diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,6070 @@
+{
+  "best_metric": 0.21704278886318207,
+  "best_model_checkpoint": "./outputs/checkpoint-10000",
+  "epoch": 5.0,
+  "global_step": 10000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9980000000000002e-05,
+      "loss": 0.6939,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9960000000000002e-05,
+      "loss": 0.625,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9940000000000002e-05,
+      "loss": 0.7085,
+      "step": 30
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9920000000000002e-05,
+      "loss": 0.7261,
+      "step": 40
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9900000000000003e-05,
+      "loss": 0.7178,
+      "step": 50
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9880000000000003e-05,
+      "loss": 0.6555,
+      "step": 60
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9860000000000003e-05,
+      "loss": 0.6719,
+      "step": 70
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9840000000000003e-05,
+      "loss": 0.7012,
+      "step": 80
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.982e-05,
+      "loss": 0.8066,
+      "step": 90
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.98e-05,
+      "loss": 0.7443,
+      "step": 100
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.978e-05,
+      "loss": 0.5879,
+      "step": 110
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.976e-05,
+      "loss": 0.625,
+      "step": 120
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.974e-05,
+      "loss": 0.6223,
+      "step": 130
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.972e-05,
+      "loss": 0.5434,
+      "step": 140
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.97e-05,
+      "loss": 0.5689,
+      "step": 150
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.968e-05,
+      "loss": 0.6387,
+      "step": 160
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.966e-05,
+      "loss": 0.6345,
+      "step": 170
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9640000000000002e-05,
+      "loss": 0.7149,
+      "step": 180
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9620000000000002e-05,
+      "loss": 0.6409,
+      "step": 190
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9600000000000002e-05,
+      "loss": 0.6555,
+      "step": 200
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9580000000000002e-05,
+      "loss": 0.6102,
+      "step": 210
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9560000000000002e-05,
+      "loss": 0.5622,
+      "step": 220
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9540000000000003e-05,
+      "loss": 0.6004,
+      "step": 230
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9520000000000003e-05,
+      "loss": 0.6103,
+      "step": 240
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95e-05,
+      "loss": 0.5665,
+      "step": 250
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948e-05,
+      "loss": 0.5606,
+      "step": 260
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.946e-05,
+      "loss": 0.6076,
+      "step": 270
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.944e-05,
+      "loss": 0.6076,
+      "step": 280
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942e-05,
+      "loss": 0.6752,
+      "step": 290
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.94e-05,
+      "loss": 0.6028,
+      "step": 300
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.938e-05,
+      "loss": 0.5248,
+      "step": 310
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.936e-05,
+      "loss": 0.5376,
+      "step": 320
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.934e-05,
+      "loss": 0.623,
+      "step": 330
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.932e-05,
+      "loss": 0.5682,
+      "step": 340
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.93e-05,
+      "loss": 0.6842,
+      "step": 350
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9280000000000002e-05,
+      "loss": 0.5146,
+      "step": 360
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9260000000000002e-05,
+      "loss": 0.6353,
+      "step": 370
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9240000000000002e-05,
+      "loss": 0.5733,
+      "step": 380
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9220000000000002e-05,
+      "loss": 0.514,
+      "step": 390
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9200000000000003e-05,
+      "loss": 0.5455,
+      "step": 400
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.918e-05,
+      "loss": 0.5864,
+      "step": 410
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.916e-05,
+      "loss": 0.5682,
+      "step": 420
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.914e-05,
+      "loss": 0.5623,
+      "step": 430
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.912e-05,
+      "loss": 0.589,
+      "step": 440
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.91e-05,
+      "loss": 0.4743,
+      "step": 450
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.908e-05,
+      "loss": 0.6108,
+      "step": 460
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.906e-05,
+      "loss": 0.5182,
+      "step": 470
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.904e-05,
+      "loss": 0.5723,
+      "step": 480
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.902e-05,
+      "loss": 0.5332,
+      "step": 490
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9e-05,
+      "loss": 0.5283,
+      "step": 500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.898e-05,
+      "loss": 0.5365,
+      "step": 510
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.896e-05,
+      "loss": 0.5412,
+      "step": 520
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.894e-05,
+      "loss": 0.5126,
+      "step": 530
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8920000000000002e-05,
+      "loss": 0.4809,
+      "step": 540
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8900000000000002e-05,
+      "loss": 0.5309,
+      "step": 550
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8880000000000002e-05,
+      "loss": 0.8295,
+      "step": 560
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.886e-05,
+      "loss": 0.6505,
+      "step": 570
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.884e-05,
+      "loss": 0.5147,
+      "step": 580
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.882e-05,
+      "loss": 0.5151,
+      "step": 590
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.88e-05,
+      "loss": 0.5735,
+      "step": 600
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.878e-05,
+      "loss": 0.5699,
+      "step": 610
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.876e-05,
+      "loss": 0.488,
+      "step": 620
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8740000000000004e-05,
+      "loss": 0.5066,
+      "step": 630
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8720000000000004e-05,
+      "loss": 0.5173,
+      "step": 640
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8700000000000004e-05,
+      "loss": 0.5082,
+      "step": 650
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8680000000000004e-05,
+      "loss": 0.4313,
+      "step": 660
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.866e-05,
+      "loss": 0.4317,
+      "step": 670
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.864e-05,
+      "loss": 0.63,
+      "step": 680
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.862e-05,
+      "loss": 0.5844,
+      "step": 690
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.86e-05,
+      "loss": 0.609,
+      "step": 700
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.858e-05,
+      "loss": 0.4647,
+      "step": 710
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8560000000000002e-05,
+      "loss": 0.5973,
+      "step": 720
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8540000000000002e-05,
+      "loss": 0.6601,
+      "step": 730
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8520000000000002e-05,
+      "loss": 0.5252,
+      "step": 740
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8500000000000002e-05,
+      "loss": 0.5705,
+      "step": 750
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8480000000000003e-05,
+      "loss": 0.5515,
+      "step": 760
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8460000000000003e-05,
+      "loss": 0.5659,
+      "step": 770
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8440000000000003e-05,
+      "loss": 0.5333,
+      "step": 780
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8420000000000003e-05,
+      "loss": 0.6172,
+      "step": 790
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8400000000000003e-05,
+      "loss": 0.5093,
+      "step": 800
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8380000000000004e-05,
+      "loss": 0.4659,
+      "step": 810
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8360000000000004e-05,
+      "loss": 0.4899,
+      "step": 820
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.834e-05,
+      "loss": 0.4629,
+      "step": 830
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.832e-05,
+      "loss": 0.6965,
+      "step": 840
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.83e-05,
+      "loss": 0.4061,
+      "step": 850
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.828e-05,
+      "loss": 0.4113,
+      "step": 860
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.826e-05,
+      "loss": 0.4064,
+      "step": 870
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.824e-05,
+      "loss": 0.5627,
+      "step": 880
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8220000000000002e-05,
+      "loss": 0.5696,
+      "step": 890
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8200000000000002e-05,
+      "loss": 0.5148,
+      "step": 900
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8180000000000002e-05,
+      "loss": 0.5677,
+      "step": 910
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8160000000000002e-05,
+      "loss": 0.5877,
+      "step": 920
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8140000000000003e-05,
+      "loss": 0.5872,
+      "step": 930
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8120000000000003e-05,
+      "loss": 0.52,
+      "step": 940
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8100000000000003e-05,
+      "loss": 0.529,
+      "step": 950
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8080000000000003e-05,
+      "loss": 0.4559,
+      "step": 960
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8060000000000003e-05,
+      "loss": 0.6564,
+      "step": 970
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8040000000000003e-05,
+      "loss": 0.5445,
+      "step": 980
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.802e-05,
+      "loss": 0.5945,
+      "step": 990
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8e-05,
+      "loss": 0.4755,
+      "step": 1000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.798e-05,
+      "loss": 0.5667,
+      "step": 1010
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.796e-05,
+      "loss": 0.511,
+      "step": 1020
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.794e-05,
+      "loss": 0.5742,
+      "step": 1030
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.792e-05,
+      "loss": 0.4517,
+      "step": 1040
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.79e-05,
+      "loss": 0.5685,
+      "step": 1050
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.788e-05,
+      "loss": 0.4295,
+      "step": 1060
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7860000000000002e-05,
+      "loss": 0.5063,
+      "step": 1070
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7840000000000002e-05,
+      "loss": 0.4871,
+      "step": 1080
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7820000000000002e-05,
+      "loss": 0.4698,
+      "step": 1090
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7800000000000002e-05,
+      "loss": 0.5343,
+      "step": 1100
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7780000000000003e-05,
+      "loss": 0.4913,
+      "step": 1110
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7760000000000003e-05,
+      "loss": 0.4589,
+      "step": 1120
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7740000000000003e-05,
+      "loss": 0.5496,
+      "step": 1130
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7720000000000003e-05,
+      "loss": 0.63,
+      "step": 1140
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.77e-05,
+      "loss": 0.5309,
+      "step": 1150
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.768e-05,
+      "loss": 0.6158,
+      "step": 1160
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.766e-05,
+      "loss": 0.5591,
+      "step": 1170
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.764e-05,
+      "loss": 0.4843,
+      "step": 1180
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.762e-05,
+      "loss": 0.5171,
+      "step": 1190
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.76e-05,
+      "loss": 0.5577,
+      "step": 1200
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.758e-05,
+      "loss": 0.4909,
+      "step": 1210
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.756e-05,
+      "loss": 0.4515,
+      "step": 1220
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.754e-05,
+      "loss": 0.4726,
+      "step": 1230
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.752e-05,
+      "loss": 0.405,
+      "step": 1240
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7500000000000002e-05,
+      "loss": 0.572,
+      "step": 1250
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7480000000000002e-05,
+      "loss": 0.6129,
