": 0.5721,
+      "step": 84140
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001257078947368421,
+      "loss": 0.5712,
+      "step": 84150
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012562894736842104,
+      "loss": 0.5693,
+      "step": 84160
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012555,
+      "loss": 0.5685,
+      "step": 84170
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012547105263157894,
+      "loss": 0.5685,
+      "step": 84180
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012539210526315789,
+      "loss": 0.5807,
+      "step": 84190
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012531315789473683,
+      "loss": 0.5684,
+      "step": 84200
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012523421052631578,
+      "loss": 0.5761,
+      "step": 84210
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012515526315789473,
+      "loss": 0.5613,
+      "step": 84220
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012507631578947368,
+      "loss": 0.5687,
+      "step": 84230
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012499736842105263,
+      "loss": 0.5559,
+      "step": 84240
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012491842105263158,
+      "loss": 0.5647,
+      "step": 84250
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012483947368421052,
+      "loss": 0.5591,
+      "step": 84260
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012476052631578944,
+      "loss": 0.5701,
+      "step": 84270
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012468157894736842,
+      "loss": 0.5585,
+      "step": 84280
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012460263157894737,
+      "loss": 0.5514,
+      "step": 84290
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012452368421052632,
+      "loss": 0.5582,
+      "step": 84300
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012444473684210524,
+      "loss": 0.562,
+      "step": 84310
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001243657894736842,
+      "loss": 0.564,
+      "step": 84320
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012428684210526316,
+      "loss": 0.5697,
+      "step": 84330
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012420789473684208,
+      "loss": 0.5632,
+      "step": 84340
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012412894736842103,
+      "loss": 0.5635,
+      "step": 84350
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012404999999999998,
+      "loss": 0.5632,
+      "step": 84360
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012397105263157895,
+      "loss": 0.5672,
+      "step": 84370
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012389210526315788,
+      "loss": 0.5793,
+      "step": 84380
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012381315789473682,
+      "loss": 0.5808,
+      "step": 84390
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012373421052631577,
+      "loss": 0.564,
+      "step": 84400
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012365526315789472,
+      "loss": 0.5755,
+      "step": 84410
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012357631578947367,
+      "loss": 0.5771,
+      "step": 84420
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012349736842105262,
+      "loss": 0.5823,
+      "step": 84430
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012341842105263157,
+      "loss": 0.5742,
+      "step": 84440
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012333947368421051,
+      "loss": 0.5693,
+      "step": 84450
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012326052631578946,
+      "loss": 0.567,
+      "step": 84460
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001231815789473684,
+      "loss": 0.5529,
+      "step": 84470
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012310263157894736,
+      "loss": 0.5767,
+      "step": 84480
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001230236842105263,
+      "loss": 0.5642,
+      "step": 84490
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012294473684210526,
+      "loss": 0.5659,
+      "step": 84500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001228657894736842,
+      "loss": 0.559,
+      "step": 84510
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012278684210526315,
+      "loss": 0.5606,
+      "step": 84520
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001227078947368421,
+      "loss": 0.5623,
+      "step": 84530
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012262894736842105,
+      "loss": 0.5589,
+      "step": 84540
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001225578947368421,
+      "loss": 0.5546,
+      "step": 84550
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012247894736842105,
+      "loss": 0.5621,
+      "step": 84560
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001224,
+      "loss": 0.5675,
+      "step": 84570
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012232105263157894,
+      "loss": 0.5709,
+      "step": 84580
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001222421052631579,
+      "loss": 0.5649,
+      "step": 84590
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012216315789473684,
+      "loss": 0.5662,
+      "step": 84600
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012208421052631576,
+      "loss": 0.567,
+      "step": 84610
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012200526315789474,
+      "loss": 0.574,
+      "step": 84620
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012192631578947368,
+      "loss": 0.5738,
+      "step": 84630
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012184736842105262,
+      "loss": 0.5599,
+      "step": 84640
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012176842105263157,
+      "loss": 0.573,
+      "step": 84650
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012168947368421051,
+      "loss": 0.5676,
+      "step": 84660
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012161052631578946,
+      "loss": 0.576,
+      "step": 84670
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012153157894736841,
+      "loss": 0.5811,
+      "step": 84680
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012145263157894736,
+      "loss": 0.5674,
+      "step": 84690
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001213736842105263,
+      "loss": 0.5682,
+      "step": 84700
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012129473684210524,
+      "loss": 0.5699,
+      "step": 84710
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001212157894736842,
+      "loss": 0.5615,
+      "step": 84720
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012113684210526315,
+      "loss": 0.5685,
+      "step": 84730
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012105789473684209,
+      "loss": 0.5693,
+      "step": 84740
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012097894736842104,
+      "loss": 0.5577,
+      "step": 84750
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001209,
+      "loss": 0.5568,
+      "step": 84760
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012082105263157895,
+      "loss": 0.5626,
+      "step": 84770
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012074210526315788,
+      "loss": 0.556,
+      "step": 84780
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012066315789473683,
+      "loss": 0.5557,
+      "step": 84790
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012058421052631578,
+      "loss": 0.5754,
+      "step": 84800
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012050526315789473,
+      "loss": 0.5728,
+      "step": 84810
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012042631578947367,
+      "loss": 0.5641,
+      "step": 84820
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012034736842105262,
+      "loss": 0.5645,
+      "step": 84830
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012026842105263157,
+      "loss": 0.5695,
+      "step": 84840
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001201894736842105,
+      "loss": 0.5729,
+      "step": 84850
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012011052631578947,
+      "loss": 0.5728,
+      "step": 84860
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012003157894736842,
+      "loss": 0.5681,
+      "step": 84870
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011995263157894735,
+      "loss": 0.5715,
+      "step": 84880
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001198736842105263,
+      "loss": 0.5862,
+      "step": 84890
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011979473684210526,
+      "loss": 0.5762,
+      "step": 84900
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011971578947368421,
+      "loss": 0.5818,
+      "step": 84910
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011963684210526314,
+      "loss": 0.5682,
+      "step": 84920
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011955789473684209,
+      "loss": 0.5733,
+      "step": 84930
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011947894736842104,
+      "loss": 0.5591,
+      "step": 84940
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001194,
+      "loss": 0.5695,
+      "step": 84950
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011932105263157894,
+      "loss": 0.5686,
+      "step": 84960
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011924210526315789,
+      "loss": 0.5588,
+      "step": 84970
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011916315789473683,
+      "loss": 0.5544,
+      "step": 84980
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011908421052631577,
+      "loss": 0.5651,
+      "step": 84990
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011900526315789473,
+      "loss": 0.5688,
+      "step": 85000
+    },
+    {
+      "epoch": 0.85,
+      "eval_accuracy": 0.8806242189602479,
+      "eval_loss": 0.54638671875,
+      "eval_runtime": 98.7484,
+      "eval_samples_per_second": 810.14,
+      "eval_steps_per_second": 1.59,
+      "step": 85000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011892631578947368,
+      "loss": 0.5437,
+      "step": 85010
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011884736842105261,
+      "loss": 0.5461,
+      "step": 85020
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011876842105263156,
+      "loss": 0.5477,
+      "step": 85030
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011868947368421052,
+      "loss": 0.5339,
+      "step": 85040
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011861052631578947,
+      "loss": 0.5559,
+      "step": 85050
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001185315789473684,
+      "loss": 0.5535,
+      "step": 85060
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011845263157894735,
+      "loss": 0.5652,
+      "step": 85070
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001183736842105263,
+      "loss": 0.5791,
+      "step": 85080
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011829473684210526,
+      "loss": 0.5687,
+      "step": 85090
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001182157894736842,
+      "loss": 0.5531,
+      "step": 85100
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011813684210526315,
+      "loss": 0.5756,
+      "step": 85110
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001180578947368421,
+      "loss": 0.577,
+      "step": 85120
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011797894736842103,
+      "loss": 0.5682,
+      "step": 85130
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011789999999999999,
+      "loss": 0.5811,
+      "step": 85140
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011782105263157894,
+      "loss": 0.5629,
+      "step": 85150
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011774210526315788,
+      "loss": 0.5632,
+      "step": 85160
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011766315789473682,
+      "loss": 0.5655,
+      "step": 85170
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011758421052631579,
+      "loss": 0.5734,
+      "step": 85180
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011750526315789473,
+      "loss": 0.5627,
+      "step": 85190
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011742631578947367,
+      "loss": 0.5543,
+      "step": 85200
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011734736842105262,
+      "loss": 0.5608,
+      "step": 85210
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011726842105263157,
+      "loss": 0.5596,
+      "step": 85220
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011718947368421053,
+      "loss": 0.5672,
+      "step": 85230
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011711052631578946,
+      "loss": 0.5619,
+      "step": 85240
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011703157894736841,
+      "loss": 0.5518,
+      "step": 85250
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011695263157894736,
+      "loss": 0.5648,
+      "step": 85260
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011687368421052632,
+      "loss": 0.5722,
+      "step": 85270
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011679473684210526,
+      "loss": 0.5628,
+      "step": 85280
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001167157894736842,
+      "loss": 0.573,
+      "step": 85290
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011663684210526315,
+      "loss": 0.5702,
+      "step": 85300
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011655789473684209,
+      "loss": 0.5746,
+      "step": 85310
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011647894736842105,
+      "loss": 0.5577,
+      "step": 85320
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001164,
+      "loss": 0.5549,
+      "step": 85330
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011632105263157893,
+      "loss": 0.5478,
+      "step": 85340
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011624210526315788,
+      "loss": 0.567,
+      "step": 85350
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011616315789473683,
+      "loss": 0.5715,
+      "step": 85360
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011608421052631579,
+      "loss": 0.5782,
+      "step": 85370
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011600526315789472,
+      "loss": 0.5762,
+      "step": 85380
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011592631578947367,
+      "loss": 0.5778,
+      "step": 85390
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011584736842105262,
+      "loss": 0.5692,
+      "step": 85400
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011576842105263158,
+      "loss": 0.5604,
+      "step": 85410
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011568947368421052,
+      "loss": 0.5752,
+      "step": 85420
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011561052631578947,
+      "loss": 0.5626,
+      "step": 85430
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011553157894736841,
+      "loss": 0.5703,
+      "step": 85440
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011545263157894735,
+      "loss": 0.5747,
+      "step": 85450
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011537368421052631,
+      "loss": 0.5706,
+      "step": 85460
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011529473684210526,
+      "loss": 0.5577,
+      "step": 85470
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001152157894736842,
+      "loss": 0.56,
+      "step": 85480
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011513684210526314,
+      "loss": 0.5659,
+      "step": 85490
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011505789473684209,
+      "loss": 0.5639,
+      "step": 85500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011497894736842105,
+      "loss": 0.5692,
+      "step": 85510
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011489999999999999,
+      "loss": 0.5699,
+      "step": 85520
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011482105263157894,
+      "loss": 0.5704,
+      "step": 85530
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011474210526315788,
+      "loss": 0.5697,
+      "step": 85540
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011466315789473685,
+      "loss": 0.572,
+      "step": 85550
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011459210526315788,
+      "loss": 0.5686,
+      "step": 85560
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011451315789473684,
+      "loss": 0.5621,
+      "step": 85570
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011443421052631578,
+      "loss": 0.5729,
+      "step": 85580
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011435526315789473,
+      "loss": 0.5724,
+      "step": 85590
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011427631578947367,
+      "loss": 0.5769,
+      "step": 85600
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011419736842105262,
+      "loss": 0.5802,
+      "step": 85610
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011411842105263157,
+      "loss": 0.5811,
+      "step": 85620
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011403947368421052,
+      "loss": 0.5598,
+      "step": 85630
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011396052631578947,
+      "loss": 0.5693,
+      "step": 85640
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001138815789473684,
+      "loss": 0.5682,
+      "step": 85650
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011380263157894735,
+      "loss": 0.5734,
+      "step": 85660
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011372368421052631,
+      "loss": 0.5715,
+      "step": 85670
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011364473684210526,
+      "loss": 0.5658,
+      "step": 85680
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001135657894736842,
+      "loss": 0.5767,
+      "step": 85690
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011348684210526314,
+      "loss": 0.5685,
+      "step": 85700
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001134078947368421,
+      "loss": 0.5574,
+      "step": 85710
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011332894736842105,
+      "loss": 0.5676,
+      "step": 85720
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011324999999999999,
+      "loss": 0.5559,
+      "step": 85730
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011317105263157894,
+      "loss": 0.5546,
+      "step": 85740
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011309210526315788,
+      "loss": 0.554,
+      "step": 85750
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011301315789473683,
+      "loss": 0.5602,
+      "step": 85760
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011293421052631578,
+      "loss": 0.5702,
+      "step": 85770
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011285526315789473,
+      "loss": 0.5716,
+      "step": 85780
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011277631578947366,
+      "loss": 0.5761,
+      "step": 85790
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011269736842105261,
+      "loss": 0.5572,
+      "step": 85800
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011261842105263157,
+      "loss": 0.5822,
+      "step": 85810
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011253947368421052,
+      "loss": 0.5724,
+      "step": 85820
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011246052631578946,
+      "loss": 0.5678,
+      "step": 85830
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001123815789473684,
+      "loss": 0.5632,
+      "step": 85840
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011230263157894737,
+      "loss": 0.5662,
+      "step": 85850
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011222368421052632,
+      "loss": 0.5786,
+      "step": 85860
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011214473684210525,
+      "loss": 0.5667,
+      "step": 85870
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001120657894736842,
+      "loss": 0.5742,
+      "step": 85880
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011198684210526315,
+      "loss": 0.5646,
+      "step": 85890
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001119078947368421,
+      "loss": 0.5775,
+      "step": 85900
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011182894736842104,
+      "loss": 0.5587,
+      "step": 85910
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011174999999999999,
+      "loss": 0.5676,
+      "step": 85920
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011167105263157893,
+      "loss": 0.5542,
+      "step": 85930
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011159210526315788,
+      "loss": 0.5562,
+      "step": 85940
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011151315789473684,
+      "loss": 0.555,
+      "step": 85950
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011143421052631579,
+      "loss": 0.5691,
+      "step": 85960
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011135526315789472,
+      "loss": 0.5806,
+      "step": 85970
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011127631578947367,
+      "loss": 0.5658,
+      "step": 85980
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011119736842105263,
+      "loss": 0.5802,
+      "step": 85990
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011111842105263158,
+      "loss": 0.5685,
+      "step": 86000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011103947368421051,
+      "loss": 0.5868,
+      "step": 86010
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011096052631578946,
+      "loss": 0.5774,
+      "step": 86020
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011088157894736841,
+      "loss": 0.5792,
+      "step": 86030
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011080263157894736,
+      "loss": 0.572,
+      "step": 86040
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001107236842105263,
+      "loss": 0.5763,
+      "step": 86050
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011064473684210525,
+      "loss": 0.5744,
+      "step": 86060
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001105657894736842,
+      "loss": 0.5807,
+      "step": 86070
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011048684210526314,
+      "loss": 0.578,
+      "step": 86080
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001104078947368421,
+      "loss": 0.5676,
+      "step": 86090
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011032894736842105,
+      "loss": 0.5795,
+      "step": 86100
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011024999999999998,
+      "loss": 0.5555,
+      "step": 86110
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011017105263157893,
+      "loss": 0.5618,
+      "step": 86120
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011009210526315789,
+      "loss": 0.572,
+      "step": 86130
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011001315789473684,
+      "loss": 0.5662,
+      "step": 86140
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010993421052631578,
+      "loss": 0.5739,
+      "step": 86150
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010985526315789472,
+      "loss": 0.5603,
+      "step": 86160
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010977631578947367,
+      "loss": 0.5634,
+      "step": 86170
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010969736842105263,
+      "loss": 0.5684,
+      "step": 86180
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010961842105263157,
+      "loss": 0.5711,
+      "step": 86190
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010953947368421052,
+      "loss": 0.5731,
+      "step": 86200
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010946052631578947,
+      "loss": 0.5663,
+      "step": 86210
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001093815789473684,
+      "loss": 0.569,
+      "step": 86220
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010930263157894736,
+      "loss": 0.566,
+      "step": 86230
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010922368421052631,
+      "loss": 0.5642,
+      "step": 86240
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010914473684210525,
+      "loss": 0.5614,
+      "step": 86250
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001090657894736842,
+      "loss": 0.554,
+      "step": 86260
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010898684210526316,
+      "loss": 0.5498,
+      "step": 86270
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001089078947368421,
+      "loss": 0.5544,
+      "step": 86280
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010882894736842104,
+      "loss": 0.5617,
+      "step": 86290
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010874999999999999,
+      "loss": 0.5575,
+      "step": 86300
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010867105263157894,
+      "loss": 0.5585,
+      "step": 86310
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001085921052631579,
+      "loss": 0.5603,
+      "step": 86320
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010851315789473683,
+      "loss": 0.5498,
+      "step": 86330
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010843421052631578,
+      "loss": 0.5574,
+      "step": 86340
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010835526315789473,
+      "loss": 0.5768,
+      "step": 86350
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010827631578947366,
+      "loss": 0.5804,
+      "step": 86360
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010819736842105263,
+      "loss": 0.5593,
+      "step": 86370
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010811842105263157,
+      "loss": 0.5625,
+      "step": 86380
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010803947368421051,
+      "loss": 0.5592,
+      "step": 86390
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010796052631578946,
+      "loss": 0.5645,
+      "step": 86400
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010788157894736842,
+      "loss": 0.5598,
+      "step": 86410
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010780263157894737,
+      "loss": 0.567,
+      "step": 86420
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001077236842105263,
+      "loss": 0.5729,
+      "step": 86430
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010764473684210525,
+      "loss": 0.5636,
+      "step": 86440
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001075657894736842,
+      "loss": 0.5656,
+      "step": 86450
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010748684210526316,
+      "loss": 0.571,
+      "step": 86460
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001074078947368421,
+      "loss": 0.5764,
+      "step": 86470
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010732894736842104,
+      "loss": 0.5652,
+      "step": 86480
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010724999999999999,
+      "loss": 0.5716,
+      "step": 86490
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010717105263157893,
+      "loss": 0.572,
+      "step": 86500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010709210526315789,
+      "loss": 0.5799,
+      "step": 86510
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010701315789473684,
+      "loss": 0.5555,
+      "step": 86520
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010693421052631578,
+      "loss": 0.568,
+      "step": 86530
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010685526315789472,
+      "loss": 0.5589,
+      "step": 86540
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010677631578947368,
+      "loss": 0.5774,
+      "step": 86550
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010670526315789472,
+      "loss": 0.5726,
+      "step": 86560
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010662631578947368,
+      "loss": 0.5817,
+      "step": 86570
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010654736842105263,
+      "loss": 0.5636,
+      "step": 86580
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010646842105263157,
+      "loss": 0.5581,
+      "step": 86590
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010638947368421051,
+      "loss": 0.5577,
+      "step": 86600
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010631052631578946,
+      "loss": 0.5483,
+      "step": 86610
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010623157894736842,
+      "loss": 0.5685,
+      "step": 86620
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010615263157894737,
+      "loss": 0.5518,
+      "step": 86630
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001060736842105263,
+      "loss": 0.5655,
+      "step": 86640
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010599473684210525,
+      "loss": 0.5541,
+      "step": 86650
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010591578947368421,
+      "loss": 0.5508,
+      "step": 86660
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010583684210526315,
+      "loss": 0.5664,
+      "step": 86670
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001057578947368421,
+      "loss": 0.5632,
+      "step": 86680
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010567894736842104,
+      "loss": 0.5748,
+      "step": 86690
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010559999999999998,
+      "loss": 0.5573,
+      "step": 86700
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010552105263157894,
+      "loss": 0.5541,
+      "step": 86710
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010544210526315789,
+      "loss": 0.5653,
+      "step": 86720
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010536315789473684,
+      "loss": 0.5746,
+      "step": 86730
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010528421052631577,
+      "loss": 0.5615,
+      "step": 86740
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010520526315789472,
+      "loss": 0.5608,
+      "step": 86750
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010512631578947368,
+      "loss": 0.5663,
+      "step": 86760
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010504736842105263,
+      "loss": 0.5796,
+      "step": 86770
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010496842105263156,
+      "loss": 0.5762,
+      "step": 86780
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010488947368421051,
+      "loss": 0.5784,
+      "step": 86790
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010481052631578947,
+      "loss": 0.5815,
+      "step": 86800
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010473157894736841,
+      "loss": 0.5618,
+      "step": 86810
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010465263157894736,
+      "loss": 0.563,
+      "step": 86820
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001045736842105263,
+      "loss": 0.5587,
+      "step": 86830
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010449473684210525,
+      "loss": 0.5658,
+      "step": 86840
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001044157894736842,
+      "loss": 0.5656,
+      "step": 86850
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010433684210526315,
+      "loss": 0.5544,
+      "step": 86860
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001042578947368421,
+      "loss": 0.5633,
+      "step": 86870
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010417894736842103,
+      "loss": 0.5526,
+      "step": 86880
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010409999999999998,
+      "loss": 0.5532,
+      "step": 86890
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010402105263157894,
+      "loss": 0.5569,
+      "step": 86900
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010394210526315789,
+      "loss": 0.5502,
+      "step": 86910
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010386315789473683,
+      "loss": 0.557,
+      "step": 86920
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010378421052631578,
+      "loss": 0.5648,
+      "step": 86930
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010370526315789474,
+      "loss": 0.5772,
+      "step": 86940
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010362631578947369,
+      "loss": 0.5599,
+      "step": 86950
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010354736842105262,
+      "loss": 0.5584,
+      "step": 86960
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010346842105263157,
+      "loss": 0.5611,
+      "step": 86970
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010338947368421052,
+      "loss": 0.5718,
+      "step": 86980
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010331052631578947,
+      "loss": 0.5668,
+      "step": 86990
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010323157894736841,
+      "loss": 0.5624,
+      "step": 87000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010315263157894736,
+      "loss": 0.5586,
+      "step": 87010
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001030736842105263,
+      "loss": 0.5577,
+      "step": 87020
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010299473684210525,
+      "loss": 0.56,
+      "step": 87030
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010291578947368421,
+      "loss": 0.5702,
+      "step": 87040
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010283684210526316,
+      "loss": 0.565,
+      "step": 87050
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010275789473684209,
+      "loss": 0.5633,
+      "step": 87060
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010267894736842104,
+      "loss": 0.5588,
+      "step": 87070
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001026,
+      "loss": 0.5518,
+      "step": 87080
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010252105263157895,
+      "loss": 0.5547,
+      "step": 87090
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010244210526315788,
+      "loss": 0.5612,
+      "step": 87100
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010236315789473683,
+      "loss": 0.5587,
+      "step": 87110
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010228421052631578,
+      "loss": 0.5513,
+      "step": 87120
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010220526315789473,
+      "loss": 0.5647,
+      "step": 87130
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010212631578947368,
+      "loss": 0.557,
+      "step": 87140
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010204736842105262,
+      "loss": 0.5613,
+      "step": 87150
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010196842105263156,
+      "loss": 0.5687,
+      "step": 87160
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010188947368421051,
+      "loss": 0.5627,
+      "step": 87170
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010181052631578947,
+      "loss": 0.5628,
+      "step": 87180
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010173157894736842,
+      "loss": 0.5652,
+      "step": 87190
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010165263157894735,
+      "loss": 0.5752,
+      "step": 87200
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001015736842105263,
+      "loss": 0.561,
+      "step": 87210
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010149473684210526,
+      "loss": 0.5793,
+      "step": 87220
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010141578947368421,
+      "loss": 0.5658,
+      "step": 87230
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010133684210526315,
+      "loss": 0.5796,
+      "step": 87240
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001012578947368421,
+      "loss": 0.5635,
+      "step": 87250
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010117894736842104,
+      "loss": 0.563,
+      "step": 87260
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001011,
+      "loss": 0.5685,
+      "step": 87270
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010102105263157894,
+      "loss": 0.5507,
+      "step": 87280
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010094210526315789,
+      "loss": 0.5601,
+      "step": 87290
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010086315789473684,
+      "loss": 0.553,
+      "step": 87300
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010078421052631577,
+      "loss": 0.5715,
+      "step": 87310
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010070526315789473,
+      "loss": 0.5684,
+      "step": 87320
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010062631578947368,
+      "loss": 0.5795,
+      "step": 87330
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010054736842105262,
+      "loss": 0.568,
+      "step": 87340
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010046842105263156,
+      "loss": 0.5803,
+      "step": 87350
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010038947368421053,
+      "loss": 0.5711,
+      "step": 87360
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010031052631578947,
+      "loss": 0.5625,
+      "step": 87370
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010023157894736841,
+      "loss": 0.5512,
+      "step": 87380
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010015263157894736,
+      "loss": 0.5513,
+      "step": 87390
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001000736842105263,
+      "loss": 0.5541,
+      "step": 87400
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.999473684210527e-05,
+      "loss": 0.5676,
+      "step": 87410
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.99157894736842e-05,
+      "loss": 0.5792,
+      "step": 87420
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.983684210526315e-05,
+      "loss": 0.5725,
+      "step": 87430
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.97578947368421e-05,
+      "loss": 0.5694,
+      "step": 87440
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.967894736842103e-05,
+      "loss": 0.5583,
+      "step": 87450
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.96e-05,
+      "loss": 0.5669,
+      "step": 87460
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.952105263157894e-05,
+      "loss": 0.571,
+      "step": 87470
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.944210526315788e-05,
+      "loss": 0.5622,
+      "step": 87480
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.936315789473683e-05,
+      "loss": 0.5697,
+      "step": 87490
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.928421052631579e-05,
+      "loss": 0.5629,
+      "step": 87500
+    },
+    {
+      "epoch": 0.88,
+      "eval_accuracy": 0.8814306303350578,
+      "eval_loss": 0.541015625,
+      "eval_runtime": 97.7663,
+      "eval_samples_per_second": 818.278,
+      "eval_steps_per_second": 1.606,
+      "step": 87500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.920526315789474e-05,
+      "loss": 0.5606,
+      "step": 87510
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.912631578947367e-05,
+      "loss": 0.5534,
+      "step": 87520
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.904736842105262e-05,
+      "loss": 0.5601,
+      "step": 87530
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.896842105263157e-05,
+      "loss": 0.5721,
+      "step": 87540
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.888947368421053e-05,
+      "loss": 0.5701,
+      "step": 87550
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.881842105263156e-05,
+      "loss": 0.5598,
+      "step": 87560
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.873947368421053e-05,
+      "loss": 0.57,
+      "step": 87570
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.866052631578946e-05,
+      "loss": 0.5681,
+      "step": 87580
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.858157894736841e-05,
+      "loss": 0.5678,
+      "step": 87590
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.850263157894736e-05,
+      "loss": 0.5571,
+      "step": 87600
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.84236842105263e-05,
+      "loss": 0.5652,
+      "step": 87610
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.834473684210525e-05,
+      "loss": 0.5726,
+      "step": 87620
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.82657894736842e-05,
+      "loss": 0.565,
+      "step": 87630
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.818684210526315e-05,
+      "loss": 0.5593,
+      "step": 87640
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.810789473684209e-05,
+      "loss": 0.5588,
+      "step": 87650
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.802894736842105e-05,
+      "loss": 0.5545,
+      "step": 87660
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.795e-05,
+      "loss": 0.5685,
+      "step": 87670
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.787105263157894e-05,
+      "loss": 0.5628,
+      "step": 87680
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.779210526315788e-05,
+      "loss": 0.5589,
+      "step": 87690
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.771315789473683e-05,
+      "loss": 0.5618,
+      "step": 87700
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.763421052631579e-05,
+      "loss": 0.5652,
+      "step": 87710
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.755526315789474e-05,
+      "loss": 0.5586,
+      "step": 87720
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.747631578947367e-05,
+      "loss": 0.5566,
+      "step": 87730
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.739736842105262e-05,
+      "loss": 0.5615,
+      "step": 87740
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.731842105263157e-05,
+      "loss": 0.5643,
+      "step": 87750
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.723947368421052e-05,
+      "loss": 0.5548,
+      "step": 87760
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.716052631578947e-05,
+      "loss": 0.5805,
+      "step": 87770
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.708157894736841e-05,
+      "loss": 0.5654,
+      "step": 87780
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.700263157894735e-05,
+      "loss": 0.573,
+      "step": 87790
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.692368421052631e-05,
+      "loss": 0.5512,
+      "step": 87800
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.684473684210526e-05,
+      "loss": 0.5689,
+      "step": 87810
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.67657894736842e-05,
+      "loss": 0.5723,
+      "step": 87820
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.668684210526314e-05,
+      "loss": 0.5695,
+      "step": 87830
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.660789473684209e-05,
+      "loss": 0.5572,
+      "step": 87840
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.652894736842105e-05,
+      "loss": 0.5727,
+      "step": 87850
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.645e-05,
+      "loss": 0.5558,
+      "step": 87860
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.637105263157893e-05,
+      "loss": 0.5543,
+      "step": 87870
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.629210526315788e-05,
+      "loss": 0.5576,
+      "step": 87880
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.621315789473684e-05,
+      "loss": 0.5601,
+      "step": 87890
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.613421052631578e-05,
+      "loss": 0.5651,
+      "step": 87900
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.605526315789473e-05,
+      "loss": 0.5664,
+      "step": 87910
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.597631578947368e-05,
+      "loss": 0.5646,
+      "step": 87920
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.589736842105261e-05,
+      "loss": 0.5531,
+      "step": 87930
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.581842105263157e-05,
+      "loss": 0.5607,
+      "step": 87940
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.573947368421052e-05,
+      "loss": 0.5465,
+      "step": 87950
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.566052631578947e-05,
+      "loss": 0.5573,
+      "step": 87960
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.55815789473684e-05,
+      "loss": 0.5427,
+      "step": 87970
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.550263157894735e-05,
+      "loss": 0.5453,
+      "step": 87980
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.542368421052631e-05,
+      "loss": 0.5498,
+      "step": 87990
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.534473684210526e-05,
+      "loss": 0.5599,
+      "step": 88000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.52657894736842e-05,
+      "loss": 0.5659,
+      "step": 88010
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.518684210526315e-05,
+      "loss": 0.5546,
+      "step": 88020
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.510789473684211e-05,
+      "loss": 0.5659,
+      "step": 88030
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.502894736842106e-05,
+      "loss": 0.5566,
+      "step": 88040
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.494999999999999e-05,
+      "loss": 0.5683,
+      "step": 88050
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.487105263157894e-05,
+      "loss": 0.5522,
+      "step": 88060
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.479210526315789e-05,
+      "loss": 0.5766,
+      "step": 88070
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.471315789473684e-05,
+      "loss": 0.5714,
+      "step": 88080
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.463421052631578e-05,
+      "loss": 0.5489,
+      "step": 88090
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.456315789473685e-05,
+      "loss": 0.5552,
+      "step": 88100
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.448421052631578e-05,
+      "loss": 0.5597,
+      "step": 88110
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.440526315789473e-05,
+      "loss": 0.5621,
+      "step": 88120
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.432631578947368e-05,
+      "loss": 0.5716,
+      "step": 88130
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.424736842105261e-05,
+      "loss": 0.5607,
+      "step": 88140
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.416842105263157e-05,
+      "loss": 0.5636,
+      "step": 88150