+      "step": 1260
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7460000000000002e-05,
+      "loss": 0.4911,
+      "step": 1270
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7440000000000002e-05,
+      "loss": 0.5328,
+      "step": 1280
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7420000000000003e-05,
+      "loss": 0.5258,
+      "step": 1290
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7400000000000003e-05,
+      "loss": 0.4596,
+      "step": 1300
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7380000000000003e-05,
+      "loss": 0.5182,
+      "step": 1310
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.736e-05,
+      "loss": 0.504,
+      "step": 1320
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.734e-05,
+      "loss": 0.5586,
+      "step": 1330
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.732e-05,
+      "loss": 0.506,
+      "step": 1340
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.73e-05,
+      "loss": 0.4121,
+      "step": 1350
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.728e-05,
+      "loss": 0.537,
+      "step": 1360
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.726e-05,
+      "loss": 0.5619,
+      "step": 1370
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.724e-05,
+      "loss": 0.4346,
+      "step": 1380
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.722e-05,
+      "loss": 0.5476,
+      "step": 1390
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.72e-05,
+      "loss": 0.4887,
+      "step": 1400
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.718e-05,
+      "loss": 0.5839,
+      "step": 1410
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7160000000000002e-05,
+      "loss": 0.5675,
+      "step": 1420
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7140000000000002e-05,
+      "loss": 0.4049,
+      "step": 1430
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7120000000000002e-05,
+      "loss": 0.4323,
+      "step": 1440
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7100000000000002e-05,
+      "loss": 0.4614,
+      "step": 1450
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7080000000000002e-05,
+      "loss": 0.5135,
+      "step": 1460
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7060000000000003e-05,
+      "loss": 0.4687,
+      "step": 1470
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.704e-05,
+      "loss": 0.4394,
+      "step": 1480
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.702e-05,
+      "loss": 0.4685,
+      "step": 1490
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7e-05,
+      "loss": 0.5203,
+      "step": 1500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.698e-05,
+      "loss": 0.5185,
+      "step": 1510
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.696e-05,
+      "loss": 0.5369,
+      "step": 1520
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.694e-05,
+      "loss": 0.4692,
+      "step": 1530
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.692e-05,
+      "loss": 0.425,
+      "step": 1540
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.69e-05,
+      "loss": 0.4624,
+      "step": 1550
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.688e-05,
+      "loss": 0.4548,
+      "step": 1560
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.686e-05,
+      "loss": 0.3428,
+      "step": 1570
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.684e-05,
+      "loss": 0.4495,
+      "step": 1580
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.682e-05,
+      "loss": 0.4178,
+      "step": 1590
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.6800000000000002e-05,
+      "loss": 0.4716,
+      "step": 1600
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6780000000000002e-05,
+      "loss": 0.5051,
+      "step": 1610
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6760000000000002e-05,
+      "loss": 0.5372,
+      "step": 1620
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6740000000000002e-05,
+      "loss": 0.5413,
+      "step": 1630
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.672e-05,
+      "loss": 0.4294,
+      "step": 1640
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.67e-05,
+      "loss": 0.4341,
+      "step": 1650
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.668e-05,
+      "loss": 0.4515,
+      "step": 1660
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.666e-05,
+      "loss": 0.5709,
+      "step": 1670
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.664e-05,
+      "loss": 0.5482,
+      "step": 1680
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.662e-05,
+      "loss": 0.4269,
+      "step": 1690
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.66e-05,
+      "loss": 0.622,
+      "step": 1700
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.658e-05,
+      "loss": 0.5156,
+      "step": 1710
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.656e-05,
+      "loss": 0.3698,
+      "step": 1720
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.654e-05,
+      "loss": 0.5114,
+      "step": 1730
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.652e-05,
+      "loss": 0.4461,
+      "step": 1740
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.65e-05,
+      "loss": 0.5658,
+      "step": 1750
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.648e-05,
+      "loss": 0.4755,
+      "step": 1760
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.646e-05,
+      "loss": 0.5131,
+      "step": 1770
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6440000000000002e-05,
+      "loss": 0.5374,
+      "step": 1780
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6420000000000002e-05,
+      "loss": 0.5224,
+      "step": 1790
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.64e-05,
+      "loss": 0.4829,
+      "step": 1800
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.638e-05,
+      "loss": 0.359,
+      "step": 1810
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.636e-05,
+      "loss": 0.3834,
+      "step": 1820
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.634e-05,
+      "loss": 0.4865,
+      "step": 1830
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.632e-05,
+      "loss": 0.4626,
+      "step": 1840
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.63e-05,
+      "loss": 0.3687,
+      "step": 1850
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.628e-05,
+      "loss": 0.5426,
+      "step": 1860
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.626e-05,
+      "loss": 0.6008,
+      "step": 1870
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6240000000000004e-05,
+      "loss": 0.5275,
+      "step": 1880
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6220000000000004e-05,
+      "loss": 0.4179,
+      "step": 1890
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.62e-05,
+      "loss": 0.4707,
+      "step": 1900
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.618e-05,
+      "loss": 0.6843,
+      "step": 1910
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.616e-05,
+      "loss": 0.5619,
+      "step": 1920
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.614e-05,
+      "loss": 0.5877,
+      "step": 1930
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.612e-05,
+      "loss": 0.6247,
+      "step": 1940
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6100000000000002e-05,
+      "loss": 0.5433,
+      "step": 1950
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6080000000000002e-05,
+      "loss": 0.4228,
+      "step": 1960
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6060000000000002e-05,
+      "loss": 0.4567,
+      "step": 1970
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6040000000000002e-05,
+      "loss": 0.5629,
+      "step": 1980
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6020000000000002e-05,
+      "loss": 0.4159,
+      "step": 1990
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.5193,
+      "step": 2000
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": 0.8533475026567482,
+      "eval_loss": 0.3346226215362549,
+      "eval_runtime": 12.1352,
+      "eval_samples_per_second": 232.629,
+      "eval_steps_per_second": 29.089,
+      "step": 2000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5980000000000003e-05,
+      "loss": 0.4334,
+      "step": 2010
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5960000000000003e-05,
+      "loss": 0.5265,
+      "step": 2020
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5940000000000003e-05,
+      "loss": 0.4527,
+      "step": 2030
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5920000000000003e-05,
+      "loss": 0.5046,
+      "step": 2040
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5900000000000004e-05,
+      "loss": 0.4549,
+      "step": 2050
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.588e-05,
+      "loss": 0.4136,
+      "step": 2060
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.586e-05,
+      "loss": 0.5247,
+      "step": 2070
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.584e-05,
+      "loss": 0.4806,
+      "step": 2080
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.582e-05,
+      "loss": 0.5402,
+      "step": 2090
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.58e-05,
+      "loss": 0.5339,
+      "step": 2100
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.578e-05,
+      "loss": 0.4736,
+      "step": 2110
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.576e-05,
+      "loss": 0.58,
+      "step": 2120
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5740000000000002e-05,
+      "loss": 0.3891,
+      "step": 2130
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5720000000000002e-05,
+      "loss": 0.5435,
+      "step": 2140
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5700000000000002e-05,
+      "loss": 0.4249,
+      "step": 2150
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5680000000000002e-05,
+      "loss": 0.5014,
+      "step": 2160
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5660000000000003e-05,
+      "loss": 0.4712,
+      "step": 2170
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.5640000000000003e-05,
+      "loss": 0.5119,
+      "step": 2180
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.5620000000000003e-05,
+      "loss": 0.5538,
+      "step": 2190
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.5600000000000003e-05,
+      "loss": 0.5055,
+      "step": 2200
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.5580000000000003e-05,
+      "loss": 0.5375,
+      "step": 2210
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.556e-05,
+      "loss": 0.5395,
+      "step": 2220
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.554e-05,
+      "loss": 0.5556,
+      "step": 2230
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.552e-05,
+      "loss": 0.4772,
+      "step": 2240
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.55e-05,
+      "loss": 0.3976,
+      "step": 2250
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.548e-05,
+      "loss": 0.5703,
+      "step": 2260
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.546e-05,
+      "loss": 0.3762,
+      "step": 2270
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.544e-05,
+      "loss": 0.511,
+      "step": 2280
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.542e-05,
+      "loss": 0.3883,
+      "step": 2290
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.54e-05,
+      "loss": 0.5396,
+      "step": 2300
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.5380000000000002e-05,
+      "loss": 0.4215,
+      "step": 2310
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.5360000000000002e-05,
+      "loss": 0.5234,
+      "step": 2320
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.5340000000000002e-05,
+      "loss": 0.5003,
+      "step": 2330
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.5320000000000002e-05,
+      "loss": 0.5139,
+      "step": 2340
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5300000000000003e-05,
+      "loss": 0.3844,
+      "step": 2350
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5280000000000003e-05,
+      "loss": 0.5582,
+      "step": 2360
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.5260000000000003e-05,
+      "loss": 0.5031,
+      "step": 2370
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.5240000000000001e-05,
+      "loss": 0.5046,
+      "step": 2380
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.5220000000000002e-05,
+      "loss": 0.4748,
+      "step": 2390
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.5200000000000002e-05,
+      "loss": 0.4242,
+      "step": 2400
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.5180000000000002e-05,
+      "loss": 0.4565,
+      "step": 2410
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.516e-05,
+      "loss": 0.4637,
+      "step": 2420
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.514e-05,
+      "loss": 0.5014,
+      "step": 2430
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.5120000000000001e-05,
+      "loss": 0.4742,
+      "step": 2440
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.5100000000000001e-05,
+      "loss": 0.455,
+      "step": 2450
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.5080000000000001e-05,
+      "loss": 0.424,
+      "step": 2460
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.5060000000000001e-05,
+      "loss": 0.4607,
+      "step": 2470
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.5040000000000002e-05,
+      "loss": 0.3498,
+      "step": 2480
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.5020000000000002e-05,
+      "loss": 0.4468,
+      "step": 2490
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.4526,
+      "step": 2500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.498e-05,
+      "loss": 0.3407,
+      "step": 2510
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.496e-05,
+      "loss": 0.5423,
+      "step": 2520
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.4940000000000001e-05,
+      "loss": 0.4411,
+      "step": 2530
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.4920000000000001e-05,
+      "loss": 0.4874,
+      "step": 2540
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.4900000000000001e-05,
+      "loss": 0.5164,
+      "step": 2550
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.4880000000000002e-05,
+      "loss": 0.3663,
+      "step": 2560
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.4860000000000002e-05,
+      "loss": 0.4915,
+      "step": 2570
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.4840000000000002e-05,
+      "loss": 0.4372,
+      "step": 2580
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.482e-05,
+      "loss": 0.3957,
+      "step": 2590
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.48e-05,
+      "loss": 0.3829,
+      "step": 2600
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.478e-05,
+      "loss": 0.4239,
+      "step": 2610
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.4760000000000001e-05,
+      "loss": 0.5329,
+      "step": 2620
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.4740000000000001e-05,
+      "loss": 0.3856,
+      "step": 2630
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.4720000000000001e-05,
+      "loss": 0.385,
+      "step": 2640
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.4700000000000002e-05,
+      "loss": 0.6578,
+      "step": 2650
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.4680000000000002e-05,
+      "loss": 0.4879,
+      "step": 2660
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.466e-05,
+      "loss": 0.4267,
+      "step": 2670
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.464e-05,
+      "loss": 0.4636,
+      "step": 2680
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.462e-05,
+      "loss": 0.4344,
+      "step": 2690
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.46e-05,
+      "loss": 0.4744,
+      "step": 2700
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4580000000000001e-05,
+      "loss": 0.582,
+      "step": 2710
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.4560000000000001e-05,
+      "loss": 0.3262,
+      "step": 2720
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.4540000000000001e-05,
+      "loss": 0.4893,
+      "step": 2730
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.4520000000000002e-05,
+      "loss": 0.3631,
+      "step": 2740
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.45e-05,
+      "loss": 0.5369,
+      "step": 2750
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.448e-05,
+      "loss": 0.4084,
+      "step": 2760
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.446e-05,
+      "loss": 0.3515,
+      "step": 2770
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.444e-05,
+      "loss": 0.3209,
+      "step": 2780
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.4420000000000001e-05,
+      "loss": 0.4764,
+      "step": 2790
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.4400000000000001e-05,
+      "loss": 0.4856,
+      "step": 2800
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.4380000000000001e-05,
+      "loss": 0.5746,
+      "step": 2810
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.4360000000000001e-05,
+      "loss": 0.4593,
+      "step": 2820
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.434e-05,
+      "loss": 0.4143,
+      "step": 2830
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.432e-05,
+      "loss": 0.3721,
+      "step": 2840
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.43e-05,
+      "loss": 0.4414,
+      "step": 2850
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.428e-05,
+      "loss": 0.5248,
+      "step": 2860
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.426e-05,
+      "loss": 0.5316,
+      "step": 2870
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4240000000000001e-05,
+      "loss": 0.3969,
+      "step": 2880
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4220000000000001e-05,
+      "loss": 0.5427,
+      "step": 2890
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4200000000000001e-05,
+      "loss": 0.6245,
+      "step": 2900
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.418e-05,
+      "loss": 0.514,
+      "step": 2910
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.416e-05,
+      "loss": 0.4506,
+      "step": 2920
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.414e-05,
+      "loss": 0.6037,
+      "step": 2930
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.412e-05,
+      "loss": 0.3866,
+      "step": 2940
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.41e-05,
+      "loss": 0.3734,
+      "step": 2950
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.408e-05,
+      "loss": 0.4857,
+      "step": 2960
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4060000000000001e-05,
+      "loss": 0.4482,
+      "step": 2970
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.4040000000000001e-05,
+      "loss": 0.3991,
+      "step": 2980
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.402e-05,
+      "loss": 0.4058,
+      "step": 2990
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.4e-05,
+      "loss": 0.4337,
+      "step": 3000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.398e-05,
+      "loss": 0.3608,
+      "step": 3010
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.396e-05,
+      "loss": 0.3672,
+      "step": 3020
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.394e-05,
+      "loss": 0.4036,
+      "step": 3030
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.392e-05,
+      "loss": 0.3101,
+      "step": 3040
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.39e-05,
+      "loss": 0.4007,
+      "step": 3050
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.3880000000000001e-05,
+      "loss": 0.3762,
+      "step": 3060
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.386e-05,
+      "loss": 0.6522,
+      "step": 3070
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.384e-05,
+      "loss": 0.5718,
+      "step": 3080
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.382e-05,
+      "loss": 0.4189,
+      "step": 3090
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.38e-05,
+      "loss": 0.3717,
+      "step": 3100
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.378e-05,
+      "loss": 0.4456,
+      "step": 3110
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.376e-05,
+      "loss": 0.3815,
+      "step": 3120
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.3740000000000002e-05,
+      "loss": 0.4982,
+      "step": 3130
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.3720000000000002e-05,
+      "loss": 0.3865,
+      "step": 3140
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.3700000000000003e-05,
+      "loss": 0.4284,
+      "step": 3150
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.3680000000000003e-05,
+      "loss": 0.469,
+      "step": 3160
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.3660000000000001e-05,
+      "loss": 0.5024,
+      "step": 3170
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.3640000000000002e-05,
+      "loss": 0.4257,
+      "step": 3180
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.3620000000000002e-05,
+      "loss": 0.4811,
+      "step": 3190
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.3600000000000002e-05,
+      "loss": 0.4264,
+      "step": 3200
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.3580000000000002e-05,
+      "loss": 0.3129,
+      "step": 3210
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.3560000000000002e-05,
+      "loss": 0.4252,
+      "step": 3220
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.3540000000000003e-05,
+      "loss": 0.459,
+      "step": 3230
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.3520000000000003e-05,
+      "loss": 0.459,
+      "step": 3240
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.3500000000000001e-05,
+      "loss": 0.5765,
+      "step": 3250
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.3480000000000001e-05,
+      "loss": 0.5409,
+      "step": 3260
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.3460000000000002e-05,
+      "loss": 0.4468,
+      "step": 3270
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.3440000000000002e-05,
+      "loss": 0.4625,
+      "step": 3280
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3420000000000002e-05,
+      "loss": 0.6015,
+      "step": 3290
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3400000000000002e-05,
+      "loss": 0.3956,
+      "step": 3300
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3380000000000002e-05,
+      "loss": 0.4767,
+      "step": 3310
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3360000000000003e-05,
+      "loss": 0.5715,
+      "step": 3320
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3340000000000001e-05,
+      "loss": 0.4869,
+      "step": 3330
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3320000000000001e-05,
+      "loss": 0.4554,
+      "step": 3340
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3300000000000001e-05,
+      "loss": 0.4013,
+      "step": 3350
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3280000000000002e-05,
+      "loss": 0.3721,
+      "step": 3360
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3260000000000002e-05,
+      "loss": 0.5452,
+      "step": 3370
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3240000000000002e-05,
+      "loss": 0.4286,
+      "step": 3380
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3220000000000002e-05,
+      "loss": 0.3666,
+      "step": 3390
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3200000000000002e-05,
+      "loss": 0.4162,
+      "step": 3400
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3180000000000001e-05,
+      "loss": 0.3919,
+      "step": 3410
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3160000000000001e-05,
+      "loss": 0.4261,
+      "step": 3420
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3140000000000001e-05,
+      "loss": 0.515,
+      "step": 3430
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3120000000000001e-05,
+      "loss": 0.3572,
+      "step": 3440
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3100000000000002e-05,
+      "loss": 0.3408,
+      "step": 3450
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3080000000000002e-05,
+      "loss": 0.4376,
+      "step": 3460
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3060000000000002e-05,
+      "loss": 0.4591,
+      "step": 3470
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3040000000000002e-05,
+      "loss": 0.4289,
+      "step": 3480
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.302e-05,
+      "loss": 0.3451,
+      "step": 3490
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 0.3713,
+      "step": 3500
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2980000000000001e-05,
+      "loss": 0.5562,
+      "step": 3510
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2960000000000001e-05,
+      "loss": 0.4459,
+      "step": 3520
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2940000000000001e-05,
+      "loss": 0.3464,
+      "step": 3530
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2920000000000002e-05,
+      "loss": 0.4756,
+      "step": 3540
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2900000000000002e-05,
+      "loss": 0.5188,
+      "step": 3550
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2880000000000002e-05,
+      "loss": 0.3863,
+      "step": 3560
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.286e-05,
+      "loss": 0.4354,
+      "step": 3570
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.284e-05,
+      "loss": 0.3893,
+      "step": 3580
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2820000000000001e-05,
+      "loss": 0.4097,
+      "step": 3590
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2800000000000001e-05,
+      "loss": 0.3204,
+      "step": 3600
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2780000000000001e-05,
+      "loss": 0.3697,
+      "step": 3610
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2760000000000001e-05,
+      "loss": 0.4393,
+      "step": 3620
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2740000000000002e-05,
+      "loss": 0.508,
+      "step": 3630
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2720000000000002e-05,
+      "loss": 0.3644,
+      "step": 3640
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.27e-05,
+      "loss": 0.3669,
+      "step": 3650
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.268e-05,
+      "loss": 0.5235,
+      "step": 3660
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.266e-05,
+      "loss": 0.4511,
+      "step": 3670
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.2640000000000001e-05,
+      "loss": 0.7004,
+      "step": 3680
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.2620000000000001e-05,
+      "loss": 0.4061,
+      "step": 3690
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.2600000000000001e-05,
+      "loss": 0.4878,
+      "step": 3700
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.2580000000000002e-05,
+      "loss": 0.394,
+      "step": 3710
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.2560000000000002e-05,
+      "loss": 0.443,
+      "step": 3720
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.254e-05,
+      "loss": 0.5277,
+      "step": 3730
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.252e-05,
+      "loss": 0.4001,
+      "step": 3740