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.408947368421052e-05,
+      "loss": 0.5482,
+      "step": 88160
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.401052631578947e-05,
+      "loss": 0.5493,
+      "step": 88170
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.39315789473684e-05,
+      "loss": 0.5473,
+      "step": 88180
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.385263157894737e-05,
+      "loss": 0.5527,
+      "step": 88190
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.377368421052631e-05,
+      "loss": 0.5507,
+      "step": 88200
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.369473684210525e-05,
+      "loss": 0.5551,
+      "step": 88210
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.36157894736842e-05,
+      "loss": 0.5569,
+      "step": 88220
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.353684210526315e-05,
+      "loss": 0.5631,
+      "step": 88230
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.345789473684211e-05,
+      "loss": 0.5582,
+      "step": 88240
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.337894736842104e-05,
+      "loss": 0.5558,
+      "step": 88250
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.329999999999999e-05,
+      "loss": 0.5542,
+      "step": 88260
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.322105263157894e-05,
+      "loss": 0.5714,
+      "step": 88270
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.314210526315787e-05,
+      "loss": 0.5604,
+      "step": 88280
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.306315789473684e-05,
+      "loss": 0.5693,
+      "step": 88290
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.298421052631578e-05,
+      "loss": 0.5553,
+      "step": 88300
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.290526315789473e-05,
+      "loss": 0.5649,
+      "step": 88310
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.282631578947367e-05,
+      "loss": 0.5605,
+      "step": 88320
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.274736842105263e-05,
+      "loss": 0.5625,
+      "step": 88330
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.266842105263158e-05,
+      "loss": 0.5614,
+      "step": 88340
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.258947368421051e-05,
+      "loss": 0.5655,
+      "step": 88350
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.251052631578946e-05,
+      "loss": 0.5665,
+      "step": 88360
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.243157894736841e-05,
+      "loss": 0.5502,
+      "step": 88370
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.235263157894737e-05,
+      "loss": 0.5712,
+      "step": 88380
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.22736842105263e-05,
+      "loss": 0.5596,
+      "step": 88390
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.219473684210525e-05,
+      "loss": 0.565,
+      "step": 88400
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.21157894736842e-05,
+      "loss": 0.5525,
+      "step": 88410
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.203684210526314e-05,
+      "loss": 0.561,
+      "step": 88420
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.19578947368421e-05,
+      "loss": 0.5502,
+      "step": 88430
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.187894736842105e-05,
+      "loss": 0.5437,
+      "step": 88440
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.18e-05,
+      "loss": 0.5445,
+      "step": 88450
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.172105263157893e-05,
+      "loss": 0.563,
+      "step": 88460
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.164210526315789e-05,
+      "loss": 0.5559,
+      "step": 88470
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.156315789473684e-05,
+      "loss": 0.5428,
+      "step": 88480
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.148421052631579e-05,
+      "loss": 0.5585,
+      "step": 88490
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.140526315789472e-05,
+      "loss": 0.555,
+      "step": 88500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.132631578947367e-05,
+      "loss": 0.5511,
+      "step": 88510
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.124736842105263e-05,
+      "loss": 0.5623,
+      "step": 88520
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.116842105263157e-05,
+      "loss": 0.5565,
+      "step": 88530
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.108947368421052e-05,
+      "loss": 0.5556,
+      "step": 88540
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.101052631578946e-05,
+      "loss": 0.5597,
+      "step": 88550
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.09315789473684e-05,
+      "loss": 0.5525,
+      "step": 88560
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.085263157894736e-05,
+      "loss": 0.5513,
+      "step": 88570
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.077368421052631e-05,
+      "loss": 0.5645,
+      "step": 88580
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.069473684210526e-05,
+      "loss": 0.565,
+      "step": 88590
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.061578947368419e-05,
+      "loss": 0.5631,
+      "step": 88600
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.053684210526315e-05,
+      "loss": 0.5731,
+      "step": 88610
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.04578947368421e-05,
+      "loss": 0.5634,
+      "step": 88620
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.037894736842105e-05,
+      "loss": 0.5574,
+      "step": 88630
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.029999999999999e-05,
+      "loss": 0.5612,
+      "step": 88640
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.022105263157893e-05,
+      "loss": 0.5519,
+      "step": 88650
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.01421052631579e-05,
+      "loss": 0.5633,
+      "step": 88660
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.006315789473683e-05,
+      "loss": 0.5493,
+      "step": 88670
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.998421052631578e-05,
+      "loss": 0.5427,
+      "step": 88680
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.990526315789473e-05,
+      "loss": 0.5537,
+      "step": 88690
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.982631578947366e-05,
+      "loss": 0.5618,
+      "step": 88700
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.974736842105262e-05,
+      "loss": 0.5594,
+      "step": 88710
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.966842105263157e-05,
+      "loss": 0.5613,
+      "step": 88720
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.958947368421052e-05,
+      "loss": 0.5563,
+      "step": 88730
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.951052631578946e-05,
+      "loss": 0.558,
+      "step": 88740
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.943157894736842e-05,
+      "loss": 0.5611,
+      "step": 88750
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.935263157894737e-05,
+      "loss": 0.5707,
+      "step": 88760
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.927368421052631e-05,
+      "loss": 0.5711,
+      "step": 88770
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.919473684210525e-05,
+      "loss": 0.5677,
+      "step": 88780
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.91157894736842e-05,
+      "loss": 0.5558,
+      "step": 88790
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.903684210526316e-05,
+      "loss": 0.5675,
+      "step": 88800
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.895789473684211e-05,
+      "loss": 0.5635,
+      "step": 88810
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.887894736842104e-05,
+      "loss": 0.5631,
+      "step": 88820
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.879999999999999e-05,
+      "loss": 0.5617,
+      "step": 88830
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.872105263157894e-05,
+      "loss": 0.5618,
+      "step": 88840
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.864210526315789e-05,
+      "loss": 0.5515,
+      "step": 88850
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.856315789473683e-05,
+      "loss": 0.5603,
+      "step": 88860
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.848421052631578e-05,
+      "loss": 0.555,
+      "step": 88870
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.840526315789472e-05,
+      "loss": 0.5655,
+      "step": 88880
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.832631578947368e-05,
+      "loss": 0.544,
+      "step": 88890
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.824736842105263e-05,
+      "loss": 0.5488,
+      "step": 88900
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.816842105263158e-05,
+      "loss": 0.5457,
+      "step": 88910
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.808947368421051e-05,
+      "loss": 0.5529,
+      "step": 88920
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.801052631578946e-05,
+      "loss": 0.545,
+      "step": 88930
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.793157894736842e-05,
+      "loss": 0.5531,
+      "step": 88940
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.785263157894737e-05,
+      "loss": 0.5613,
+      "step": 88950
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.77736842105263e-05,
+      "loss": 0.5542,
+      "step": 88960
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.769473684210525e-05,
+      "loss": 0.5573,
+      "step": 88970
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.76157894736842e-05,
+      "loss": 0.5621,
+      "step": 88980
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.753684210526315e-05,
+      "loss": 0.5609,
+      "step": 88990
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.74578947368421e-05,
+      "loss": 0.5511,
+      "step": 89000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.737894736842105e-05,
+      "loss": 0.551,
+      "step": 89010
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.729999999999998e-05,
+      "loss": 0.5672,
+      "step": 89020
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.722105263157894e-05,
+      "loss": 0.5635,
+      "step": 89030
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.714210526315789e-05,
+      "loss": 0.5622,
+      "step": 89040
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.706315789473684e-05,
+      "loss": 0.5481,
+      "step": 89050
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.698421052631577e-05,
+      "loss": 0.5586,
+      "step": 89060
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.690526315789472e-05,
+      "loss": 0.5593,
+      "step": 89070
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.682631578947368e-05,
+      "loss": 0.5607,
+      "step": 89080
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.674736842105263e-05,
+      "loss": 0.5487,
+      "step": 89090
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.666842105263157e-05,
+      "loss": 0.5549,
+      "step": 89100
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.658947368421052e-05,
+      "loss": 0.5402,
+      "step": 89110
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.651052631578946e-05,
+      "loss": 0.5462,
+      "step": 89120
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.643157894736841e-05,
+      "loss": 0.5393,
+      "step": 89130
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.635263157894736e-05,
+      "loss": 0.5485,
+      "step": 89140
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.627368421052631e-05,
+      "loss": 0.5348,
+      "step": 89150
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.619473684210524e-05,
+      "loss": 0.5457,
+      "step": 89160
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.61157894736842e-05,
+      "loss": 0.5443,
+      "step": 89170
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.603684210526315e-05,
+      "loss": 0.5416,
+      "step": 89180
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.59578947368421e-05,
+      "loss": 0.5547,
+      "step": 89190
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.587894736842104e-05,
+      "loss": 0.5567,
+      "step": 89200
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.579999999999998e-05,
+      "loss": 0.5465,
+      "step": 89210
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.572105263157895e-05,
+      "loss": 0.5436,
+      "step": 89220
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.56421052631579e-05,
+      "loss": 0.5476,
+      "step": 89230
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.556315789473683e-05,
+      "loss": 0.5521,
+      "step": 89240
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.548421052631578e-05,
+      "loss": 0.5449,
+      "step": 89250
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.540526315789473e-05,
+      "loss": 0.5518,
+      "step": 89260
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.532631578947369e-05,
+      "loss": 0.5415,
+      "step": 89270
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.524736842105262e-05,
+      "loss": 0.5642,
+      "step": 89280
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.516842105263157e-05,
+      "loss": 0.5591,
+      "step": 89290
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.508947368421052e-05,
+      "loss": 0.5477,
+      "step": 89300
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.501052631578947e-05,
+      "loss": 0.5412,
+      "step": 89310
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.493157894736842e-05,
+      "loss": 0.557,
+      "step": 89320
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.485263157894736e-05,
+      "loss": 0.5437,
+      "step": 89330
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.47736842105263e-05,
+      "loss": 0.5496,
+      "step": 89340
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.469473684210525e-05,
+      "loss": 0.5582,
+      "step": 89350
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.461578947368421e-05,
+      "loss": 0.5559,
+      "step": 89360
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.453684210526316e-05,
+      "loss": 0.5559,
+      "step": 89370
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.445789473684209e-05,
+      "loss": 0.5458,
+      "step": 89380
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.437894736842104e-05,
+      "loss": 0.5466,
+      "step": 89390
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.43e-05,
+      "loss": 0.5434,
+      "step": 89400
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.422105263157895e-05,
+      "loss": 0.5571,
+      "step": 89410
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.414210526315789e-05,
+      "loss": 0.5477,
+      "step": 89420
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.406315789473683e-05,
+      "loss": 0.5559,
+      "step": 89430
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.398421052631578e-05,
+      "loss": 0.5469,
+      "step": 89440
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.390526315789473e-05,
+      "loss": 0.5594,
+      "step": 89450
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.382631578947368e-05,
+      "loss": 0.5549,
+      "step": 89460
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.374736842105263e-05,
+      "loss": 0.5616,
+      "step": 89470
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.366842105263156e-05,
+      "loss": 0.5534,
+      "step": 89480
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.358947368421051e-05,
+      "loss": 0.5546,
+      "step": 89490
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.351052631578947e-05,
+      "loss": 0.5519,
+      "step": 89500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.343157894736842e-05,
+      "loss": 0.5491,
+      "step": 89510
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.335263157894736e-05,
+      "loss": 0.5394,
+      "step": 89520
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.32736842105263e-05,
+      "loss": 0.5497,
+      "step": 89530
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.319473684210527e-05,
+      "loss": 0.56,
+      "step": 89540
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.311578947368421e-05,
+      "loss": 0.5481,
+      "step": 89550
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.303684210526315e-05,
+      "loss": 0.5543,
+      "step": 89560
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.29578947368421e-05,
+      "loss": 0.5625,
+      "step": 89570
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.287894736842104e-05,
+      "loss": 0.5411,
+      "step": 89580
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.28e-05,
+      "loss": 0.5489,
+      "step": 89590
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.272105263157894e-05,
+      "loss": 0.542,
+      "step": 89600
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.264210526315789e-05,
+      "loss": 0.5521,
+      "step": 89610
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.256315789473682e-05,
+      "loss": 0.5438,
+      "step": 89620
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.248421052631577e-05,
+      "loss": 0.5569,
+      "step": 89630
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.240526315789473e-05,
+      "loss": 0.5563,
+      "step": 89640
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.232631578947368e-05,
+      "loss": 0.5735,
+      "step": 89650
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.224736842105262e-05,
+      "loss": 0.5714,
+      "step": 89660
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.216842105263157e-05,
+      "loss": 0.5744,
+      "step": 89670
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.208947368421053e-05,
+      "loss": 0.5674,
+      "step": 89680
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.201052631578948e-05,
+      "loss": 0.5549,
+      "step": 89690
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.193157894736841e-05,
+      "loss": 0.5675,
+      "step": 89700
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.185263157894736e-05,
+      "loss": 0.5662,
+      "step": 89710
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.177368421052631e-05,
+      "loss": 0.5685,
+      "step": 89720
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.169473684210527e-05,
+      "loss": 0.5617,
+      "step": 89730
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.16157894736842e-05,
+      "loss": 0.5657,
+      "step": 89740
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.153684210526315e-05,
+      "loss": 0.5598,
+      "step": 89750
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.14578947368421e-05,
+      "loss": 0.5597,
+      "step": 89760
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.137894736842104e-05,
+      "loss": 0.5574,
+      "step": 89770
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.13e-05,
+      "loss": 0.5665,
+      "step": 89780
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.122105263157895e-05,
+      "loss": 0.568,
+      "step": 89790
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.114210526315788e-05,
+      "loss": 0.5616,
+      "step": 89800
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.106315789473683e-05,
+      "loss": 0.5512,
+      "step": 89810
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.098421052631579e-05,
+      "loss": 0.5602,
+      "step": 89820
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.090526315789474e-05,
+      "loss": 0.5565,
+      "step": 89830
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.082631578947367e-05,
+      "loss": 0.5572,
+      "step": 89840
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.074736842105262e-05,
+      "loss": 0.5627,
+      "step": 89850
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.066842105263157e-05,
+      "loss": 0.5689,
+      "step": 89860
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.058947368421053e-05,
+      "loss": 0.5642,
+      "step": 89870
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.051052631578947e-05,
+      "loss": 0.5434,
+      "step": 89880
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.043157894736842e-05,
+      "loss": 0.5498,
+      "step": 89890
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.035263157894736e-05,
+      "loss": 0.5351,
+      "step": 89900
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.02736842105263e-05,
+      "loss": 0.5417,
+      "step": 89910
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.019473684210526e-05,
+      "loss": 0.5542,
+      "step": 89920
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.011578947368421e-05,
+      "loss": 0.5592,
+      "step": 89930
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.003684210526314e-05,
+      "loss": 0.5733,
+      "step": 89940
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.995789473684209e-05,
+      "loss": 0.571,
+      "step": 89950
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.987894736842105e-05,
+      "loss": 0.5549,
+      "step": 89960
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.98e-05,
+      "loss": 0.5582,
+      "step": 89970
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.972105263157894e-05,
+      "loss": 0.564,
+      "step": 89980
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.964210526315788e-05,
+      "loss": 0.565,
+      "step": 89990
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.956315789473683e-05,
+      "loss": 0.5724,
+      "step": 90000
+    },
+    {
+      "epoch": 0.9,
+      "eval_accuracy": 0.8820988856910535,
+      "eval_loss": 0.537109375,
+      "eval_runtime": 97.2134,
+      "eval_samples_per_second": 822.932,
+      "eval_steps_per_second": 1.615,
+      "step": 90000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.94842105263158e-05,
+      "loss": 0.5633,
+      "step": 90010
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.940526315789473e-05,
+      "loss": 0.555,
+      "step": 90020
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.932631578947368e-05,
+      "loss": 0.5656,
+      "step": 90030
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.924736842105263e-05,
+      "loss": 0.5636,
+      "step": 90040
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.916842105263156e-05,
+      "loss": 0.5523,
+      "step": 90050
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.908947368421052e-05,
+      "loss": 0.5567,
+      "step": 90060
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.901052631578947e-05,
+      "loss": 0.5561,
+      "step": 90070
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.89315789473684e-05,
+      "loss": 0.5568,
+      "step": 90080
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.885263157894735e-05,
+      "loss": 0.5597,
+      "step": 90090
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.878157894736842e-05,
+      "loss": 0.5579,
+      "step": 90100
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.870263157894735e-05,
+      "loss": 0.5622,
+      "step": 90110
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.862368421052631e-05,
+      "loss": 0.5563,
+      "step": 90120
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.854473684210526e-05,
+      "loss": 0.5649,
+      "step": 90130