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.25e-05,
+      "loss": 0.4226,
+      "step": 3750
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.248e-05,
+      "loss": 0.497,
+      "step": 3760
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.2460000000000001e-05,
+      "loss": 0.4695,
+      "step": 3770
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.2440000000000001e-05,
+      "loss": 0.4494,
+      "step": 3780
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2420000000000001e-05,
+      "loss": 0.4569,
+      "step": 3790
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2400000000000002e-05,
+      "loss": 0.535,
+      "step": 3800
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.2380000000000002e-05,
+      "loss": 0.4378,
+      "step": 3810
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.236e-05,
+      "loss": 0.408,
+      "step": 3820
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.234e-05,
+      "loss": 0.3572,
+      "step": 3830
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.232e-05,
+      "loss": 0.3581,
+      "step": 3840
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.23e-05,
+      "loss": 0.3117,
+      "step": 3850
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.2280000000000001e-05,
+      "loss": 0.4045,
+      "step": 3860
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.2260000000000001e-05,
+      "loss": 0.354,
+      "step": 3870
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.2240000000000001e-05,
+      "loss": 0.4286,
+      "step": 3880
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.2220000000000002e-05,
+      "loss": 0.5221,
+      "step": 3890
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.22e-05,
+      "loss": 0.376,
+      "step": 3900
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.218e-05,
+      "loss": 0.3027,
+      "step": 3910
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.216e-05,
+      "loss": 0.4867,
+      "step": 3920
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.214e-05,
+      "loss": 0.4892,
+      "step": 3930
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2120000000000001e-05,
+      "loss": 0.3886,
+      "step": 3940
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.2100000000000001e-05,
+      "loss": 0.3565,
+      "step": 3950
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.2080000000000001e-05,
+      "loss": 0.3524,
+      "step": 3960
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.2060000000000001e-05,
+      "loss": 0.3861,
+      "step": 3970
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.204e-05,
+      "loss": 0.4929,
+      "step": 3980
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.202e-05,
+      "loss": 0.3813,
+      "step": 3990
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.2e-05,
+      "loss": 0.337,
+      "step": 4000
+    },
+    {
+      "epoch": 2.0,
+      "eval_accuracy": 0.8777895855472901,
+      "eval_loss": 0.28915584087371826,
+      "eval_runtime": 12.0602,
+      "eval_samples_per_second": 234.076,
+      "eval_steps_per_second": 29.27,
+      "step": 4000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.198e-05,
+      "loss": 0.4788,
+      "step": 4010
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.196e-05,
+      "loss": 0.4925,
+      "step": 4020
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.1940000000000001e-05,
+      "loss": 0.4162,
+      "step": 4030
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.1920000000000001e-05,
+      "loss": 0.4735,
+      "step": 4040
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.1900000000000001e-05,
+      "loss": 0.5041,
+      "step": 4050
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.188e-05,
+      "loss": 0.4502,
+      "step": 4060
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.186e-05,
+      "loss": 0.404,
+      "step": 4070
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.184e-05,
+      "loss": 0.41,
+      "step": 4080
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.182e-05,
+      "loss": 0.3369,
+      "step": 4090
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.18e-05,
+      "loss": 0.3297,
+      "step": 4100
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.178e-05,
+      "loss": 0.4311,
+      "step": 4110
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.1760000000000001e-05,
+      "loss": 0.4848,
+      "step": 4120
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.1740000000000001e-05,
+      "loss": 0.4646,
+      "step": 4130
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.172e-05,
+      "loss": 0.2932,
+      "step": 4140
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.17e-05,
+      "loss": 0.5245,
+      "step": 4150
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.168e-05,
+      "loss": 0.4599,
+      "step": 4160
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.166e-05,
+      "loss": 0.3636,
+      "step": 4170
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.164e-05,
+      "loss": 0.4461,
+      "step": 4180
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.162e-05,
+      "loss": 0.4158,
+      "step": 4190
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.16e-05,
+      "loss": 0.4235,
+      "step": 4200
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.1580000000000001e-05,
+      "loss": 0.5746,
+      "step": 4210
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.156e-05,
+      "loss": 0.4843,
+      "step": 4220
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.154e-05,
+      "loss": 0.438,
+      "step": 4230
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.152e-05,
+      "loss": 0.4038,
+      "step": 4240
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.15e-05,
+      "loss": 0.3282,
+      "step": 4250
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.148e-05,
+      "loss": 0.3476,
+      "step": 4260
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.146e-05,
+      "loss": 0.5096,
+      "step": 4270
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.144e-05,
+      "loss": 0.4123,
+      "step": 4280
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.142e-05,
+      "loss": 0.3776,
+      "step": 4290
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.14e-05,
+      "loss": 0.4905,
+      "step": 4300
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.138e-05,
+      "loss": 0.4365,
+      "step": 4310
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.136e-05,
+      "loss": 0.3964,
+      "step": 4320
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.134e-05,
+      "loss": 0.4531,
+      "step": 4330
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.132e-05,
+      "loss": 0.4326,
+      "step": 4340
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.13e-05,
+      "loss": 0.3562,
+      "step": 4350
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.128e-05,
+      "loss": 0.3907,
+      "step": 4360
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.126e-05,
+      "loss": 0.4127,
+      "step": 4370
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.1240000000000002e-05,
+      "loss": 0.4155,
+      "step": 4380
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.1220000000000003e-05,
+      "loss": 0.4167,
+      "step": 4390
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.1200000000000001e-05,
+      "loss": 0.3452,
+      "step": 4400
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.1180000000000001e-05,
+      "loss": 0.3245,
+      "step": 4410
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.1160000000000002e-05,
+      "loss": 0.3444,
+      "step": 4420
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.1140000000000002e-05,
+      "loss": 0.4127,
+      "step": 4430
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.1120000000000002e-05,
+      "loss": 0.3968,
+      "step": 4440
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.1100000000000002e-05,
+      "loss": 0.3726,
+      "step": 4450
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.1080000000000002e-05,
+      "loss": 0.3779,
+      "step": 4460
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.1060000000000003e-05,
+      "loss": 0.6474,
+      "step": 4470
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.1040000000000001e-05,
+      "loss": 0.3387,
+      "step": 4480
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.1020000000000001e-05,
+      "loss": 0.5002,
+      "step": 4490
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 0.3347,
+      "step": 4500
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.0980000000000002e-05,
+      "loss": 0.4875,
+      "step": 4510
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.0960000000000002e-05,
+      "loss": 0.4971,
+      "step": 4520
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.0940000000000002e-05,
+      "loss": 0.3277,
+      "step": 4530
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.0920000000000002e-05,
+      "loss": 0.4733,
+      "step": 4540
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.0900000000000002e-05,
+      "loss": 0.4148,
+      "step": 4550
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.0880000000000001e-05,
+      "loss": 0.4388,
+      "step": 4560
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.0860000000000001e-05,
+      "loss": 0.4897,
+      "step": 4570
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.0840000000000001e-05,
+      "loss": 0.5448,
+      "step": 4580
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.0820000000000001e-05,
+      "loss": 0.3224,
+      "step": 4590
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.0800000000000002e-05,
+      "loss": 0.5522,
+      "step": 4600
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.0780000000000002e-05,
+      "loss": 0.3744,
+      "step": 4610
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.0760000000000002e-05,
+      "loss": 0.4594,
+      "step": 4620
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.0740000000000002e-05,
+      "loss": 0.4284,
+      "step": 4630
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.072e-05,
+      "loss": 0.4829,
+      "step": 4640
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.0700000000000001e-05,
+      "loss": 0.3484,
+      "step": 4650
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.0680000000000001e-05,
+      "loss": 0.3768,
+      "step": 4660
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.0660000000000001e-05,
+      "loss": 0.4383,
+      "step": 4670
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0640000000000001e-05,
+      "loss": 0.4789,
+      "step": 4680
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0620000000000002e-05,
+      "loss": 0.2911,
+      "step": 4690
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0600000000000002e-05,
+      "loss": 0.5925,
+      "step": 4700
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0580000000000002e-05,
+      "loss": 0.3896,
+      "step": 4710
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.056e-05,
+      "loss": 0.4255,
+      "step": 4720
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.054e-05,
+      "loss": 0.4014,
+      "step": 4730
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0520000000000001e-05,
+      "loss": 0.3283,
+      "step": 4740
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0500000000000001e-05,
+      "loss": 0.4424,
+      "step": 4750
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0480000000000001e-05,
+      "loss": 0.3849,
+      "step": 4760
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0460000000000001e-05,
+      "loss": 0.4048,
+      "step": 4770
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0440000000000002e-05,
+      "loss": 0.2175,
+      "step": 4780
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0420000000000002e-05,
+      "loss": 0.538,
+      "step": 4790
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.04e-05,
+      "loss": 0.3498,
+      "step": 4800
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.038e-05,
+      "loss": 0.5187,
+      "step": 4810
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.036e-05,
+      "loss": 0.4999,
+      "step": 4820
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0340000000000001e-05,
+      "loss": 0.4838,
+      "step": 4830
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0320000000000001e-05,
+      "loss": 0.3557,
+      "step": 4840
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0300000000000001e-05,
+      "loss": 0.4256,
+      "step": 4850
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0280000000000002e-05,
+      "loss": 0.3425,
+      "step": 4860
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0260000000000002e-05,
+      "loss": 0.3676,
+      "step": 4870
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.024e-05,
+      "loss": 0.3826,
+      "step": 4880
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.022e-05,
+      "loss": 0.3211,
+      "step": 4890
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.02e-05,
+      "loss": 0.3308,
+      "step": 4900
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.018e-05,
+      "loss": 0.4076,
+      "step": 4910
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0160000000000001e-05,
+      "loss": 0.4811,
+      "step": 4920
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0140000000000001e-05,
+      "loss": 0.2422,
+      "step": 4930
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0120000000000001e-05,
+      "loss": 0.3318,
+      "step": 4940
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0100000000000002e-05,
+      "loss": 0.4771,
+      "step": 4950
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.008e-05,
+      "loss": 0.5033,
+      "step": 4960
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.006e-05,
+      "loss": 0.5547,