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.846578947368421e-05,
+      "loss": 0.5568,
+      "step": 90140
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.838684210526314e-05,
+      "loss": 0.5674,
+      "step": 90150
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.830789473684209e-05,
+      "loss": 0.5641,
+      "step": 90160
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.822894736842105e-05,
+      "loss": 0.5598,
+      "step": 90170
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.815e-05,
+      "loss": 0.567,
+      "step": 90180
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.807105263157894e-05,
+      "loss": 0.5678,
+      "step": 90190
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.799210526315789e-05,
+      "loss": 0.5584,
+      "step": 90200
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.791315789473683e-05,
+      "loss": 0.566,
+      "step": 90210
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.783421052631578e-05,
+      "loss": 0.5538,
+      "step": 90220
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.775526315789473e-05,
+      "loss": 0.5639,
+      "step": 90230
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.767631578947368e-05,
+      "loss": 0.5665,
+      "step": 90240
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.759736842105261e-05,
+      "loss": 0.5691,
+      "step": 90250
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.751842105263158e-05,
+      "loss": 0.5702,
+      "step": 90260
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.743947368421052e-05,
+      "loss": 0.5608,
+      "step": 90270
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.736052631578947e-05,
+      "loss": 0.5412,
+      "step": 90280
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.72815789473684e-05,
+      "loss": 0.554,
+      "step": 90290
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.720263157894735e-05,
+      "loss": 0.5615,
+      "step": 90300
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.712368421052632e-05,
+      "loss": 0.5645,
+      "step": 90310
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.704473684210526e-05,
+      "loss": 0.5545,
+      "step": 90320
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.69657894736842e-05,
+      "loss": 0.5475,
+      "step": 90330
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.688684210526315e-05,
+      "loss": 0.5529,
+      "step": 90340
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.68078947368421e-05,
+      "loss": 0.5619,
+      "step": 90350
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.672894736842104e-05,
+      "loss": 0.5584,
+      "step": 90360
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.664999999999999e-05,
+      "loss": 0.5619,
+      "step": 90370
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.657105263157894e-05,
+      "loss": 0.5522,
+      "step": 90380
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.649210526315788e-05,
+      "loss": 0.5677,
+      "step": 90390
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.641315789473684e-05,
+      "loss": 0.5508,
+      "step": 90400
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.633421052631579e-05,
+      "loss": 0.5749,
+      "step": 90410
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.625526315789473e-05,
+      "loss": 0.5545,
+      "step": 90420
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.617631578947367e-05,
+      "loss": 0.5567,
+      "step": 90430
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.609736842105262e-05,
+      "loss": 0.5543,
+      "step": 90440
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.601842105263158e-05,
+      "loss": 0.5609,
+      "step": 90450
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.593947368421053e-05,
+      "loss": 0.5603,
+      "step": 90460
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.586052631578946e-05,
+      "loss": 0.5612,
+      "step": 90470
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.578157894736841e-05,
+      "loss": 0.5648,
+      "step": 90480
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.570263157894736e-05,
+      "loss": 0.5653,
+      "step": 90490
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.562368421052632e-05,
+      "loss": 0.5616,
+      "step": 90500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.554473684210526e-05,
+      "loss": 0.5541,
+      "step": 90510
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.54657894736842e-05,
+      "loss": 0.563,
+      "step": 90520
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.538684210526315e-05,
+      "loss": 0.5585,
+      "step": 90530
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.53078947368421e-05,
+      "loss": 0.544,
+      "step": 90540
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.522894736842105e-05,
+      "loss": 0.5487,
+      "step": 90550
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.515e-05,
+      "loss": 0.5548,
+      "step": 90560
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.507105263157893e-05,
+      "loss": 0.5547,
+      "step": 90570
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.49921052631579e-05,
+      "loss": 0.5555,
+      "step": 90580
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.491315789473683e-05,
+      "loss": 0.5652,
+      "step": 90590
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.483421052631579e-05,
+      "loss": 0.5609,
+      "step": 90600
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.475526315789473e-05,
+      "loss": 0.5549,
+      "step": 90610
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.467631578947369e-05,
+      "loss": 0.5575,
+      "step": 90620
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.459736842105262e-05,
+      "loss": 0.5562,
+      "step": 90630
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.451842105263157e-05,
+      "loss": 0.5562,
+      "step": 90640
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.443947368421052e-05,
+      "loss": 0.5541,
+      "step": 90650
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.436052631578947e-05,
+      "loss": 0.5454,
+      "step": 90660
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.428157894736841e-05,
+      "loss": 0.5686,
+      "step": 90670
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.420263157894736e-05,
+      "loss": 0.5682,
+      "step": 90680
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.412368421052631e-05,
+      "loss": 0.5579,
+      "step": 90690
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.404473684210526e-05,
+      "loss": 0.562,
+      "step": 90700
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.39657894736842e-05,
+      "loss": 0.5628,
+      "step": 90710
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.388684210526316e-05,
+      "loss": 0.5559,
+      "step": 90720
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.380789473684209e-05,
+      "loss": 0.5576,
+      "step": 90730
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.372894736842105e-05,
+      "loss": 0.5532,
+      "step": 90740
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.364999999999999e-05,
+      "loss": 0.5491,
+      "step": 90750
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.357105263157895e-05,
+      "loss": 0.5621,
+      "step": 90760
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.349210526315788e-05,
+      "loss": 0.5491,
+      "step": 90770
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.341315789473683e-05,
+      "loss": 0.5474,
+      "step": 90780
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.333421052631578e-05,
+      "loss": 0.561,
+      "step": 90790
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.325526315789473e-05,
+      "loss": 0.5569,
+      "step": 90800
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.317631578947368e-05,
+      "loss": 0.5555,
+      "step": 90810
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.309736842105263e-05,
+      "loss": 0.5637,
+      "step": 90820
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.301842105263157e-05,
+      "loss": 0.5755,
+      "step": 90830
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.293947368421052e-05,
+      "loss": 0.5626,
+      "step": 90840
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.286052631578946e-05,
+      "loss": 0.5753,
+      "step": 90850
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.278157894736842e-05,
+      "loss": 0.5672,
+      "step": 90860
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.270263157894735e-05,
+      "loss": 0.5644,
+      "step": 90870
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.262368421052632e-05,
+      "loss": 0.5543,
+      "step": 90880
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.254473684210525e-05,
+      "loss": 0.5625,
+      "step": 90890
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.246578947368421e-05,
+      "loss": 0.5621,
+      "step": 90900
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.238684210526315e-05,
+      "loss": 0.565,
+      "step": 90910
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.23078947368421e-05,
+      "loss": 0.5587,
+      "step": 90920
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.222894736842104e-05,
+      "loss": 0.5577,
+      "step": 90930
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.214999999999999e-05,
+      "loss": 0.5575,
+      "step": 90940
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.207105263157894e-05,
+      "loss": 0.5588,
+      "step": 90950
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.199210526315789e-05,
+      "loss": 0.5574,
+      "step": 90960
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.191315789473684e-05,
+      "loss": 0.5623,
+      "step": 90970
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.183421052631579e-05,
+      "loss": 0.5494,
+      "step": 90980
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.175526315789473e-05,
+      "loss": 0.5557,
+      "step": 90990
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.167631578947368e-05,
+      "loss": 0.5554,
+      "step": 91000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.159736842105262e-05,
+      "loss": 0.5635,
+      "step": 91010
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.151842105263158e-05,
+      "loss": 0.5585,
+      "step": 91020
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.143947368421051e-05,
+      "loss": 0.5526,
+      "step": 91030
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.136052631578947e-05,
+      "loss": 0.5539,
+      "step": 91040
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.128157894736841e-05,
+      "loss": 0.5575,
+      "step": 91050
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.120263157894736e-05,
+      "loss": 0.5483,
+      "step": 91060
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.11236842105263e-05,
+      "loss": 0.5647,
+      "step": 91070
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.104473684210525e-05,
+      "loss": 0.5564,
+      "step": 91080
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.09657894736842e-05,
+      "loss": 0.5727,
+      "step": 91090
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.089473684210525e-05,
+      "loss": 0.5502,
+      "step": 91100
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.081578947368421e-05,
+      "loss": 0.5648,
+      "step": 91110
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.073684210526315e-05,
+      "loss": 0.5551,
+      "step": 91120
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.06578947368421e-05,
+      "loss": 0.5649,
+      "step": 91130
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.057894736842104e-05,
+      "loss": 0.5595,
+      "step": 91140
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.049999999999999e-05,
+      "loss": 0.5592,
+      "step": 91150
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.042105263157894e-05,
+      "loss": 0.5569,
+      "step": 91160
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.034210526315789e-05,
+      "loss": 0.5577,
+      "step": 91170
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.026315789473684e-05,
+      "loss": 0.567,
+      "step": 91180
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.018421052631579e-05,
+      "loss": 0.5574,
+      "step": 91190
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.010526315789473e-05,
+      "loss": 0.5472,
+      "step": 91200
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.002631578947368e-05,
+      "loss": 0.5459,
+      "step": 91210
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.994736842105262e-05,
+      "loss": 0.5495,
+      "step": 91220
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.986842105263158e-05,
+      "loss": 0.5771,
+      "step": 91230
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.978947368421051e-05,
+      "loss": 0.5726,
+      "step": 91240
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.971052631578948e-05,
+      "loss": 0.5685,
+      "step": 91250
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.963157894736841e-05,
+      "loss": 0.568,
+      "step": 91260
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.955263157894737e-05,
+      "loss": 0.5633,
+      "step": 91270
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.947368421052631e-05,
+      "loss": 0.544,
+      "step": 91280
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.939473684210526e-05,
+      "loss": 0.5422,
+      "step": 91290
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.93157894736842e-05,
+      "loss": 0.5397,
+      "step": 91300
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.923684210526315e-05,
+      "loss": 0.565,
+      "step": 91310
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.91578947368421e-05,
+      "loss": 0.5302,
+      "step": 91320
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.907894736842105e-05,
+      "loss": 0.5405,
+      "step": 91330
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.9e-05,
+      "loss": 0.5276,
+      "step": 91340
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.892105263157894e-05,
+      "loss": 0.5401,
+      "step": 91350
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.884210526315788e-05,
+      "loss": 0.5361,
+      "step": 91360
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.876315789473684e-05,
+      "loss": 0.5328,
+      "step": 91370
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.868421052631578e-05,
+      "loss": 0.5281,
+      "step": 91380
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.860526315789474e-05,
+      "loss": 0.5363,
+      "step": 91390
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.852631578947367e-05,
+      "loss": 0.5341,
+      "step": 91400
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.844736842105263e-05,
+      "loss": 0.545,
+      "step": 91410
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.836842105263157e-05,
+      "loss": 0.5369,
+      "step": 91420
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.828947368421052e-05,
+      "loss": 0.5564,
+      "step": 91430
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.821052631578947e-05,
+      "loss": 0.5442,
+      "step": 91440
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.813157894736841e-05,
+      "loss": 0.5697,
+      "step": 91450
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.805263157894736e-05,
+      "loss": 0.5594,
+      "step": 91460
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.797368421052631e-05,
+      "loss": 0.5676,
+      "step": 91470
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.789473684210526e-05,
+      "loss": 0.561,
+      "step": 91480
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.781578947368421e-05,
+      "loss": 0.5583,
+      "step": 91490
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.773684210526316e-05,
+      "loss": 0.5535,
+      "step": 91500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.76578947368421e-05,
+      "loss": 0.5649,
+      "step": 91510
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.757894736842104e-05,
+      "loss": 0.559,
+      "step": 91520
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.75e-05,
+      "loss": 0.5572,
+      "step": 91530
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.742105263157894e-05,
+      "loss": 0.5607,
+      "step": 91540
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.73421052631579e-05,
+      "loss": 0.5493,
+      "step": 91550
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.726315789473683e-05,
+      "loss": 0.5643,
+      "step": 91560
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.71842105263158e-05,
+      "loss": 0.5615,
+      "step": 91570
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.710526315789473e-05,
+      "loss": 0.5482,
+      "step": 91580
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.702631578947368e-05,
+      "loss": 0.5503,
+      "step": 91590
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.694736842105263e-05,
+      "loss": 0.5373,
+      "step": 91600
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.686842105263157e-05,
+      "loss": 0.538,
+      "step": 91610
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.678947368421052e-05,
+      "loss": 0.541,
+      "step": 91620
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.671052631578947e-05,
+      "loss": 0.5373,
+      "step": 91630
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.663157894736842e-05,
+      "loss": 0.5468,
+      "step": 91640
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.655263157894737e-05,
+      "loss": 0.5584,
+      "step": 91650
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.64736842105263e-05,
+      "loss": 0.5783,
+      "step": 91660
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.639473684210526e-05,
+      "loss": 0.5676,
+      "step": 91670
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.63157894736842e-05,
+      "loss": 0.5649,
+      "step": 91680
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.623684210526316e-05,
+      "loss": 0.5692,
+      "step": 91690
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.61578947368421e-05,
+      "loss": 0.573,
+      "step": 91700
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.607894736842106e-05,
+      "loss": 0.5864,
+      "step": 91710
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.599999999999999e-05,
+      "loss": 0.5886,
+      "step": 91720
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.592105263157894e-05,
+      "loss": 0.5783,
+      "step": 91730
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.584210526315789e-05,
+      "loss": 0.5594,
+      "step": 91740
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.576315789473684e-05,
+      "loss": 0.5523,
+      "step": 91750
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.568421052631578e-05,
+      "loss": 0.5319,
+      "step": 91760
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.560526315789473e-05,
+      "loss": 0.5527,
+      "step": 91770
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.552631578947368e-05,
+      "loss": 0.545,
+      "step": 91780
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.544736842105263e-05,
+      "loss": 0.5667,
+      "step": 91790
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.536842105263156e-05,
+      "loss": 0.5477,
+      "step": 91800
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.528947368421053e-05,
+      "loss": 0.5548,
+      "step": 91810
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.521052631578946e-05,
+      "loss": 0.5453,
+      "step": 91820
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.513157894736842e-05,
+      "loss": 0.5416,
+      "step": 91830
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.505263157894736e-05,
+      "loss": 0.5347,
+      "step": 91840
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.497368421052632e-05,
+      "loss": 0.5378,
+      "step": 91850
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.489473684210525e-05,
+      "loss": 0.5349,
+      "step": 91860
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.48157894736842e-05,
+      "loss": 0.5472,
+      "step": 91870
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.473684210526315e-05,
+      "loss": 0.5486,
+      "step": 91880
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.46578947368421e-05,
+      "loss": 0.5664,
+      "step": 91890
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.457894736842105e-05,
+      "loss": 0.5623,
+      "step": 91900
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.45e-05,
+      "loss": 0.5685,
+      "step": 91910
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.442105263157894e-05,
+      "loss": 0.5764,
+      "step": 91920
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.434210526315789e-05,
+      "loss": 0.5662,
+      "step": 91930
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.426315789473683e-05,
+      "loss": 0.5745,
+      "step": 91940
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.418421052631579e-05,
+      "loss": 0.5882,
+      "step": 91950
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.410526315789472e-05,
+      "loss": 0.5799,
+      "step": 91960
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.402631578947369e-05,
+      "loss": 0.5643,
+      "step": 91970
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.394736842105262e-05,
+      "loss": 0.5662,
+      "step": 91980
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.386842105263158e-05,
+      "loss": 0.5588,
+      "step": 91990
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.378947368421052e-05,
+      "loss": 0.5582,
+      "step": 92000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.371052631578947e-05,
+      "loss": 0.5585,
+      "step": 92010
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.363157894736841e-05,
+      "loss": 0.5563,
+      "step": 92020
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.355263157894736e-05,
+      "loss": 0.5533,
+      "step": 92030
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.347368421052631e-05,
+      "loss": 0.559,
+      "step": 92040
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.339473684210526e-05,
+      "loss": 0.5579,
+      "step": 92050
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.33157894736842e-05,
+      "loss": 0.5512,
+      "step": 92060
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.323684210526315e-05,
+      "loss": 0.5438,
+      "step": 92070
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.315789473684209e-05,
+      "loss": 0.539,
+      "step": 92080
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.307894736842105e-05,
+      "loss": 0.548,
+      "step": 92090
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.30078947368421e-05,
+      "loss": 0.5482,
+      "step": 92100
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.292894736842105e-05,
+      "loss": 0.5486,
+      "step": 92110
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.285e-05,
+      "loss": 0.5571,
+      "step": 92120
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.277105263157894e-05,
+      "loss": 0.5558,
+      "step": 92130
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.269210526315789e-05,