+      "step": 4970
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.004e-05,
+      "loss": 0.3199,
+      "step": 4980
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.002e-05,
+      "loss": 0.4568,
+      "step": 4990
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1e-05,
+      "loss": 0.2876,
+      "step": 5000
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.980000000000001e-06,
+      "loss": 0.4552,
+      "step": 5010
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.960000000000001e-06,
+      "loss": 0.4002,
+      "step": 5020
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.940000000000001e-06,
+      "loss": 0.4293,
+      "step": 5030
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.920000000000002e-06,
+      "loss": 0.3541,
+      "step": 5040
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.9e-06,
+      "loss": 0.4607,
+      "step": 5050
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.88e-06,
+      "loss": 0.393,
+      "step": 5060
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.86e-06,
+      "loss": 0.3946,
+      "step": 5070
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.84e-06,
+      "loss": 0.4723,
+      "step": 5080
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.820000000000001e-06,
+      "loss": 0.4753,
+      "step": 5090
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.800000000000001e-06,
+      "loss": 0.4871,
+      "step": 5100
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.780000000000001e-06,
+      "loss": 0.3223,
+      "step": 5110
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.760000000000001e-06,
+      "loss": 0.3411,
+      "step": 5120
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.74e-06,
+      "loss": 0.3945,
+      "step": 5130
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.72e-06,
+      "loss": 0.4348,
+      "step": 5140
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.7e-06,
+      "loss": 0.4065,
+      "step": 5150
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.68e-06,
+      "loss": 0.4037,
+      "step": 5160
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.66e-06,
+      "loss": 0.4085,
+      "step": 5170
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.640000000000001e-06,
+      "loss": 0.4172,
+      "step": 5180
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.620000000000001e-06,
+      "loss": 0.3582,
+      "step": 5190
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 0.3995,
+      "step": 5200
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.58e-06,
+      "loss": 0.2695,
+      "step": 5210
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.56e-06,
+      "loss": 0.3823,
+      "step": 5220
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.54e-06,
+      "loss": 0.3865,
+      "step": 5230
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.52e-06,
+      "loss": 0.5069,
+      "step": 5240
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.5e-06,
+      "loss": 0.478,
+      "step": 5250
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.48e-06,
+      "loss": 0.4872,
+      "step": 5260
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.460000000000001e-06,
+      "loss": 0.315,
+      "step": 5270
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.440000000000001e-06,
+      "loss": 0.4556,
+      "step": 5280
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.42e-06,
+      "loss": 0.2827,
+      "step": 5290
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.4e-06,
+      "loss": 0.3753,
+      "step": 5300
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.38e-06,
+      "loss": 0.3867,
+      "step": 5310
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.360000000000002e-06,
+      "loss": 0.3859,
+      "step": 5320
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.340000000000002e-06,
+      "loss": 0.3779,
+      "step": 5330
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.32e-06,
+      "loss": 0.3769,
+      "step": 5340
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.3e-06,
+      "loss": 0.4974,
+      "step": 5350
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.280000000000001e-06,
+      "loss": 0.4267,
+      "step": 5360
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.260000000000001e-06,
+      "loss": 0.4678,
+      "step": 5370
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.240000000000001e-06,
+      "loss": 0.4145,
+      "step": 5380
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.220000000000002e-06,
+      "loss": 0.3775,
+      "step": 5390
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.200000000000002e-06,
+      "loss": 0.254,
+      "step": 5400
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.180000000000002e-06,
+      "loss": 0.4195,
+      "step": 5410
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.16e-06,
+      "loss": 0.4089,
+      "step": 5420
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.14e-06,
+      "loss": 0.2697,
+      "step": 5430
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.12e-06,
+      "loss": 0.4161,
+      "step": 5440
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.100000000000001e-06,
+      "loss": 0.2266,
+      "step": 5450
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.080000000000001e-06,
+      "loss": 0.3939,
+      "step": 5460
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.060000000000001e-06,
+      "loss": 0.3836,
+      "step": 5470
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.040000000000002e-06,
+      "loss": 0.3154,
+      "step": 5480
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.020000000000002e-06,
+      "loss": 0.4146,
+      "step": 5490
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9e-06,
+      "loss": 0.4668,
+      "step": 5500
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.98e-06,
+      "loss": 0.377,
+      "step": 5510
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.96e-06,
+      "loss": 0.4411,
+      "step": 5520
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.94e-06,
+      "loss": 0.3943,
+      "step": 5530
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.920000000000001e-06,
+      "loss": 0.2114,
+      "step": 5540
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.900000000000001e-06,
+      "loss": 0.4031,
+      "step": 5550
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.880000000000001e-06,
+      "loss": 0.4905,
+      "step": 5560
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.860000000000002e-06,
+      "loss": 0.4883,
+      "step": 5570
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.84e-06,
+      "loss": 0.4772,
+      "step": 5580
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.82e-06,
+      "loss": 0.4445,
+      "step": 5590
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.8e-06,
+      "loss": 0.4179,
+      "step": 5600
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.78e-06,
+      "loss": 0.4053,
+      "step": 5610
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.76e-06,
+      "loss": 0.367,
+      "step": 5620
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.740000000000001e-06,
+      "loss": 0.4299,
+      "step": 5630
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.720000000000001e-06,
+      "loss": 0.3171,
+      "step": 5640
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.700000000000001e-06,
+      "loss": 0.5525,
+      "step": 5650
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.68e-06,
+      "loss": 0.3091,
+      "step": 5660
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.66e-06,
+      "loss": 0.4636,
+      "step": 5670
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.64e-06,
+      "loss": 0.3944,
+      "step": 5680
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.62e-06,
+      "loss": 0.3552,
+      "step": 5690
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.6e-06,
+      "loss": 0.4895,
+      "step": 5700
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.580000000000001e-06,
+      "loss": 0.2151,
+      "step": 5710
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.560000000000001e-06,
+      "loss": 0.2243,
+      "step": 5720
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.540000000000001e-06,
+      "loss": 0.2747,
+      "step": 5730
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.52e-06,
+      "loss": 0.3228,
+      "step": 5740
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.5e-06,
+      "loss": 0.3405,
+      "step": 5750
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.48e-06,
+      "loss": 0.3393,
+      "step": 5760
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.46e-06,
+      "loss": 0.496,
+      "step": 5770
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.44e-06,
+      "loss": 0.4909,
+      "step": 5780
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 8.42e-06,
+      "loss": 0.3902,
+      "step": 5790
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 0.1537,
+      "step": 5800
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.380000000000001e-06,
+      "loss": 0.3553,
+      "step": 5810
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.36e-06,
+      "loss": 0.3674,
+      "step": 5820
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.34e-06,
+      "loss": 0.2641,
+      "step": 5830
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.32e-06,
+      "loss": 0.4141,
+      "step": 5840
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.3e-06,
+      "loss": 0.4156,
+      "step": 5850
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.28e-06,
+      "loss": 0.3066,
+      "step": 5860
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.26e-06,
+      "loss": 0.3093,
+      "step": 5870
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.24e-06,
+      "loss": 0.3933,
+      "step": 5880
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.220000000000001e-06,
+      "loss": 0.4485,
+      "step": 5890
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.2e-06,
+      "loss": 0.3116,
+      "step": 5900
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.18e-06,
+      "loss": 0.3577,
+      "step": 5910
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.16e-06,
+      "loss": 0.3518,
+      "step": 5920
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.14e-06,
+      "loss": 0.4512,
+      "step": 5930
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.120000000000002e-06,
+      "loss": 0.4969,
+      "step": 5940
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.1e-06,
+      "loss": 0.6094,
+      "step": 5950
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.08e-06,
+      "loss": 0.3628,
+      "step": 5960
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.06e-06,
+      "loss": 0.4789,
+      "step": 5970
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.040000000000001e-06,
+      "loss": 0.3781,
+      "step": 5980
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.020000000000001e-06,
+      "loss": 0.3777,
+      "step": 5990
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.3771,
+      "step": 6000
+    },
+    {
+      "epoch": 3.0,
+      "eval_accuracy": 0.8969181721572795,
+      "eval_loss": 0.2492985874414444,
+      "eval_runtime": 12.192,
+      "eval_samples_per_second": 231.546,
+      "eval_steps_per_second": 28.953,
+      "step": 6000
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.980000000000002e-06,
+      "loss": 0.3267,
+      "step": 6010
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.960000000000002e-06,
+      "loss": 0.363,
+      "step": 6020
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.94e-06,
+      "loss": 0.3494,
+      "step": 6030
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.92e-06,
+      "loss": 0.2584,
+      "step": 6040
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.9e-06,
+      "loss": 0.2958,
+      "step": 6050
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.88e-06,
+      "loss": 0.3289,
+      "step": 6060
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.860000000000001e-06,
+      "loss": 0.3998,
+      "step": 6070
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.840000000000001e-06,
+      "loss": 0.4226,
+      "step": 6080
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.820000000000001e-06,
+      "loss": 0.5205,
+      "step": 6090
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.800000000000002e-06,
+      "loss": 0.2656,
+      "step": 6100
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 7.78e-06,
+      "loss": 0.3355,
+      "step": 6110
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 7.76e-06,
+      "loss": 0.3279,
+      "step": 6120
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 7.74e-06,
+      "loss": 0.2957,
+      "step": 6130
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 7.72e-06,
+      "loss": 0.3718,
+      "step": 6140
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 7.7e-06,
+      "loss": 0.3685,
+      "step": 6150
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 7.680000000000001e-06,
+      "loss": 0.4784,
+      "step": 6160
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 7.660000000000001e-06,
+      "loss": 0.2879,
+      "step": 6170
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 7.640000000000001e-06,
+      "loss": 0.2827,
+      "step": 6180
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 7.620000000000001e-06,
+      "loss": 0.4386,
+      "step": 6190
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 7.600000000000001e-06,
+      "loss": 0.436,
+      "step": 6200
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 7.58e-06,
+      "loss": 0.3106,
+      "step": 6210
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 7.5600000000000005e-06,
+      "loss": 0.3643,
+      "step": 6220
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 7.540000000000001e-06,
+      "loss": 0.4393,
+      "step": 6230
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 7.520000000000001e-06,
+      "loss": 0.3892,
+      "step": 6240
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 0.5483,
+      "step": 6250
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 7.48e-06,
+      "loss": 0.434,
+      "step": 6260
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 7.4600000000000006e-06,
+      "loss": 0.3603,
+      "step": 6270
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 7.440000000000001e-06,
+      "loss": 0.4386,
+      "step": 6280
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 7.420000000000001e-06,
+      "loss": 0.4268,
+      "step": 6290
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 7.4e-06,
+      "loss": 0.432,
+      "step": 6300
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 7.3800000000000005e-06,
+      "loss": 0.3501,
+      "step": 6310
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 7.360000000000001e-06,
+      "loss": 0.3151,
+      "step": 6320
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 7.340000000000001e-06,
+      "loss": 0.4465,
+      "step": 6330
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 7.32e-06,
+      "loss": 0.4318,
+      "step": 6340
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 7.3e-06,
+      "loss": 0.337,
+      "step": 6350
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 7.280000000000001e-06,
+      "loss": 0.3532,
+      "step": 6360
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 7.260000000000001e-06,
+      "loss": 0.3917,
+      "step": 6370
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 7.24e-06,
+      "loss": 0.4249,
+      "step": 6380
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 7.22e-06,
+      "loss": 0.3245,
+      "step": 6390
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 7.2000000000000005e-06,
+      "loss": 0.3392,
+      "step": 6400
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 7.180000000000001e-06,
+      "loss": 0.3983,
+      "step": 6410
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 7.16e-06,
+      "loss": 0.28,
+      "step": 6420
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 7.14e-06,
+      "loss": 0.4035,
+      "step": 6430
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 7.1200000000000004e-06,
+      "loss": 0.3861,
+      "step": 6440
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 7.100000000000001e-06,
+      "loss": 0.6548,
+      "step": 6450
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 7.08e-06,
+      "loss": 0.4171,
+      "step": 6460
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 7.06e-06,
+      "loss": 0.4112,
+      "step": 6470
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 7.04e-06,
+      "loss": 0.2889,
+      "step": 6480
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 7.0200000000000006e-06,
+      "loss": 0.3209,
+      "step": 6490
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 7e-06,
+      "loss": 0.4533,
+      "step": 6500
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 6.98e-06,
+      "loss": 0.4109,
+      "step": 6510
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.96e-06,
+      "loss": 0.4078,
+      "step": 6520
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.9400000000000005e-06,
+      "loss": 0.4347,
+      "step": 6530
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.92e-06,
+      "loss": 0.3825,
+      "step": 6540
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.9e-06,
+      "loss": 0.3999,
+      "step": 6550
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.88e-06,
+      "loss": 0.3498,
+      "step": 6560
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.860000000000001e-06,
+      "loss": 0.2962,
+      "step": 6570
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.8400000000000014e-06,
+      "loss": 0.4853,
+      "step": 6580
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.820000000000001e-06,
+      "loss": 0.4502,
+      "step": 6590
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.800000000000001e-06,
+      "loss": 0.4225,
+      "step": 6600
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.780000000000001e-06,
+      "loss": 0.3751,
+      "step": 6610
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.760000000000001e-06,
+      "loss": 0.3618,
+      "step": 6620
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.740000000000001e-06,
+      "loss": 0.2274,
+      "step": 6630
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.720000000000001e-06,
+      "loss": 0.3132,
+      "step": 6640
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.700000000000001e-06,
+      "loss": 0.2964,
+      "step": 6650
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.680000000000001e-06,
+      "loss": 0.4183,
+      "step": 6660
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.660000000000001e-06,
+      "loss": 0.3342,
+      "step": 6670
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.640000000000001e-06,
+      "loss": 0.4125,
+      "step": 6680
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.620000000000001e-06,
+      "loss": 0.4458,
+      "step": 6690
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.600000000000001e-06,
+      "loss": 0.338,
+      "step": 6700
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.5800000000000005e-06,
+      "loss": 0.2804,
+      "step": 6710
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.560000000000001e-06,
+      "loss": 0.2932,
+      "step": 6720
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.540000000000001e-06,
+      "loss": 0.3589,
+      "step": 6730
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.520000000000001e-06,
+      "loss": 0.361,
+      "step": 6740
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.5000000000000004e-06,
+      "loss": 0.3964,
+      "step": 6750
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.480000000000001e-06,
+      "loss": 0.3585,
+      "step": 6760
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.460000000000001e-06,
+      "loss": 0.3569,
+      "step": 6770
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.440000000000001e-06,
+      "loss": 0.3413,
+      "step": 6780
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.42e-06,
+      "loss": 0.3815,
+      "step": 6790
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 0.3529,
+      "step": 6800
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.380000000000001e-06,
+      "loss": 0.2655,
+      "step": 6810
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.360000000000001e-06,
+      "loss": 0.3011,
+      "step": 6820
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.34e-06,
+      "loss": 0.2936,
+      "step": 6830
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.3200000000000005e-06,
+      "loss": 0.2918,
+      "step": 6840
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.300000000000001e-06,
+      "loss": 0.432,
+      "step": 6850
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.280000000000001e-06,
+      "loss": 0.4691,
+      "step": 6860
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.26e-06,
+      "loss": 0.4513,
+      "step": 6870
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.24e-06,
+      "loss": 0.4222,
+      "step": 6880
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.220000000000001e-06,
+      "loss": 0.3333,
+      "step": 6890
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.200000000000001e-06,
+      "loss": 0.2511,
+      "step": 6900
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.18e-06,
+      "loss": 0.3171,
+      "step": 6910
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.16e-06,
+      "loss": 0.4494,
+      "step": 6920
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.1400000000000005e-06,
+      "loss": 0.2583,
+      "step": 6930
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.120000000000001e-06,
+      "loss": 0.3131,
+      "step": 6940
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.1e-06,
+      "loss": 0.551,
+      "step": 6950
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.08e-06,
+      "loss": 0.2501,
+      "step": 6960
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.0600000000000004e-06,
+      "loss": 0.3497,
+      "step": 6970
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.040000000000001e-06,
+      "loss": 0.5224,
+      "step": 6980
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.02e-06,
+      "loss": 0.3128,
+      "step": 6990
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6e-06,
+      "loss": 0.3826,
+      "step": 7000
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 5.98e-06,
+      "loss": 0.3125,
+      "step": 7010
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 5.9600000000000005e-06,
+      "loss": 0.4111,
+      "step": 7020
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5.94e-06,
+      "loss": 0.2608,
+      "step": 7030
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5.92e-06,
+      "loss": 0.3991,
+      "step": 7040
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5.9e-06,
+      "loss": 0.3986,
+      "step": 7050
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.8800000000000005e-06,
+      "loss": 0.3782,
+      "step": 7060
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.86e-06,
+      "loss": 0.4044,
+      "step": 7070
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.84e-06,
+      "loss": 0.4661,
+      "step": 7080
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.82e-06,
+      "loss": 0.3813,
+      "step": 7090
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.8e-06,
+      "loss": 0.426,
+      "step": 7100
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.78e-06,
+      "loss": 0.2647,
+      "step": 7110
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.76e-06,
+      "loss": 0.4087,
+      "step": 7120
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.74e-06,
+      "loss": 0.3746,
+      "step": 7130
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.72e-06,
+      "loss": 0.4835,
+      "step": 7140
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.7e-06,
+      "loss": 0.2269,
+      "step": 7150
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.68e-06,
+      "loss": 0.2984,
+      "step": 7160
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.66e-06,
+      "loss": 0.3343,
+      "step": 7170
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.64e-06,
+      "loss": 0.5323,
+      "step": 7180
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.620000000000001e-06,
+      "loss": 0.4122,
+      "step": 7190
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 0.4582,
+      "step": 7200
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.580000000000001e-06,
+      "loss": 0.308,
+      "step": 7210
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.560000000000001e-06,
+      "loss": 0.3218,
+      "step": 7220
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.540000000000001e-06,
+      "loss": 0.4305,
+      "step": 7230
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.5200000000000005e-06,
+      "loss": 0.3995,
+      "step": 7240
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.500000000000001e-06,
+      "loss": 0.3022,
+      "step": 7250
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.480000000000001e-06,
+      "loss": 0.4294,
+      "step": 7260
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.460000000000001e-06,
+      "loss": 0.3508,
+      "step": 7270
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.4400000000000004e-06,
+      "loss": 0.5502,
+      "step": 7280
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.420000000000001e-06,
+      "loss": 0.3474,
+      "step": 7290
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.400000000000001e-06,
+      "loss": 0.4551,
+      "step": 7300
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.380000000000001e-06,
+      "loss": 0.376,
+      "step": 7310
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.36e-06,
+      "loss": 0.3818,
+      "step": 7320
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.3400000000000005e-06,
+      "loss": 0.4401,
+      "step": 7330
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.320000000000001e-06,
+      "loss": 0.2996,
+      "step": 7340
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.300000000000001e-06,
+      "loss": 0.4456,
+      "step": 7350
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 5.28e-06,
+      "loss": 0.3195,
+      "step": 7360
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 5.2600000000000005e-06,
+      "loss": 0.4861,
+      "step": 7370
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 5.240000000000001e-06,
+      "loss": 0.4056,
+      "step": 7380
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 5.220000000000001e-06,
+      "loss": 0.3041,
+      "step": 7390
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5.2e-06,
+      "loss": 0.3005,
+      "step": 7400
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 5.18e-06,
+      "loss": 0.2771,
+      "step": 7410
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 5.1600000000000006e-06,
+      "loss": 0.2574,
+      "step": 7420
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 5.140000000000001e-06,
+      "loss": 0.3663,
+      "step": 7430
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 5.12e-06,
+      "loss": 0.3586,
+      "step": 7440
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 5.1e-06,
+      "loss": 0.2552,
+      "step": 7450
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 5.0800000000000005e-06,
+      "loss": 0.5182,
+      "step": 7460
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 5.060000000000001e-06,
+      "loss": 0.3337,
+      "step": 7470