+      "loss": 0.5622,
+      "step": 92140
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.261315789473684e-05,
+      "loss": 0.5607,
+      "step": 92150
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.253421052631579e-05,
+      "loss": 0.5745,
+      "step": 92160
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.245526315789472e-05,
+      "loss": 0.5646,
+      "step": 92170
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.237631578947369e-05,
+      "loss": 0.5684,
+      "step": 92180
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.229736842105262e-05,
+      "loss": 0.5744,
+      "step": 92190
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.221842105263157e-05,
+      "loss": 0.5742,
+      "step": 92200
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.213947368421052e-05,
+      "loss": 0.5668,
+      "step": 92210
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.206052631578947e-05,
+      "loss": 0.5487,
+      "step": 92220
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.198157894736841e-05,
+      "loss": 0.5567,
+      "step": 92230
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.190263157894736e-05,
+      "loss": 0.5486,
+      "step": 92240
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.182368421052631e-05,
+      "loss": 0.5625,
+      "step": 92250
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.174473684210526e-05,
+      "loss": 0.5562,
+      "step": 92260
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.166578947368421e-05,
+      "loss": 0.549,
+      "step": 92270
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.158684210526316e-05,
+      "loss": 0.5432,
+      "step": 92280
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.15078947368421e-05,
+      "loss": 0.5462,
+      "step": 92290
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.142894736842105e-05,
+      "loss": 0.5472,
+      "step": 92300
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.134999999999999e-05,
+      "loss": 0.5388,
+      "step": 92310
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.127105263157895e-05,
+      "loss": 0.535,
+      "step": 92320
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.119210526315788e-05,
+      "loss": 0.5383,
+      "step": 92330
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.111315789473685e-05,
+      "loss": 0.5431,
+      "step": 92340
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.103421052631578e-05,
+      "loss": 0.5502,
+      "step": 92350
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.0955263157894735e-05,
+      "loss": 0.5704,
+      "step": 92360
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.087631578947368e-05,
+      "loss": 0.5768,
+      "step": 92370
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.0797368421052625e-05,
+      "loss": 0.576,
+      "step": 92380
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.071842105263157e-05,
+      "loss": 0.581,
+      "step": 92390
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.063947368421052e-05,
+      "loss": 0.5734,
+      "step": 92400
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.056052631578947e-05,
+      "loss": 0.5678,
+      "step": 92410
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.048157894736841e-05,
+      "loss": 0.5702,
+      "step": 92420
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.0402631578947367e-05,
+      "loss": 0.5698,
+      "step": 92430
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.032368421052631e-05,
+      "loss": 0.579,
+      "step": 92440
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.0244736842105256e-05,
+      "loss": 0.5601,
+      "step": 92450
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.0165789473684205e-05,
+      "loss": 0.5581,
+      "step": 92460
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.008684210526315e-05,
+      "loss": 0.562,
+      "step": 92470
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.00078947368421e-05,
+      "loss": 0.5585,
+      "step": 92480
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.992894736842104e-05,
+      "loss": 0.5743,
+      "step": 92490
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.985e-05,
+      "loss": 0.5482,
+      "step": 92500
+    },
+    {
+      "epoch": 0.93,
+      "eval_accuracy": 0.8834968490752123,
+      "eval_loss": 0.52978515625,
+      "eval_runtime": 97.1569,
+      "eval_samples_per_second": 823.411,
+      "eval_steps_per_second": 1.616,
+      "step": 92500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.977105263157894e-05,
+      "loss": 0.5555,
+      "step": 92510
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.969210526315789e-05,
+      "loss": 0.5454,
+      "step": 92520
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.9613157894736836e-05,
+      "loss": 0.5387,
+      "step": 92530
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.9534210526315784e-05,
+      "loss": 0.5603,
+      "step": 92540
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.945526315789473e-05,
+      "loss": 0.5328,
+      "step": 92550
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.9376315789473674e-05,
+      "loss": 0.5441,
+      "step": 92560
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.929736842105263e-05,
+      "loss": 0.557,
+      "step": 92570
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.921842105263157e-05,
+      "loss": 0.5579,
+      "step": 92580
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.913947368421052e-05,
+      "loss": 0.5615,
+      "step": 92590
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.906052631578947e-05,
+      "loss": 0.5585,
+      "step": 92600
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.8981578947368416e-05,
+      "loss": 0.5563,
+      "step": 92610
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.8902631578947364e-05,
+      "loss": 0.5681,
+      "step": 92620
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.882368421052631e-05,
+      "loss": 0.5568,
+      "step": 92630
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.874473684210526e-05,
+      "loss": 0.5636,
+      "step": 92640
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.86657894736842e-05,
+      "loss": 0.552,
+      "step": 92650
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.858684210526315e-05,
+      "loss": 0.5579,
+      "step": 92660
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.85078947368421e-05,
+      "loss": 0.5406,
+      "step": 92670
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.842894736842105e-05,
+      "loss": 0.5514,
+      "step": 92680
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.8349999999999995e-05,
+      "loss": 0.5416,
+      "step": 92690
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.8271052631578944e-05,
+      "loss": 0.5424,
+      "step": 92700
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.819210526315789e-05,
+      "loss": 0.5331,
+      "step": 92710
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.8113157894736833e-05,
+      "loss": 0.545,
+      "step": 92720
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.803421052631578e-05,
+      "loss": 0.5257,
+      "step": 92730
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.795526315789473e-05,
+      "loss": 0.5363,
+      "step": 92740
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.787631578947368e-05,
+      "loss": 0.5345,
+      "step": 92750
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.779736842105263e-05,
+      "loss": 0.5373,
+      "step": 92760
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.7718421052631575e-05,
+      "loss": 0.5339,
+      "step": 92770
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.763947368421052e-05,
+      "loss": 0.5269,
+      "step": 92780
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.7560526315789465e-05,
+      "loss": 0.5483,
+      "step": 92790
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.748157894736842e-05,
+      "loss": 0.5536,
+      "step": 92800
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.740263157894736e-05,
+      "loss": 0.5591,
+      "step": 92810
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.732368421052631e-05,
+      "loss": 0.5613,
+      "step": 92820
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.724473684210526e-05,
+      "loss": 0.562,
+      "step": 92830
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.7165789473684206e-05,
+      "loss": 0.5623,
+      "step": 92840
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.7086842105263155e-05,
+      "loss": 0.5591,
+      "step": 92850
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.70078947368421e-05,
+      "loss": 0.5741,
+      "step": 92860
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.692894736842105e-05,
+      "loss": 0.5545,
+      "step": 92870
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.684999999999999e-05,
+      "loss": 0.5488,
+      "step": 92880
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.677105263157894e-05,
+      "loss": 0.5598,
+      "step": 92890
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.669210526315789e-05,
+      "loss": 0.5502,
+      "step": 92900
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.661315789473684e-05,
+      "loss": 0.5513,
+      "step": 92910
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.6534210526315786e-05,
+      "loss": 0.5383,
+      "step": 92920
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.6455263157894734e-05,
+      "loss": 0.5458,
+      "step": 92930
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.637631578947368e-05,
+      "loss": 0.536,
+      "step": 92940
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.6297368421052624e-05,
+      "loss": 0.5503,
+      "step": 92950
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.621842105263157e-05,
+      "loss": 0.5452,
+      "step": 92960
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.613947368421052e-05,
+      "loss": 0.5285,
+      "step": 92970
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.606052631578947e-05,
+      "loss": 0.5303,
+      "step": 92980
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.598157894736842e-05,
+      "loss": 0.5272,
+      "step": 92990
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.5902631578947366e-05,
+      "loss": 0.5252,
+      "step": 93000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.5823684210526314e-05,
+      "loss": 0.5261,
+      "step": 93010
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.5744736842105255e-05,
+      "loss": 0.5248,
+      "step": 93020
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.5665789473684204e-05,
+      "loss": 0.53,
+      "step": 93030
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.558684210526315e-05,
+      "loss": 0.5336,
+      "step": 93040
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.55078947368421e-05,
+      "loss": 0.5329,
+      "step": 93050
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.542894736842105e-05,
+      "loss": 0.5422,
+      "step": 93060
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.535e-05,
+      "loss": 0.5743,
+      "step": 93070
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.5271052631578945e-05,
+      "loss": 0.5587,
+      "step": 93080
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.5192105263157894e-05,
+      "loss": 0.5608,
+      "step": 93090
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.5113157894736835e-05,
+      "loss": 0.5682,
+      "step": 93100
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.504210526315789e-05,
+      "loss": 0.5697,
+      "step": 93110
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.496315789473683e-05,
+      "loss": 0.567,
+      "step": 93120
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.4884210526315786e-05,
+      "loss": 0.5714,
+      "step": 93130
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.480526315789473e-05,
+      "loss": 0.5616,
+      "step": 93140
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.472631578947368e-05,
+      "loss": 0.5684,
+      "step": 93150
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.4647368421052625e-05,
+      "loss": 0.559,
+      "step": 93160
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.456842105263158e-05,
+      "loss": 0.5566,
+      "step": 93170
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.448947368421052e-05,
+      "loss": 0.5645,
+      "step": 93180
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.441052631578946e-05,
+      "loss": 0.5635,
+      "step": 93190
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.433157894736842e-05,
+      "loss": 0.5472,
+      "step": 93200
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.425263157894736e-05,
+      "loss": 0.5437,
+      "step": 93210
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.4173684210526314e-05,
+      "loss": 0.5467,
+      "step": 93220
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.4094736842105256e-05,
+      "loss": 0.5473,
+      "step": 93230
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.401578947368421e-05,
+      "loss": 0.5332,
+      "step": 93240
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.393684210526315e-05,
+      "loss": 0.5336,
+      "step": 93250
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.3857894736842094e-05,
+      "loss": 0.5374,
+      "step": 93260
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.377894736842105e-05,
+      "loss": 0.5484,
+      "step": 93270
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.369999999999999e-05,
+      "loss": 0.5483,
+      "step": 93280
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.3621052631578946e-05,
+      "loss": 0.5549,
+      "step": 93290
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.354210526315789e-05,
+      "loss": 0.5658,
+      "step": 93300
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.346315789473684e-05,
+      "loss": 0.5748,
+      "step": 93310
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.3384210526315784e-05,
+      "loss": 0.5751,
+      "step": 93320
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.3305263157894725e-05,
+      "loss": 0.5793,
+      "step": 93330
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.322631578947368e-05,
+      "loss": 0.5721,
+      "step": 93340
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.314736842105262e-05,
+      "loss": 0.576,
+      "step": 93350
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.306842105263158e-05,
+      "loss": 0.5655,
+      "step": 93360
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.298947368421052e-05,
+      "loss": 0.5695,
+      "step": 93370
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.2910526315789474e-05,
+      "loss": 0.5617,
+      "step": 93380
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.2831578947368415e-05,
+      "loss": 0.5572,
+      "step": 93390
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.2752631578947364e-05,
+      "loss": 0.5574,
+      "step": 93400
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.267368421052631e-05,
+      "loss": 0.5526,
+      "step": 93410
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.259473684210525e-05,
+      "loss": 0.5534,
+      "step": 93420
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.251578947368421e-05,
+      "loss": 0.551,
+      "step": 93430
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.243684210526315e-05,
+      "loss": 0.5488,
+      "step": 93440
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.2357894736842105e-05,
+      "loss": 0.5515,
+      "step": 93450
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.2278947368421047e-05,
+      "loss": 0.5414,
+      "step": 93460
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.2199999999999995e-05,
+      "loss": 0.5343,
+      "step": 93470
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.212105263157894e-05,
+      "loss": 0.5388,
+      "step": 93480
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.2042105263157885e-05,
+      "loss": 0.5437,
+      "step": 93490
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.196315789473684e-05,
+      "loss": 0.5417,
+      "step": 93500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.188421052631578e-05,
+      "loss": 0.5515,
+      "step": 93510
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.1805263157894736e-05,
+      "loss": 0.5587,
+      "step": 93520
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.172631578947368e-05,
+      "loss": 0.5602,
+      "step": 93530
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.1647368421052626e-05,
+      "loss": 0.5655,
+      "step": 93540
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.1568421052631575e-05,
+      "loss": 0.5536,
+      "step": 93550
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.1489473684210516e-05,
+      "loss": 0.5767,
+      "step": 93560
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.141052631578947e-05,
+      "loss": 0.5697,
+      "step": 93570
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.133157894736841e-05,
+      "loss": 0.5656,
+      "step": 93580
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.125263157894737e-05,
+      "loss": 0.5627,
+      "step": 93590
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.117368421052631e-05,
+      "loss": 0.5727,
+      "step": 93600
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.109473684210526e-05,
+      "loss": 0.5581,
+      "step": 93610
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.1015789473684206e-05,
+      "loss": 0.5502,
+      "step": 93620
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.0936842105263154e-05,
+      "loss": 0.5505,
+      "step": 93630
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.08578947368421e-05,
+      "loss": 0.5572,
+      "step": 93640
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.0778947368421044e-05,
+      "loss": 0.5467,
+      "step": 93650
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.07e-05,
+      "loss": 0.5507,
+      "step": 93660
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.062105263157894e-05,
+      "loss": 0.5508,
+      "step": 93670
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.054210526315789e-05,
+      "loss": 0.5531,
+      "step": 93680
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.046315789473684e-05,
+      "loss": 0.5446,
+      "step": 93690
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.0384210526315786e-05,
+      "loss": 0.5452,
+      "step": 93700
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.0305263157894734e-05,
+      "loss": 0.5435,
+      "step": 93710
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.0226315789473675e-05,
+      "loss": 0.5439,
+      "step": 93720
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.014736842105263e-05,
+      "loss": 0.5391,
+      "step": 93730
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.006842105263157e-05,
+      "loss": 0.5426,
+      "step": 93740
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.998947368421052e-05,
+      "loss": 0.5356,
+      "step": 93750
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.991052631578947e-05,
+      "loss": 0.5665,
+      "step": 93760
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.983157894736842e-05,
+      "loss": 0.5653,
+      "step": 93770
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.9752631578947365e-05,
+      "loss": 0.5651,
+      "step": 93780
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.967368421052631e-05,
+      "loss": 0.5734,
+      "step": 93790
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.959473684210526e-05,
+      "loss": 0.5607,
+      "step": 93800
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.95157894736842e-05,
+      "loss": 0.5729,
+      "step": 93810
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.943684210526315e-05,
+      "loss": 0.5705,
+      "step": 93820
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.93578947368421e-05,
+      "loss": 0.564,
+      "step": 93830
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.927894736842105e-05,
+      "loss": 0.5598,
+      "step": 93840
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.9199999999999997e-05,
+      "loss": 0.551,
+      "step": 93850
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.9121052631578945e-05,
+      "loss": 0.5414,
+      "step": 93860
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.904210526315789e-05,
+      "loss": 0.5508,
+      "step": 93870
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.8963157894736835e-05,
+      "loss": 0.5563,
+      "step": 93880
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.888421052631578e-05,
+      "loss": 0.5493,
+      "step": 93890
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.880526315789473e-05,
+      "loss": 0.5347,
+      "step": 93900
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.872631578947368e-05,
+      "loss": 0.5435,
+      "step": 93910
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.864736842105263e-05,
+      "loss": 0.5397,
+      "step": 93920
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.8568421052631576e-05,
+      "loss": 0.5361,
+      "step": 93930
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.8489473684210524e-05,
+      "loss": 0.533,
+      "step": 93940
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.8410526315789466e-05,
+      "loss": 0.5266,
+      "step": 93950
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.8331578947368414e-05,
+      "loss": 0.5381,
+      "step": 93960
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.825263157894736e-05,
+      "loss": 0.5453,
+      "step": 93970
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.817368421052631e-05,
+      "loss": 0.5353,
+      "step": 93980
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.809473684210526e-05,
+      "loss": 0.5374,
+      "step": 93990
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.801578947368421e-05,
+      "loss": 0.5461,
+      "step": 94000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.7936842105263156e-05,
+      "loss": 0.5481,
+      "step": 94010
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.78578947368421e-05,
+      "loss": 0.5603,
+      "step": 94020
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.7778947368421046e-05,
+      "loss": 0.5625,
+      "step": 94030
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.7699999999999994e-05,
+      "loss": 0.568,
+      "step": 94040
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.762105263157894e-05,
+      "loss": 0.5599,
+      "step": 94050
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.754210526315789e-05,
+      "loss": 0.5612,
+      "step": 94060
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.746315789473684e-05,
+      "loss": 0.5454,
+      "step": 94070
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.738421052631579e-05,
+      "loss": 0.5487,
+      "step": 94080
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.7305263157894735e-05,
+      "loss": 0.5448,
+      "step": 94090
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.722631578947368e-05,
+      "loss": 0.5491,
+      "step": 94100
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.715526315789473e-05,