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 5.04e-06,
+      "loss": 0.3506,
+      "step": 7480
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 5.02e-06,
+      "loss": 0.3552,
+      "step": 7490
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 5e-06,
+      "loss": 0.3652,
+      "step": 7500
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.980000000000001e-06,
+      "loss": 0.2553,
+      "step": 7510
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.960000000000001e-06,
+      "loss": 0.2788,
+      "step": 7520
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.94e-06,
+      "loss": 0.3503,
+      "step": 7530
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.92e-06,
+      "loss": 0.429,
+      "step": 7540
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.9000000000000005e-06,
+      "loss": 0.2884,
+      "step": 7550
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.880000000000001e-06,
+      "loss": 0.4194,
+      "step": 7560
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.86e-06,
+      "loss": 0.2595,
+      "step": 7570
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.84e-06,
+      "loss": 0.3946,
+      "step": 7580
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.8200000000000004e-06,
+      "loss": 0.3062,
+      "step": 7590
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 0.3616,
+      "step": 7600
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.78e-06,
+      "loss": 0.3569,
+      "step": 7610
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.76e-06,
+      "loss": 0.3033,
+      "step": 7620
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.74e-06,
+      "loss": 0.3695,
+      "step": 7630
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.7200000000000005e-06,
+      "loss": 0.2913,
+      "step": 7640
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.7e-06,
+      "loss": 0.2956,
+      "step": 7650
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.680000000000001e-06,
+      "loss": 0.4334,
+      "step": 7660
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.66e-06,
+      "loss": 0.429,
+      "step": 7670
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.6400000000000005e-06,
+      "loss": 0.2591,
+      "step": 7680
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.620000000000001e-06,
+      "loss": 0.3912,
+      "step": 7690
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.600000000000001e-06,
+      "loss": 0.5294,
+      "step": 7700
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.58e-06,
+      "loss": 0.3006,
+      "step": 7710
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.56e-06,
+      "loss": 0.2757,
+      "step": 7720
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.540000000000001e-06,
+      "loss": 0.3146,
+      "step": 7730
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.520000000000001e-06,
+      "loss": 0.4105,
+      "step": 7740
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.5e-06,
+      "loss": 0.4373,
+      "step": 7750
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.48e-06,
+      "loss": 0.3716,
+      "step": 7760
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.4600000000000005e-06,
+      "loss": 0.5191,
+      "step": 7770
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.440000000000001e-06,
+      "loss": 0.2644,
+      "step": 7780
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.42e-06,
+      "loss": 0.3725,
+      "step": 7790
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.4e-06,
+      "loss": 0.4628,
+      "step": 7800
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.38e-06,
+      "loss": 0.393,
+      "step": 7810
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.360000000000001e-06,
+      "loss": 0.3997,
+      "step": 7820
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.34e-06,
+      "loss": 0.4044,
+      "step": 7830
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.32e-06,
+      "loss": 0.4409,
+      "step": 7840
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.3e-06,
+      "loss": 0.3767,
+      "step": 7850
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.2800000000000005e-06,
+      "loss": 0.3694,
+      "step": 7860
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.26e-06,
+      "loss": 0.3515,
+      "step": 7870
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.24e-06,
+      "loss": 0.3994,
+      "step": 7880
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.22e-06,
+      "loss": 0.333,
+      "step": 7890
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 0.2682,
+      "step": 7900
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.18e-06,
+      "loss": 0.3332,
+      "step": 7910
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.16e-06,
+      "loss": 0.3456,
+      "step": 7920
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.14e-06,
+      "loss": 0.4017,
+      "step": 7930
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.12e-06,
+      "loss": 0.2923,
+      "step": 7940
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.1e-06,
+      "loss": 0.3038,
+      "step": 7950
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.08e-06,
+      "loss": 0.3672,
+      "step": 7960
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.060000000000001e-06,
+      "loss": 0.4095,
+      "step": 7970
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.04e-06,
+      "loss": 0.376,
+      "step": 7980
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.0200000000000005e-06,
+      "loss": 0.3436,
+      "step": 7990
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.3819,
+      "step": 8000
+    },
+    {
+      "epoch": 4.0,
+      "eval_accuracy": 0.9100247963159759,
+      "eval_loss": 0.22746825218200684,
+      "eval_runtime": 11.9848,
+      "eval_samples_per_second": 235.548,
+      "eval_steps_per_second": 29.454,
+      "step": 8000
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.980000000000001e-06,
+      "loss": 0.421,
+      "step": 8010
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.96e-06,
+      "loss": 0.3546,
+      "step": 8020
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.94e-06,
+      "loss": 0.3224,
+      "step": 8030
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.920000000000001e-06,
+      "loss": 0.2738,
+      "step": 8040
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.900000000000001e-06,
+      "loss": 0.3581,
+      "step": 8050
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.88e-06,
+      "loss": 0.2254,
+      "step": 8060
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.86e-06,
+      "loss": 0.3602,
+      "step": 8070
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.8400000000000005e-06,
+      "loss": 0.3515,
+      "step": 8080
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.820000000000001e-06,
+      "loss": 0.3209,
+      "step": 8090
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.8000000000000005e-06,
+      "loss": 0.3708,
+      "step": 8100
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.7800000000000002e-06,
+      "loss": 0.3035,
+      "step": 8110
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.7600000000000004e-06,
+      "loss": 0.3209,
+      "step": 8120
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.74e-06,
+      "loss": 0.3928,
+      "step": 8130
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.7200000000000004e-06,
+      "loss": 0.3642,
+      "step": 8140
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.7e-06,
+      "loss": 0.3093,
+      "step": 8150
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.6800000000000003e-06,
+      "loss": 0.3347,
+      "step": 8160
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.66e-06,
+      "loss": 0.2622,
+      "step": 8170
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.6400000000000003e-06,
+      "loss": 0.2705,
+      "step": 8180
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.62e-06,
+      "loss": 0.2752,
+      "step": 8190
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.6000000000000003e-06,
+      "loss": 0.331,
+      "step": 8200
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.58e-06,
+      "loss": 0.3249,
+      "step": 8210
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.5600000000000002e-06,
+      "loss": 0.4114,
+      "step": 8220
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.54e-06,
+      "loss": 0.4724,
+      "step": 8230
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.52e-06,
+      "loss": 0.3165,
+      "step": 8240
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.5e-06,
+      "loss": 0.3408,
+      "step": 8250
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.48e-06,
+      "loss": 0.5024,
+      "step": 8260
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.46e-06,
+      "loss": 0.3348,
+      "step": 8270
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.44e-06,
+      "loss": 0.3261,
+      "step": 8280
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.4200000000000007e-06,
+      "loss": 0.2399,
+      "step": 8290
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.4000000000000005e-06,
+      "loss": 0.3624,
+      "step": 8300
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.3800000000000007e-06,
+      "loss": 0.5088,
+      "step": 8310
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.3600000000000004e-06,
+      "loss": 0.4345,
+      "step": 8320
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.3400000000000006e-06,
+      "loss": 0.3422,
+      "step": 8330
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.3200000000000004e-06,
+      "loss": 0.3658,
+      "step": 8340
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.3000000000000006e-06,
+      "loss": 0.4656,
+      "step": 8350
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.2800000000000004e-06,
+      "loss": 0.2924,
+      "step": 8360
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.2600000000000006e-06,
+      "loss": 0.308,
+      "step": 8370
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.2400000000000003e-06,
+      "loss": 0.4511,
+      "step": 8380
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.2200000000000005e-06,
+      "loss": 0.3178,
+      "step": 8390
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 0.5311,
+      "step": 8400
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.1800000000000005e-06,
+      "loss": 0.426,
+      "step": 8410
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.1600000000000002e-06,
+      "loss": 0.3695,
+      "step": 8420
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.1400000000000004e-06,
+      "loss": 0.3946,
+      "step": 8430
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.12e-06,
+      "loss": 0.2517,
+      "step": 8440
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.1000000000000004e-06,
+      "loss": 0.4627,
+      "step": 8450
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.08e-06,
+      "loss": 0.2384,
+      "step": 8460
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.0600000000000003e-06,
+      "loss": 0.2364,
+      "step": 8470
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.04e-06,
+      "loss": 0.315,
+      "step": 8480
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 3.0200000000000003e-06,
+      "loss": 0.327,
+      "step": 8490
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 3e-06,
+      "loss": 0.3416,
+      "step": 8500
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.9800000000000003e-06,
+      "loss": 0.4286,
+      "step": 8510
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.96e-06,
+      "loss": 0.2526,
+      "step": 8520
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.9400000000000002e-06,
+      "loss": 0.2432,
+      "step": 8530
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.92e-06,
+      "loss": 0.3154,
+      "step": 8540
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.9e-06,
+      "loss": 0.1894,
+      "step": 8550
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.88e-06,
+      "loss": 0.3152,
+      "step": 8560
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.86e-06,
+      "loss": 0.4447,
+      "step": 8570
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.84e-06,
+      "loss": 0.283,
+      "step": 8580
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.82e-06,
+      "loss": 0.3148,
+      "step": 8590
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 0.4559,
+      "step": 8600
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.7800000000000005e-06,
+      "loss": 0.2617,
+      "step": 8610
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.7600000000000003e-06,
+      "loss": 0.3346,
+      "step": 8620
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.7400000000000004e-06,
+      "loss": 0.2664,
+      "step": 8630
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.7200000000000002e-06,
+      "loss": 0.3194,
+      "step": 8640
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.7000000000000004e-06,
+      "loss": 0.3718,
+      "step": 8650
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.68e-06,
+      "loss": 0.2523,
+      "step": 8660
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.6600000000000004e-06,
+      "loss": 0.333,
+      "step": 8670
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.64e-06,
+      "loss": 0.5717,
+      "step": 8680
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.6200000000000003e-06,
+      "loss": 0.3764,
+      "step": 8690
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.6e-06,
+      "loss": 0.4653,
+      "step": 8700
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.5800000000000003e-06,
+      "loss": 0.4164,
+      "step": 8710
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.56e-06,
+      "loss": 0.4843,
+      "step": 8720