+      "loss": 0.5526,
+      "step": 94110
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.707631578947369e-05,
+      "loss": 0.5447,
+      "step": 94120
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.699736842105263e-05,
+      "loss": 0.5488,
+      "step": 94130
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.691842105263157e-05,
+      "loss": 0.5453,
+      "step": 94140
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.6839473684210525e-05,
+      "loss": 0.5392,
+      "step": 94150
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.6760526315789467e-05,
+      "loss": 0.5412,
+      "step": 94160
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.668157894736842e-05,
+      "loss": 0.5335,
+      "step": 94170
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.660263157894736e-05,
+      "loss": 0.5324,
+      "step": 94180
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.652368421052632e-05,
+      "loss": 0.5441,
+      "step": 94190
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.644473684210526e-05,
+      "loss": 0.5295,
+      "step": 94200
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.63657894736842e-05,
+      "loss": 0.5475,
+      "step": 94210
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.6286842105263156e-05,
+      "loss": 0.5501,
+      "step": 94220
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.62078947368421e-05,
+      "loss": 0.567,
+      "step": 94230
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.612894736842105e-05,
+      "loss": 0.5567,
+      "step": 94240
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.6049999999999994e-05,
+      "loss": 0.5674,
+      "step": 94250
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.597105263157895e-05,
+      "loss": 0.5685,
+      "step": 94260
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.589210526315789e-05,
+      "loss": 0.5618,
+      "step": 94270
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.581315789473683e-05,
+      "loss": 0.5669,
+      "step": 94280
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.573421052631579e-05,
+      "loss": 0.5627,
+      "step": 94290
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.565526315789473e-05,
+      "loss": 0.5577,
+      "step": 94300
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.5576315789473684e-05,
+      "loss": 0.5517,
+      "step": 94310
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.5497368421052626e-05,
+      "loss": 0.5455,
+      "step": 94320
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.541842105263158e-05,
+      "loss": 0.5478,
+      "step": 94330
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.533947368421052e-05,
+      "loss": 0.5443,
+      "step": 94340
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.5260526315789464e-05,
+      "loss": 0.5539,
+      "step": 94350
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.518157894736842e-05,
+      "loss": 0.5513,
+      "step": 94360
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.510263157894736e-05,
+      "loss": 0.5428,
+      "step": 94370
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.5023684210526316e-05,
+      "loss": 0.5497,
+      "step": 94380
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.494473684210526e-05,
+      "loss": 0.5517,
+      "step": 94390
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.486578947368421e-05,
+      "loss": 0.5433,
+      "step": 94400
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.4786842105263154e-05,
+      "loss": 0.5405,
+      "step": 94410
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.4707894736842095e-05,
+      "loss": 0.5429,
+      "step": 94420
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.462894736842105e-05,
+      "loss": 0.5556,
+      "step": 94430
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.454999999999999e-05,
+      "loss": 0.5511,
+      "step": 94440
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.447105263157895e-05,
+      "loss": 0.5604,
+      "step": 94450
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.439210526315789e-05,
+      "loss": 0.5699,
+      "step": 94460
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.4313157894736844e-05,
+      "loss": 0.5701,
+      "step": 94470
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.4234210526315785e-05,
+      "loss": 0.5673,
+      "step": 94480
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.415526315789473e-05,
+      "loss": 0.5775,
+      "step": 94490
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.407631578947368e-05,
+      "loss": 0.5747,
+      "step": 94500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.399736842105262e-05,
+      "loss": 0.5858,
+      "step": 94510
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.391842105263158e-05,
+      "loss": 0.5853,
+      "step": 94520
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.383947368421052e-05,
+      "loss": 0.5728,
+      "step": 94530
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.3760526315789475e-05,
+      "loss": 0.5671,
+      "step": 94540
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.3681578947368416e-05,
+      "loss": 0.5611,
+      "step": 94550
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.360263157894736e-05,
+      "loss": 0.5633,
+      "step": 94560
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.352368421052631e-05,
+      "loss": 0.5577,
+      "step": 94570
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.3444736842105255e-05,
+      "loss": 0.5653,
+      "step": 94580
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.336578947368421e-05,
+      "loss": 0.5482,
+      "step": 94590
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.328684210526315e-05,
+      "loss": 0.5551,
+      "step": 94600
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.3207894736842106e-05,
+      "loss": 0.5538,
+      "step": 94610
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.312894736842105e-05,
+      "loss": 0.5382,
+      "step": 94620
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.3049999999999996e-05,
+      "loss": 0.5542,
+      "step": 94630
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.2971052631578944e-05,
+      "loss": 0.5521,
+      "step": 94640
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.2892105263157886e-05,
+      "loss": 0.5469,
+      "step": 94650
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.281315789473684e-05,
+      "loss": 0.5476,
+      "step": 94660
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.273421052631578e-05,
+      "loss": 0.5429,
+      "step": 94670
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.265526315789474e-05,
+      "loss": 0.5527,
+      "step": 94680
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.257631578947368e-05,
+      "loss": 0.5632,
+      "step": 94690
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.249736842105263e-05,
+      "loss": 0.5661,
+      "step": 94700
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.2418421052631576e-05,
+      "loss": 0.5854,
+      "step": 94710
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.233947368421052e-05,
+      "loss": 0.5721,
+      "step": 94720
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.226052631578947e-05,
+      "loss": 0.5783,
+      "step": 94730
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.2181578947368414e-05,
+      "loss": 0.5672,
+      "step": 94740
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.210263157894737e-05,
+      "loss": 0.572,
+      "step": 94750
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.202368421052631e-05,
+      "loss": 0.5643,
+      "step": 94760
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.194473684210526e-05,
+      "loss": 0.5759,
+      "step": 94770
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.186578947368421e-05,
+      "loss": 0.5706,
+      "step": 94780
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.178684210526315e-05,
+      "loss": 0.5665,
+      "step": 94790
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.1707894736842104e-05,
+      "loss": 0.5499,
+      "step": 94800
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.1628947368421045e-05,
+      "loss": 0.548,
+      "step": 94810
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.155e-05,
+      "loss": 0.5503,
+      "step": 94820
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.147105263157894e-05,
+      "loss": 0.558,
+      "step": 94830
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.139210526315789e-05,
+      "loss": 0.5583,
+      "step": 94840
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.131315789473684e-05,
+      "loss": 0.548,
+      "step": 94850
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.123421052631579e-05,
+      "loss": 0.5449,
+      "step": 94860
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.1155263157894735e-05,
+      "loss": 0.5324,
+      "step": 94870
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.1076315789473677e-05,
+      "loss": 0.5373,
+      "step": 94880
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.099736842105263e-05,
+      "loss": 0.5574,
+      "step": 94890
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.091842105263157e-05,
+      "loss": 0.5446,
+      "step": 94900
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.083947368421052e-05,
+      "loss": 0.5565,
+      "step": 94910
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.076052631578947e-05,
+      "loss": 0.5674,
+      "step": 94920
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.068157894736842e-05,
+      "loss": 0.5634,
+      "step": 94930
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.0602631578947366e-05,
+      "loss": 0.5778,
+      "step": 94940
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.052368421052631e-05,
+      "loss": 0.5761,
+      "step": 94950
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.044473684210526e-05,
+      "loss": 0.571,
+      "step": 94960
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.0365789473684205e-05,
+      "loss": 0.5681,
+      "step": 94970
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.028684210526315e-05,
+      "loss": 0.5841,
+      "step": 94980
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.02078947368421e-05,
+      "loss": 0.5761,
+      "step": 94990
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.012894736842105e-05,
+      "loss": 0.5829,
+      "step": 95000
+    },
+    {
+      "epoch": 0.95,
+      "eval_accuracy": 0.8839367508449035,
+      "eval_loss": 0.52685546875,
+      "eval_runtime": 98.3189,
+      "eval_samples_per_second": 813.679,
+      "eval_steps_per_second": 1.597,
+      "step": 95000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.005e-05,
+      "loss": 0.5568,
+      "step": 95010
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.997105263157894e-05,
+      "loss": 0.5498,
+      "step": 95020
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.9892105263157894e-05,
+      "loss": 0.5406,
+      "step": 95030
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.9813157894736836e-05,
+      "loss": 0.5532,
+      "step": 95040
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.9734210526315784e-05,
+      "loss": 0.5551,
+      "step": 95050
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.965526315789473e-05,
+      "loss": 0.5595,
+      "step": 95060
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.957631578947368e-05,
+      "loss": 0.5469,
+      "step": 95070
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.949736842105263e-05,
+      "loss": 0.5454,
+      "step": 95080
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.941842105263158e-05,
+      "loss": 0.5446,
+      "step": 95090
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.9339473684210526e-05,
+      "loss": 0.535,
+      "step": 95100
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.9268421052631574e-05,
+      "loss": 0.536,
+      "step": 95110
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.918947368421053e-05,
+      "loss": 0.5379,
+      "step": 95120
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.911052631578947e-05,
+      "loss": 0.5307,
+      "step": 95130
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.903157894736841e-05,
+      "loss": 0.5353,
+      "step": 95140
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.895263157894737e-05,
+      "loss": 0.5446,
+      "step": 95150
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.887368421052631e-05,
+      "loss": 0.5689,
+      "step": 95160
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.8794736842105264e-05,
+      "loss": 0.568,
+      "step": 95170
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.8715789473684205e-05,
+      "loss": 0.5719,
+      "step": 95180
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.863684210526316e-05,
+      "loss": 0.5674,
+      "step": 95190
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.85578947368421e-05,
+      "loss": 0.5561,
+      "step": 95200
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.847894736842104e-05,
+      "loss": 0.5712,
+      "step": 95210
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.84e-05,
+      "loss": 0.5603,
+      "step": 95220
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.832105263157894e-05,
+      "loss": 0.5641,
+      "step": 95230
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.8242105263157895e-05,
+      "loss": 0.5608,
+      "step": 95240
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.8163157894736836e-05,
+      "loss": 0.5598,
+      "step": 95250
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.808421052631579e-05,
+      "loss": 0.5671,
+      "step": 95260
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.800526315789473e-05,
+      "loss": 0.5663,
+      "step": 95270
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.7926315789473675e-05,
+      "loss": 0.5518,
+      "step": 95280
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.784736842105263e-05,
+      "loss": 0.5732,
+      "step": 95290
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.776842105263157e-05,
+      "loss": 0.5565,
+      "step": 95300
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.7689473684210526e-05,
+      "loss": 0.5531,
+      "step": 95310
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.761052631578947e-05,
+      "loss": 0.5512,
+      "step": 95320
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.753157894736842e-05,
+      "loss": 0.5591,
+      "step": 95330
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.7452631578947364e-05,
+      "loss": 0.553,
+      "step": 95340
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.737368421052631e-05,
+      "loss": 0.5402,
+      "step": 95350
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.729473684210526e-05,
+      "loss": 0.5408,
+      "step": 95360
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.721578947368421e-05,
+      "loss": 0.5403,
+      "step": 95370
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.713684210526316e-05,
+      "loss": 0.5423,
+      "step": 95380
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.70578947368421e-05,
+      "loss": 0.5352,
+      "step": 95390
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.697894736842105e-05,
+      "loss": 0.5481,
+      "step": 95400
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.6899999999999996e-05,
+      "loss": 0.5533,
+      "step": 95410
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.6821052631578944e-05,
+      "loss": 0.548,
+      "step": 95420
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.674210526315789e-05,
+      "loss": 0.5597,
+      "step": 95430
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.666315789473684e-05,
+      "loss": 0.5572,
+      "step": 95440
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.658421052631579e-05,
+      "loss": 0.5442,
+      "step": 95450
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.650526315789473e-05,
+      "loss": 0.554,
+      "step": 95460
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.642631578947368e-05,
+      "loss": 0.5457,
+      "step": 95470
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.634736842105263e-05,
+      "loss": 0.5552,
+      "step": 95480
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.6268421052631575e-05,
+      "loss": 0.5455,
+      "step": 95490
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.6189473684210524e-05,
+      "loss": 0.5406,
+      "step": 95500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.611052631578947e-05,
+      "loss": 0.5361,
+      "step": 95510
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.603157894736842e-05,
+      "loss": 0.5373,
+      "step": 95520
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.595263157894736e-05,
+      "loss": 0.5197,
+      "step": 95530
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.587368421052631e-05,
+      "loss": 0.5248,
+      "step": 95540
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.579473684210526e-05,
+      "loss": 0.5302,
+      "step": 95550
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.571578947368421e-05,
+      "loss": 0.5375,
+      "step": 95560
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5636842105263155e-05,
+      "loss": 0.5219,
+      "step": 95570
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.55578947368421e-05,
+      "loss": 0.5273,
+      "step": 95580
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.547894736842105e-05,
+      "loss": 0.5231,
+      "step": 95590
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.539999999999999e-05,
+      "loss": 0.5261,
+      "step": 95600
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.532105263157894e-05,
+      "loss": 0.5414,
+      "step": 95610
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.524210526315789e-05,
+      "loss": 0.539,
+      "step": 95620
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.516315789473684e-05,
+      "loss": 0.5431,
+      "step": 95630
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5084210526315786e-05,
+      "loss": 0.5294,
+      "step": 95640
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5005263157894735e-05,
+      "loss": 0.5412,
+      "step": 95650
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.492631578947368e-05,
+      "loss": 0.5516,
+      "step": 95660
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4847368421052624e-05,
+      "loss": 0.5552,
+      "step": 95670
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.476842105263157e-05,
+      "loss": 0.55,
+      "step": 95680
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.468947368421052e-05,
+      "loss": 0.5607,
+      "step": 95690
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.461052631578947e-05,
+      "loss": 0.5547,
+      "step": 95700
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.453157894736842e-05,
+      "loss": 0.5573,
+      "step": 95710
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4452631578947366e-05,
+      "loss": 0.5617,
+      "step": 95720
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4373684210526314e-05,
+      "loss": 0.565,
+      "step": 95730
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4294736842105256e-05,
+      "loss": 0.5638,
+      "step": 95740
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4215789473684204e-05,
+      "loss": 0.5517,
+      "step": 95750
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.413684210526315e-05,
+      "loss": 0.5676,
+      "step": 95760
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.40578947368421e-05,
+      "loss": 0.5414,
+      "step": 95770
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.397894736842105e-05,
+      "loss": 0.5531,
+      "step": 95780
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.39e-05,
+      "loss": 0.5515,
+      "step": 95790
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3821052631578946e-05,
+      "loss": 0.5547,
+      "step": 95800
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.374210526315789e-05,
+      "loss": 0.5481,
+      "step": 95810
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3663157894736835e-05,
+      "loss": 0.5464,
+      "step": 95820
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3584210526315784e-05,
+      "loss": 0.5296,
+      "step": 95830
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.350526315789473e-05,
+      "loss": 0.5566,
+      "step": 95840
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.342631578947368e-05,
+      "loss": 0.5197,
+      "step": 95850
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.334736842105263e-05,
+      "loss": 0.5252,
+      "step": 95860
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.326842105263158e-05,
+      "loss": 0.5438,
+      "step": 95870
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.318947368421052e-05,
+      "loss": 0.5394,
+      "step": 95880
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.311052631578947e-05,
+      "loss": 0.5398,
+      "step": 95890
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3031578947368415e-05,
+      "loss": 0.5438,
+      "step": 95900
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.295263157894736e-05,
+      "loss": 0.5639,
+      "step": 95910
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.287368421052631e-05,
+      "loss": 0.5493,
+      "step": 95920
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.279473684210526e-05,
+      "loss": 0.5642,
+      "step": 95930
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.271578947368421e-05,
+      "loss": 0.5573,
+      "step": 95940
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.2636842105263157e-05,
+      "loss": 0.5567,
+      "step": 95950
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.25578947368421e-05,
+      "loss": 0.5554,
+      "step": 95960
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.2478947368421046e-05,
+      "loss": 0.5513,
+      "step": 95970
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.2399999999999995e-05,
+      "loss": 0.5468,
+      "step": 95980
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.232105263157894e-05,
+      "loss": 0.5452,
+      "step": 95990
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.224210526315789e-05,
+      "loss": 0.5277,
+      "step": 96000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.216315789473684e-05,
+      "loss": 0.5449,
+      "step": 96010
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.208421052631579e-05,
+      "loss": 0.5374,
+      "step": 96020
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.200526315789473e-05,
+      "loss": 0.5532,
+      "step": 96030
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.192631578947368e-05,
+      "loss": 0.5255,
+      "step": 96040
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.1847368421052626e-05,
+      "loss": 0.5336,
+      "step": 96050
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.1768421052631574e-05,
+      "loss": 0.5327,
+      "step": 96060
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.168947368421052e-05,
+      "loss": 0.53,
+      "step": 96070
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.161052631578947e-05,
+      "loss": 0.5198,
+      "step": 96080
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.153157894736842e-05,
+      "loss": 0.5226,
+      "step": 96090
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.145263157894736e-05,
+      "loss": 0.5224,
+      "step": 96100
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.138157894736842e-05,
+      "loss": 0.5325,
+      "step": 96110
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.1302631578947364e-05,
+      "loss": 0.5368,
+      "step": 96120
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.122368421052631e-05,
+      "loss": 0.5431,
+      "step": 96130
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.114473684210526e-05,
+      "loss": 0.5476,
+      "step": 96140
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.106578947368421e-05,
+      "loss": 0.5619,
+      "step": 96150
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.098684210526316e-05,
+      "loss": 0.5504,
+      "step": 96160
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.0907894736842105e-05,
+      "loss": 0.5671,
+      "step": 96170
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.0828947368421054e-05,
+      "loss": 0.5477,
+      "step": 96180
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.0749999999999995e-05,
+      "loss": 0.5622,
+      "step": 96190
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.0671052631578944e-05,
+      "loss": 0.5505,
+      "step": 96200
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.059210526315789e-05,
+      "loss": 0.5486,
+      "step": 96210
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.051315789473684e-05,
+      "loss": 0.5309,
+      "step": 96220
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.043421052631579e-05,
+      "loss": 0.5347,
+      "step": 96230
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.0355263157894737e-05,
+      "loss": 0.5366,
+      "step": 96240
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.0276315789473682e-05,
+      "loss": 0.5322,
+      "step": 96250
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.0197368421052627e-05,
+      "loss": 0.5508,
+      "step": 96260
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.0118421052631575e-05,
+      "loss": 0.5386,
+      "step": 96270
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.0039473684210523e-05,
+      "loss": 0.5458,
+      "step": 96280
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.996052631578947e-05,
+      "loss": 0.5432,
+      "step": 96290
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.988157894736842e-05,
+      "loss": 0.5317,
+      "step": 96300
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.9802631578947368e-05,
+      "loss": 0.5331,
+      "step": 96310
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.9723684210526316e-05,
+      "loss": 0.5273,
+      "step": 96320
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.9644736842105258e-05,
+      "loss": 0.5326,
+      "step": 96330
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.9565789473684206e-05,
+      "loss": 0.5359,
+      "step": 96340
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.9486842105263155e-05,
+      "loss": 0.5455,
+      "step": 96350
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.9407894736842103e-05,
+      "loss": 0.5487,
+      "step": 96360
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.932894736842105e-05,
+      "loss": 0.5657,
+      "step": 96370
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.925e-05,
+      "loss": 0.5638,
+      "step": 96380
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.9171052631578948e-05,
+      "loss": 0.5571,
+      "step": 96390
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.909210526315789e-05,
+      "loss": 0.5652,
+      "step": 96400
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.9013157894736838e-05,
+      "loss": 0.5728,
+      "step": 96410
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8934210526315786e-05,
+      "loss": 0.5647,
+      "step": 96420
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8855263157894734e-05,
+      "loss": 0.5567,
+      "step": 96430
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8776315789473683e-05,
+      "loss": 0.558,
+      "step": 96440
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.869736842105263e-05,
+      "loss": 0.5515,
+      "step": 96450
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.861842105263158e-05,
+      "loss": 0.5398,
+      "step": 96460
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.853947368421052e-05,
+      "loss": 0.5497,
+      "step": 96470
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.846052631578947e-05,
+      "loss": 0.5523,
+      "step": 96480
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8381578947368417e-05,
+      "loss": 0.557,
+      "step": 96490
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8302631578947366e-05,
+      "loss": 0.5566,
+      "step": 96500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8223684210526314e-05,
+      "loss": 0.5379,
+      "step": 96510
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8144736842105262e-05,
+      "loss": 0.5407,
+      "step": 96520
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.806578947368421e-05,
+      "loss": 0.5258,
+      "step": 96530
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.798684210526316e-05,
+      "loss": 0.54,
+      "step": 96540
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.79078947368421e-05,
+      "loss": 0.524,
+      "step": 96550
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.782894736842105e-05,
+      "loss": 0.5204,
+      "step": 96560
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7749999999999997e-05,
+      "loss": 0.5338,
+      "step": 96570
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7671052631578945e-05,
+      "loss": 0.541,
+      "step": 96580
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7592105263157893e-05,
+      "loss": 0.5475,
+      "step": 96590
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7513157894736842e-05,
+      "loss": 0.5553,
+      "step": 96600
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.743421052631579e-05,
+      "loss": 0.565,
+      "step": 96610
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.735526315789473e-05,
+      "loss": 0.5593,
+      "step": 96620
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.727631578947368e-05,
+      "loss": 0.5644,
+      "step": 96630
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7197368421052628e-05,
+      "loss": 0.571,
+      "step": 96640
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7118421052631577e-05,
+      "loss": 0.569,
+      "step": 96650
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7039473684210525e-05,
+      "loss": 0.5711,
+      "step": 96660
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6960526315789473e-05,
+      "loss": 0.5648,
+      "step": 96670
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.688157894736842e-05,
+      "loss": 0.5571,
+      "step": 96680
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6802631578947363e-05,
+      "loss": 0.5504,
+      "step": 96690
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.672368421052631e-05,
+      "loss": 0.5558,
+      "step": 96700
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.664473684210526e-05,
+      "loss": 0.5563,
+      "step": 96710
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6565789473684208e-05,
+      "loss": 0.5604,
+      "step": 96720
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6486842105263156e-05,
+      "loss": 0.5428,
+      "step": 96730
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6407894736842104e-05,
+      "loss": 0.5404,
+      "step": 96740
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6328947368421053e-05,
+      "loss": 0.5365,
+      "step": 96750
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6249999999999998e-05,
+      "loss": 0.5401,
+      "step": 96760
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6171052631578943e-05,
+      "loss": 0.5342,
+      "step": 96770
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.609210526315789e-05,
+      "loss": 0.5354,
+      "step": 96780
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.601315789473684e-05,
+      "loss": 0.5306,
+      "step": 96790
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5934210526315788e-05,
+      "loss": 0.536,
+      "step": 96800
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5855263157894736e-05,
+      "loss": 0.5567,
+      "step": 96810
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5776315789473684e-05,
+      "loss": 0.5455,
+      "step": 96820
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.569736842105263e-05,
+      "loss": 0.5507,
+      "step": 96830
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5618421052631574e-05,
+      "loss": 0.5625,
+      "step": 96840
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5539473684210522e-05,
+      "loss": 0.5549,
+      "step": 96850
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.546052631578947e-05,
+      "loss": 0.5499,
+      "step": 96860
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.538157894736842e-05,
+      "loss": 0.5542,
+      "step": 96870
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5302631578947367e-05,
+      "loss": 0.5479,
+      "step": 96880
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5223684210526315e-05,
+      "loss": 0.5515,
+      "step": 96890
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.514473684210526e-05,
+      "loss": 0.5341,
+      "step": 96900
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.506578947368421e-05,
+      "loss": 0.5378,
+      "step": 96910
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4986842105263154e-05,
+      "loss": 0.5312,
+      "step": 96920
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4907894736842102e-05,
+      "loss": 0.5169,
+      "step": 96930
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.482894736842105e-05,
+      "loss": 0.5368,
+      "step": 96940
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.475e-05,
+      "loss": 0.5416,
+      "step": 96950
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4671052631578947e-05,
+      "loss": 0.5266,
+      "step": 96960
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4592105263157892e-05,
+      "loss": 0.5371,
+      "step": 96970
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.451315789473684e-05,
+      "loss": 0.547,
+      "step": 96980
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.443421052631579e-05,
+      "loss": 0.5425,
+      "step": 96990
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4355263157894733e-05,
+      "loss": 0.5337,
+      "step": 97000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.427631578947368e-05,
+      "loss": 0.5488,
+      "step": 97010
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.419736842105263e-05,
+      "loss": 0.5378,
+      "step": 97020
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4118421052631578e-05,
+      "loss": 0.5393,
+      "step": 97030
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4039473684210523e-05,
+      "loss": 0.5527,
+      "step": 97040
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.396052631578947e-05,
+      "loss": 0.5498,
+      "step": 97050
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.388157894736842e-05,
+      "loss": 0.5658,
+      "step": 97060
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3802631578947365e-05,
+      "loss": 0.561,
+      "step": 97070
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3723684210526313e-05,
+      "loss": 0.5657,
+      "step": 97080
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.364473684210526e-05,
+      "loss": 0.5672,
+      "step": 97090
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.356578947368421e-05,
+      "loss": 0.5773,
+      "step": 97100
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.349473684210526e-05,
+      "loss": 0.5737,
+      "step": 97110
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.341578947368421e-05,
+      "loss": 0.5672,
+      "step": 97120
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3336842105263154e-05,
+      "loss": 0.5636,
+      "step": 97130
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3257894736842102e-05,
+      "loss": 0.5599,
+      "step": 97140
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.317894736842105e-05,
+      "loss": 0.5634,
+      "step": 97150
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.31e-05,
+      "loss": 0.5527,
+      "step": 97160
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3021052631578944e-05,
+      "loss": 0.5462,
+      "step": 97170
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2942105263157892e-05,
+      "loss": 0.5676,
+      "step": 97180
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.286315789473684e-05,
+      "loss": 0.5544,
+      "step": 97190
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.278421052631579e-05,
+      "loss": 0.5497,
+      "step": 97200
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2705263157894734e-05,
+      "loss": 0.5432,
+      "step": 97210
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2626315789473682e-05,
+      "loss": 0.5352,
+      "step": 97220
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.254736842105263e-05,
+      "loss": 0.5462,
+      "step": 97230
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.246842105263158e-05,
+      "loss": 0.5296,
+      "step": 97240
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2389473684210524e-05,
+      "loss": 0.5315,
+      "step": 97250
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2310526315789472e-05,
+      "loss": 0.5399,
+      "step": 97260
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.223157894736842e-05,
+      "loss": 0.542,
+      "step": 97270
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2152631578947365e-05,
+      "loss": 0.5405,
+      "step": 97280
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2073684210526313e-05,
+      "loss": 0.5418,
+      "step": 97290
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1994736842105262e-05,
+      "loss": 0.5631,
+      "step": 97300
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.191578947368421e-05,
+      "loss": 0.5695,
+      "step": 97310
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.183684210526316e-05,
+      "loss": 0.5689,
+      "step": 97320
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1757894736842103e-05,
+      "loss": 0.5672,
+      "step": 97330
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.167894736842105e-05,
+      "loss": 0.5668,
+      "step": 97340
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1599999999999996e-05,
+      "loss": 0.5717,
+      "step": 97350
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1521052631578945e-05,
+      "loss": 0.5578,
+      "step": 97360
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1442105263157893e-05,
+      "loss": 0.5658,
+      "step": 97370
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.136315789473684e-05,
+      "loss": 0.5531,
+      "step": 97380
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.128421052631579e-05,
+      "loss": 0.5455,
+      "step": 97390
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1205263157894735e-05,
+      "loss": 0.5528,
+      "step": 97400
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1126315789473683e-05,
+      "loss": 0.5488,
+      "step": 97410
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1047368421052628e-05,
+      "loss": 0.5441,
+      "step": 97420
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0968421052631576e-05,
+      "loss": 0.5538,
+      "step": 97430
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0889473684210524e-05,
+      "loss": 0.5341,
+      "step": 97440
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0810526315789473e-05,
+      "loss": 0.5429,
+      "step": 97450
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.073157894736842e-05,
+      "loss": 0.5401,
+      "step": 97460
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.065263157894737e-05,
+      "loss": 0.5352,
+      "step": 97470
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0573684210526314e-05,
+      "loss": 0.5236,
+      "step": 97480
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.049473684210526e-05,
+      "loss": 0.5251,
+      "step": 97490
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0415789473684207e-05,
+      "loss": 0.5369,
+      "step": 97500
+    },
+    {
+      "epoch": 0.97,
+      "eval_accuracy": 0.8848131360716049,
+      "eval_loss": 0.5224609375,
+      "eval_runtime": 97.6227,
+      "eval_samples_per_second": 819.482,
+      "eval_steps_per_second": 1.608,
+      "step": 97500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0336842105263156e-05,
+      "loss": 0.5494,
+      "step": 97510
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0257894736842104e-05,
+      "loss": 0.5459,
+      "step": 97520
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0178947368421052e-05,
+      "loss": 0.5491,
+      "step": 97530
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.01e-05,
+      "loss": 0.5605,
+      "step": 97540
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.002105263157895e-05,
+      "loss": 0.5561,
+      "step": 97550
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.994210526315789e-05,
+      "loss": 0.5638,
+      "step": 97560
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.986315789473684e-05,
+      "loss": 0.5807,
+      "step": 97570
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9784210526315787e-05,
+      "loss": 0.5669,
+      "step": 97580
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9705263157894735e-05,
+      "loss": 0.568,
+      "step": 97590
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9626315789473684e-05,
+      "loss": 0.5584,
+      "step": 97600
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9547368421052632e-05,
+      "loss": 0.5545,
+      "step": 97610
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.946842105263158e-05,
+      "loss": 0.5602,
+      "step": 97620
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9389473684210522e-05,
+      "loss": 0.5527,
+      "step": 97630
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.931052631578947e-05,
+      "loss": 0.5406,
+      "step": 97640
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.923157894736842e-05,
+      "loss": 0.5477,
+      "step": 97650
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9152631578947367e-05,
+      "loss": 0.5427,
+      "step": 97660
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9073684210526315e-05,
+      "loss": 0.547,
+      "step": 97670
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8994736842105263e-05,
+      "loss": 0.5386,
+      "step": 97680
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.891578947368421e-05,
+      "loss": 0.546,
+      "step": 97690
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8836842105263153e-05,
+      "loss": 0.5366,
+      "step": 97700
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.87578947368421e-05,
+      "loss": 0.546,
+      "step": 97710
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.867894736842105e-05,
+      "loss": 0.5417,
+      "step": 97720
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8599999999999998e-05,
+      "loss": 0.5293,
+      "step": 97730
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8521052631578946e-05,
+      "loss": 0.5449,
+      "step": 97740
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.844210526315789e-05,
+      "loss": 0.5507,
+      "step": 97750
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.836315789473684e-05,
+      "loss": 0.5582,
+      "step": 97760
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8284210526315788e-05,
+      "loss": 0.5668,
+      "step": 97770
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8205263157894736e-05,
+      "loss": 0.5599,
+      "step": 97780
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.812631578947368e-05,
+      "loss": 0.5694,
+      "step": 97790
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.804736842105263e-05,
+      "loss": 0.5687,
+      "step": 97800
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7968421052631578e-05,
+      "loss": 0.5715,
+      "step": 97810
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7889473684210526e-05,
+      "loss": 0.5682,
+      "step": 97820
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.781052631578947e-05,
+      "loss": 0.5604,
+      "step": 97830
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.773157894736842e-05,
+      "loss": 0.5544,
+      "step": 97840
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7652631578947368e-05,
+      "loss": 0.562,
+      "step": 97850
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7573684210526312e-05,
+      "loss": 0.5514,
+      "step": 97860
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.749473684210526e-05,
+      "loss": 0.5468,
+      "step": 97870
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.741578947368421e-05,
+      "loss": 0.5509,
+      "step": 97880
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7336842105263157e-05,
+      "loss": 0.5626,
+      "step": 97890
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7257894736842102e-05,
+      "loss": 0.5478,
+      "step": 97900
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.717894736842105e-05,
+      "loss": 0.5339,
+      "step": 97910
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.71e-05,
+      "loss": 0.5395,
+      "step": 97920
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7021052631578944e-05,
+      "loss": 0.5329,
+      "step": 97930
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6942105263157892e-05,
+      "loss": 0.5265,
+      "step": 97940
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.686315789473684e-05,
+      "loss": 0.5336,
+      "step": 97950
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.678421052631579e-05,
+      "loss": 0.5251,
+      "step": 97960
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6705263157894734e-05,
+      "loss": 0.5393,
+      "step": 97970
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6626315789473682e-05,
+      "loss": 0.5467,
+      "step": 97980
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.654736842105263e-05,
+      "loss": 0.5554,
+      "step": 97990
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6468421052631575e-05,
+      "loss": 0.5688,
+      "step": 98000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6389473684210523e-05,
+      "loss": 0.5638,
+      "step": 98010
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6310526315789472e-05,
+      "loss": 0.5665,
+      "step": 98020
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.623157894736842e-05,
+      "loss": 0.5707,
+      "step": 98030
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6152631578947365e-05,
+      "loss": 0.5603,
+      "step": 98040
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6073684210526313e-05,
+      "loss": 0.5643,
+      "step": 98050
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.599473684210526e-05,
+      "loss": 0.5695,
+      "step": 98060
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5915789473684207e-05,
+      "loss": 0.5653,
+      "step": 98070
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5836842105263155e-05,
+      "loss": 0.5578,
+      "step": 98080
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5757894736842103e-05,
+      "loss": 0.5433,
+      "step": 98090
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.567894736842105e-05,
+      "loss": 0.5342,
+      "step": 98100
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5599999999999996e-05,
+      "loss": 0.5509,
+      "step": 98110
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.552894736842105e-05,
+      "loss": 0.5387,
+      "step": 98120
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.545e-05,
+      "loss": 0.5577,
+      "step": 98130
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5371052631578944e-05,
+      "loss": 0.5427,
+      "step": 98140
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5292105263157893e-05,
+      "loss": 0.5401,
+      "step": 98150
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5213157894736841e-05,
+      "loss": 0.5331,
+      "step": 98160
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5134210526315788e-05,
+      "loss": 0.5166,
+      "step": 98170
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5055263157894736e-05,
+      "loss": 0.5319,
+      "step": 98180
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4976315789473684e-05,
+      "loss": 0.5298,
+      "step": 98190
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4897368421052629e-05,
+      "loss": 0.5225,
+      "step": 98200
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4818421052631577e-05,
+      "loss": 0.5472,
+      "step": 98210
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4739473684210526e-05,
+      "loss": 0.5575,
+      "step": 98220
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4660526315789474e-05,
+      "loss": 0.5471,
+      "step": 98230
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4581578947368419e-05,
+      "loss": 0.565,
+      "step": 98240
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4502631578947367e-05,
+      "loss": 0.5484,
+      "step": 98250
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4423684210526316e-05,
+      "loss": 0.5682,
+      "step": 98260
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.434473684210526e-05,
+      "loss": 0.5639,
+      "step": 98270
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4265789473684209e-05,
+      "loss": 0.5628,
+      "step": 98280
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4186842105263157e-05,
+      "loss": 0.5645,
+      "step": 98290
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4107894736842105e-05,
+      "loss": 0.558,
+      "step": 98300
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.402894736842105e-05,
+      "loss": 0.5468,
+      "step": 98310
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3949999999999999e-05,
+      "loss": 0.555,
+      "step": 98320
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3871052631578947e-05,
+      "loss": 0.5515,
+      "step": 98330
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3792105263157892e-05,
+      "loss": 0.5408,
+      "step": 98340
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.371315789473684e-05,
+      "loss": 0.5386,
+      "step": 98350
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3634210526315788e-05,
+      "loss": 0.54,
+      "step": 98360
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3555263157894737e-05,
+      "loss": 0.5423,
+      "step": 98370
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3476315789473682e-05,
+      "loss": 0.5279,
+      "step": 98380
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.339736842105263e-05,
+      "loss": 0.537,
+      "step": 98390
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3318421052631578e-05,
+      "loss": 0.536,
+      "step": 98400
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3239473684210527e-05,
+      "loss": 0.5422,
+      "step": 98410
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3160526315789471e-05,
+      "loss": 0.544,
+      "step": 98420
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.308157894736842e-05,
+      "loss": 0.5422,
+      "step": 98430
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3002631578947368e-05,
+      "loss": 0.5319,
+      "step": 98440
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2923684210526315e-05,
+      "loss": 0.5468,
+      "step": 98450
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2844736842105261e-05,
+      "loss": 0.5449,
+      "step": 98460
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.276578947368421e-05,
+      "loss": 0.5569,
+      "step": 98470
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2686842105263158e-05,
+      "loss": 0.5599,
+      "step": 98480
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2607894736842104e-05,
+      "loss": 0.5608,
+      "step": 98490
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2528947368421051e-05,
+      "loss": 0.5747,
+      "step": 98500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.245e-05,
+      "loss": 0.5713,
+      "step": 98510
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2371052631578946e-05,
+      "loss": 0.5587,
+      "step": 98520
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2292105263157893e-05,
+      "loss": 0.568,
+      "step": 98530
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2213157894736841e-05,
+      "loss": 0.5496,
+      "step": 98540
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.213421052631579e-05,
+      "loss": 0.5667,
+      "step": 98550
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2055263157894736e-05,
+      "loss": 0.5402,
+      "step": 98560
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1976315789473682e-05,
+      "loss": 0.5294,
+      "step": 98570
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.189736842105263e-05,
+      "loss": 0.5306,
+      "step": 98580
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1818421052631577e-05,
+      "loss": 0.5317,
+      "step": 98590
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1739473684210526e-05,
+      "loss": 0.5424,
+      "step": 98600
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1660526315789472e-05,
+      "loss": 0.5175,
+      "step": 98610
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.158157894736842e-05,
+      "loss": 0.5112,
+      "step": 98620
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1502631578947367e-05,
+      "loss": 0.5215,
+      "step": 98630
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1423684210526315e-05,
+      "loss": 0.5233,
+      "step": 98640
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1344736842105262e-05,
+      "loss": 0.5117,
+      "step": 98650
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1265789473684209e-05,
+      "loss": 0.5186,
+      "step": 98660
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1186842105263157e-05,
+      "loss": 0.5214,
+      "step": 98670
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1107894736842105e-05,
+      "loss": 0.5307,
+      "step": 98680
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1028947368421052e-05,
+      "loss": 0.5245,
+      "step": 98690
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0949999999999998e-05,
+      "loss": 0.5476,
+      "step": 98700
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0871052631578947e-05,
+      "loss": 0.5443,
+      "step": 98710
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0792105263157895e-05,
+      "loss": 0.5479,
+      "step": 98720
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.071315789473684e-05,
+      "loss": 0.5547,
+      "step": 98730
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0634210526315788e-05,
+      "loss": 0.5645,
+      "step": 98740
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0555263157894737e-05,
+      "loss": 0.5748,
+      "step": 98750
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0476315789473683e-05,
+      "loss": 0.5618,
+      "step": 98760
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.039736842105263e-05,
+      "loss": 0.5506,
+      "step": 98770
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0318421052631578e-05,
+      "loss": 0.529,
+      "step": 98780
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0239473684210526e-05,
+      "loss": 0.536,
+      "step": 98790
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0160526315789473e-05,
+      "loss": 0.5472,
+      "step": 98800
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.008157894736842e-05,
+      "loss": 0.5439,
+      "step": 98810
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0002631578947368e-05,
+      "loss": 0.5529,
+      "step": 98820
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.923684210526316e-06,
+      "loss": 0.5439,
+      "step": 98830
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.844736842105261e-06,
+      "loss": 0.533,
+      "step": 98840
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.76578947368421e-06,
+      "loss": 0.5391,
+      "step": 98850
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.686842105263158e-06,
+      "loss": 0.5274,
+      "step": 98860
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.607894736842106e-06,
+      "loss": 0.5355,
+      "step": 98870
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.528947368421051e-06,
+      "loss": 0.5272,
+      "step": 98880
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.45e-06,
+      "loss": 0.5291,
+      "step": 98890
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.371052631578946e-06,
+      "loss": 0.5353,
+      "step": 98900
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.292105263157894e-06,
+      "loss": 0.5427,
+      "step": 98910
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.21315789473684e-06,
+      "loss": 0.5431,
+      "step": 98920
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.134210526315789e-06,
+      "loss": 0.5431,
+      "step": 98930
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.055263157894736e-06,
+      "loss": 0.5569,
+      "step": 98940
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.976315789473684e-06,
+      "loss": 0.563,
+      "step": 98950
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.89736842105263e-06,
+      "loss": 0.5641,
+      "step": 98960
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.818421052631577e-06,
+      "loss": 0.5705,
+      "step": 98970
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.74736842105263e-06,
+      "loss": 0.5701,
+      "step": 98980
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.668421052631579e-06,
+      "loss": 0.5524,
+      "step": 98990
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.589473684210525e-06,
+      "loss": 0.5472,
+      "step": 99000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.510526315789472e-06,
+      "loss": 0.5477,
+      "step": 99010
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.43157894736842e-06,
+      "loss": 0.5383,
+      "step": 99020
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.352631578947367e-06,
+      "loss": 0.5362,
+      "step": 99030
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.273684210526315e-06,
+      "loss": 0.5428,
+      "step": 99040
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.194736842105262e-06,
+      "loss": 0.5357,
+      "step": 99050
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.11578947368421e-06,
+      "loss": 0.5365,
+      "step": 99060
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.036842105263157e-06,
+      "loss": 0.5419,
+      "step": 99070
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.957894736842103e-06,
+      "loss": 0.5353,
+      "step": 99080
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.878947368421052e-06,
+      "loss": 0.5401,
+      "step": 99090
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.799999999999998e-06,
+      "loss": 0.5318,
+      "step": 99100
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.721052631578946e-06,
+      "loss": 0.5423,
+      "step": 99110
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.642105263157893e-06,
+      "loss": 0.5223,
+      "step": 99120
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.563157894736842e-06,
+      "loss": 0.5314,
+      "step": 99130
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.484210526315789e-06,
+      "loss": 0.5375,
+      "step": 99140
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.405263157894736e-06,
+      "loss": 0.5475,
+      "step": 99150
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.326315789473684e-06,
+      "loss": 0.5481,
+      "step": 99160
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.24736842105263e-06,
+      "loss": 0.5519,
+      "step": 99170
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.168421052631579e-06,
+      "loss": 0.5675,
+      "step": 99180
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.089473684210525e-06,
+      "loss": 0.5612,
+      "step": 99190
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.0105263157894736e-06,
+      "loss": 0.5572,
+      "step": 99200
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.93157894736842e-06,
+      "loss": 0.5683,
+      "step": 99210
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.8526315789473685e-06,
+      "loss": 0.5542,
+      "step": 99220
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.773684210526315e-06,
+      "loss": 0.5614,
+      "step": 99230
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.694736842105262e-06,
+      "loss": 0.5456,
+      "step": 99240
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.61578947368421e-06,
+      "loss": 0.5442,
+      "step": 99250
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.536842105263157e-06,
+      "loss": 0.5416,
+      "step": 99260
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.457894736842105e-06,
+      "loss": 0.5463,
+      "step": 99270
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.3789473684210515e-06,
+      "loss": 0.5546,
+      "step": 99280
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.3e-06,
+      "loss": 0.5339,
+      "step": 99290
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.2210526315789464e-06,
+      "loss": 0.5405,
+      "step": 99300
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.142105263157894e-06,
+      "loss": 0.533,
+      "step": 99310
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.063157894736841e-06,
+      "loss": 0.5315,
+      "step": 99320
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.984210526315789e-06,
+      "loss": 0.541,
+      "step": 99330
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.905263157894736e-06,
+      "loss": 0.5386,
+      "step": 99340
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.826315789473684e-06,
+      "loss": 0.5231,
+      "step": 99350
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.747368421052631e-06,
+      "loss": 0.5433,
+      "step": 99360
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.668421052631579e-06,
+      "loss": 0.5502,
+      "step": 99370
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.589473684210526e-06,
+      "loss": 0.5589,
+      "step": 99380
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.5105263157894735e-06,
+      "loss": 0.5584,
+      "step": 99390
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.43157894736842e-06,
+      "loss": 0.5622,
+      "step": 99400
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.3526315789473684e-06,
+      "loss": 0.562,
+      "step": 99410
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.273684210526315e-06,
+      "loss": 0.572,
+      "step": 99420
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.194736842105263e-06,
+      "loss": 0.5653,
+      "step": 99430
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.11578947368421e-06,
+      "loss": 0.5621,
+      "step": 99440
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.036842105263158e-06,
+      "loss": 0.5488,
+      "step": 99450
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.957894736842105e-06,
+      "loss": 0.5379,
+      "step": 99460
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.8789473684210515e-06,
+      "loss": 0.5541,
+      "step": 99470
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.8e-06,
+      "loss": 0.548,
+      "step": 99480
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.721052631578946e-06,
+      "loss": 0.5427,
+      "step": 99490
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.642105263157895e-06,
+      "loss": 0.5592,
+      "step": 99500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.563157894736841e-06,
+      "loss": 0.5521,
+      "step": 99510
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.484210526315789e-06,
+      "loss": 0.5467,
+      "step": 99520
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.405263157894736e-06,
+      "loss": 0.5407,
+      "step": 99530
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.326315789473684e-06,
+      "loss": 0.5413,
+      "step": 99540
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.247368421052631e-06,
+      "loss": 0.5337,
+      "step": 99550
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.168421052631579e-06,
+      "loss": 0.5315,
+      "step": 99560
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.089473684210526e-06,
+      "loss": 0.5279,
+      "step": 99570
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.0105263157894735e-06,
+      "loss": 0.5294,
+      "step": 99580
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.93157894736842e-06,
+      "loss": 0.5355,
+      "step": 99590
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.8526315789473676e-06,
+      "loss": 0.5406,
+      "step": 99600
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7736842105263154e-06,
+      "loss": 0.5498,
+      "step": 99610
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.694736842105263e-06,
+      "loss": 0.55,
+      "step": 99620
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6157894736842103e-06,
+      "loss": 0.5627,
+      "step": 99630
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.536842105263158e-06,
+      "loss": 0.5577,
+      "step": 99640
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.4578947368421053e-06,
+      "loss": 0.5669,
+      "step": 99650
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.378947368421052e-06,
+      "loss": 0.5615,
+      "step": 99660
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.2999999999999993e-06,
+      "loss": 0.5618,
+      "step": 99670
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.2210526315789468e-06,
+      "loss": 0.551,
+      "step": 99680
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.1421052631578942e-06,
+      "loss": 0.5513,
+      "step": 99690
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.0631578947368417e-06,
+      "loss": 0.5531,
+      "step": 99700
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.984210526315789e-06,
+      "loss": 0.5431,
+      "step": 99710
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9052631578947366e-06,
+      "loss": 0.537,
+      "step": 99720
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.826315789473684e-06,
+      "loss": 0.5366,
+      "step": 99730
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.747368421052631e-06,
+      "loss": 0.5352,
+      "step": 99740
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6684210526315785e-06,
+      "loss": 0.5337,
+      "step": 99750
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.589473684210526e-06,
+      "loss": 0.5264,
+      "step": 99760
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5105263157894735e-06,
+      "loss": 0.5306,
+      "step": 99770
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.431578947368421e-06,
+      "loss": 0.537,
+      "step": 99780
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3526315789473684e-06,
+      "loss": 0.5183,
+      "step": 99790
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2736842105263154e-06,
+      "loss": 0.525,
+      "step": 99800
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.194736842105263e-06,
+      "loss": 0.5179,
+      "step": 99810
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1157894736842103e-06,
+      "loss": 0.525,
+      "step": 99820
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0368421052631578e-06,
+      "loss": 0.5342,
+      "step": 99830
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.957894736842105e-06,
+      "loss": 0.553,
+      "step": 99840
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8789473684210525e-06,
+      "loss": 0.5481,
+      "step": 99850
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8e-06,
+      "loss": 0.5443,
+      "step": 99860
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7210526315789474e-06,
+      "loss": 0.5509,
+      "step": 99870
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6421052631578944e-06,
+      "loss": 0.5453,
+      "step": 99880
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5631578947368419e-06,
+      "loss": 0.5505,
+      "step": 99890
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4842105263157893e-06,
+      "loss": 0.5574,
+      "step": 99900
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4052631578947368e-06,
+      "loss": 0.5573,
+      "step": 99910
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.326315789473684e-06,
+      "loss": 0.553,
+      "step": 99920
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2473684210526315e-06,
+      "loss": 0.5504,
+      "step": 99930
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.168421052631579e-06,
+      "loss": 0.5388,
+      "step": 99940
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0894736842105262e-06,
+      "loss": 0.534,
+      "step": 99950
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0105263157894736e-06,
+      "loss": 0.5417,
+      "step": 99960
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.31578947368421e-07,
+      "loss": 0.5396,
+      "step": 99970
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.526315789473684e-07,
+      "loss": 0.5378,
+      "step": 99980
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.736842105263157e-07,
+      "loss": 0.5274,
+      "step": 99990
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.947368421052631e-07,
+      "loss": 0.532,
+      "step": 100000
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": 0.8854931609881757,
+      "eval_loss": 0.51953125,
+      "eval_runtime": 99.2784,
+      "eval_samples_per_second": 805.815,
+      "eval_steps_per_second": 1.581,
+      "step": 100000
+    },
+    {
+      "epoch": 1.0,
+      "step": 100000,
+      "total_flos": 3.58736203874304e+19,
+      "train_loss": 0.8820991159725189,
+      "train_runtime": 81002.545,
+      "train_samples_per_second": 1264.158,
+      "train_steps_per_second": 1.235
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 100000,
+  "num_train_epochs": 9223372036854775807,
+  "save_steps": 2500,
+  "total_flos": 3.58736203874304e+19,
+  "trial_name": null,
+  "trial_params": null
+}