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.5400000000000002e-06,
+      "loss": 0.2491,
+      "step": 8730
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.52e-06,
+      "loss": 0.4553,
+      "step": 8740
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.5e-06,
+      "loss": 0.4461,
+      "step": 8750
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.4800000000000004e-06,
+      "loss": 0.457,
+      "step": 8760
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.46e-06,
+      "loss": 0.3921,
+      "step": 8770
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.4400000000000004e-06,
+      "loss": 0.5787,
+      "step": 8780
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.42e-06,
+      "loss": 0.4045,
+      "step": 8790
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 0.2649,
+      "step": 8800
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.38e-06,
+      "loss": 0.3951,
+      "step": 8810
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.3600000000000003e-06,
+      "loss": 0.3285,
+      "step": 8820
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.3400000000000005e-06,
+      "loss": 0.3446,
+      "step": 8830
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.3200000000000002e-06,
+      "loss": 0.2856,
+      "step": 8840
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.3000000000000004e-06,
+      "loss": 0.2773,
+      "step": 8850
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.28e-06,
+      "loss": 0.282,
+      "step": 8860
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.2600000000000004e-06,
+      "loss": 0.4473,
+      "step": 8870
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.24e-06,
+      "loss": 0.3314,
+      "step": 8880
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.2200000000000003e-06,
+      "loss": 0.3704,
+      "step": 8890
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.2e-06,
+      "loss": 0.3082,
+      "step": 8900
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.1800000000000003e-06,
+      "loss": 0.3567,
+      "step": 8910
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.16e-06,
+      "loss": 0.467,
+      "step": 8920
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.1400000000000003e-06,
+      "loss": 0.4065,
+      "step": 8930
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.12e-06,
+      "loss": 0.3587,
+      "step": 8940
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.1000000000000002e-06,
+      "loss": 0.1997,
+      "step": 8950
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.08e-06,
+      "loss": 0.3872,
+      "step": 8960
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.06e-06,
+      "loss": 0.3959,
+      "step": 8970
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.04e-06,
+      "loss": 0.2842,
+      "step": 8980
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.02e-06,
+      "loss": 0.2899,
+      "step": 8990
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 0.3252,
+      "step": 9000
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.98e-06,
+      "loss": 0.3935,
+      "step": 9010
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.9600000000000003e-06,
+      "loss": 0.2214,
+      "step": 9020
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.94e-06,
+      "loss": 0.2897,
+      "step": 9030
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.9200000000000003e-06,
+      "loss": 0.371,
+      "step": 9040
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.9000000000000002e-06,
+      "loss": 0.391,
+      "step": 9050
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.8800000000000002e-06,
+      "loss": 0.3473,
+      "step": 9060
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.8600000000000002e-06,
+      "loss": 0.3959,
+      "step": 9070
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.8400000000000002e-06,
+      "loss": 0.3049,
+      "step": 9080
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.8200000000000002e-06,
+      "loss": 0.3039,
+      "step": 9090
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.8000000000000001e-06,
+      "loss": 0.3924,
+      "step": 9100
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.7800000000000001e-06,
+      "loss": 0.3453,
+      "step": 9110
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.76e-06,
+      "loss": 0.3879,
+      "step": 9120
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.74e-06,
+      "loss": 0.2591,
+      "step": 9130
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.72e-06,
+      "loss": 0.334,
+      "step": 9140
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.7000000000000002e-06,
+      "loss": 0.291,
+      "step": 9150
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.6800000000000002e-06,
+      "loss": 0.2824,
+      "step": 9160
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.6600000000000002e-06,
+      "loss": 0.3436,
+      "step": 9170
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.6400000000000002e-06,
+      "loss": 0.4787,
+      "step": 9180
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.6200000000000002e-06,
+      "loss": 0.2834,
+      "step": 9190
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 0.3493,
+      "step": 9200
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.5800000000000001e-06,
+      "loss": 0.3324,
+      "step": 9210
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.56e-06,
+      "loss": 0.4284,
+      "step": 9220
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.54e-06,
+      "loss": 0.2592,
+      "step": 9230
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.52e-06,
+      "loss": 0.3601,
+      "step": 9240
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.5e-06,
+      "loss": 0.3911,
+      "step": 9250
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.48e-06,
+      "loss": 0.3995,
+      "step": 9260
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.46e-06,
+      "loss": 0.3506,
+      "step": 9270
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.44e-06,
+      "loss": 0.3215,
+      "step": 9280
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.42e-06,
+      "loss": 0.4489,
+      "step": 9290
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 0.3886,
+      "step": 9300
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.3800000000000001e-06,
+      "loss": 0.4151,
+      "step": 9310
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.3600000000000001e-06,
+      "loss": 0.2327,
+      "step": 9320
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.34e-06,
+      "loss": 0.3776,
+      "step": 9330
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.32e-06,
+      "loss": 0.2364,
+      "step": 9340
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.3e-06,
+      "loss": 0.3731,
+      "step": 9350
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.28e-06,
+      "loss": 0.2075,
+      "step": 9360
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.26e-06,
+      "loss": 0.2859,
+      "step": 9370
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.2400000000000002e-06,
+      "loss": 0.4468,
+      "step": 9380
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.2200000000000002e-06,
+      "loss": 0.4261,
+      "step": 9390
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 0.3312,
+      "step": 9400
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.1800000000000001e-06,
+      "loss": 0.245,
+      "step": 9410
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.1600000000000001e-06,
+      "loss": 0.4722,
+      "step": 9420
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.14e-06,
+      "loss": 0.3234,
+      "step": 9430
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.12e-06,
+      "loss": 0.3657,
+      "step": 9440
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.1e-06,
+      "loss": 0.3333,
+      "step": 9450
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.08e-06,
+      "loss": 0.4043,
+      "step": 9460
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.06e-06,
+      "loss": 0.3975,
+      "step": 9470
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.04e-06,
+      "loss": 0.309,
+      "step": 9480
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.02e-06,
+      "loss": 0.2178,
+      "step": 9490
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 0.3706,
+      "step": 9500
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 9.800000000000001e-07,
+      "loss": 0.3555,
+      "step": 9510
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 9.600000000000001e-07,
+      "loss": 0.2577,
+      "step": 9520
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 9.400000000000001e-07,
+      "loss": 0.3322,
+      "step": 9530
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 9.200000000000001e-07,
+      "loss": 0.2272,
+      "step": 9540
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.000000000000001e-07,
+      "loss": 0.3338,
+      "step": 9550
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 8.8e-07,
+      "loss": 0.3833,
+      "step": 9560
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.6e-07,
+      "loss": 0.3448,
+      "step": 9570
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.400000000000001e-07,
+      "loss": 0.4303,
+      "step": 9580
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.200000000000001e-07,
+      "loss": 0.223,
+      "step": 9590
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 0.3101,
+      "step": 9600
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 7.8e-07,
+      "loss": 0.4332,
+      "step": 9610
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.6e-07,
+      "loss": 0.2678,
+      "step": 9620
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.4e-07,
+      "loss": 0.3285,
+      "step": 9630
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.2e-07,
+      "loss": 0.3571,
+      "step": 9640
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 7.000000000000001e-07,
+      "loss": 0.2149,
+      "step": 9650
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.800000000000001e-07,
+      "loss": 0.4648,
+      "step": 9660
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.6e-07,
+      "loss": 0.239,
+      "step": 9670
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 6.4e-07,
+      "loss": 0.4276,
+      "step": 9680
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 6.200000000000001e-07,
+      "loss": 0.1763,
+      "step": 9690
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 0.3054,
+      "step": 9700
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.800000000000001e-07,
+      "loss": 0.2566,
+      "step": 9710
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.6e-07,
+      "loss": 0.3284,
+      "step": 9720
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.4e-07,
+      "loss": 0.3322,
+      "step": 9730
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.2e-07,
+      "loss": 0.44,
+      "step": 9740
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 0.3967,
+      "step": 9750
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.800000000000001e-07,
+      "loss": 0.2981,
+      "step": 9760
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.6000000000000004e-07,
+      "loss": 0.2687,
+      "step": 9770
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.4e-07,
+      "loss": 0.297,
+      "step": 9780
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.2000000000000006e-07,
+      "loss": 0.3544,
+      "step": 9790
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 0.3498,
+      "step": 9800
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.8e-07,
+      "loss": 0.3002,
+      "step": 9810
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.6e-07,
+      "loss": 0.4046,
+      "step": 9820
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.4000000000000003e-07,
+      "loss": 0.3375,
+      "step": 9830
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.2e-07,
+      "loss": 0.3961,
+      "step": 9840
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.0000000000000004e-07,
+      "loss": 0.3055,
+      "step": 9850
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.8e-07,
+      "loss": 0.2224,
+      "step": 9860
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.6e-07,
+      "loss": 0.349,
+      "step": 9870
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.4000000000000003e-07,
+      "loss": 0.1733,
+      "step": 9880
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.2e-07,
+      "loss": 0.215,
+      "step": 9890
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 0.4088,
+      "step": 9900
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.8e-07,
+      "loss": 0.3428,
+      "step": 9910
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.6e-07,
+      "loss": 0.2119,
+      "step": 9920
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.4e-07,
+      "loss": 0.2488,
+      "step": 9930
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.2000000000000002e-07,
+      "loss": 0.6147,
+      "step": 9940
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.0000000000000001e-07,
+      "loss": 0.2948,
+      "step": 9950
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 8e-08,
+      "loss": 0.2339,
+      "step": 9960
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 6.000000000000001e-08,
+      "loss": 0.2891,
+      "step": 9970
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4e-08,
+      "loss": 0.3026,
+      "step": 9980
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2e-08,
+      "loss": 0.2569,
+      "step": 9990
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.0,
+      "loss": 0.3581,
+      "step": 10000
+    },
+    {
+      "epoch": 5.0,
+      "eval_accuracy": 0.9107332624867163,
+      "eval_loss": 0.21704278886318207,
+      "eval_runtime": 12.0318,
+      "eval_samples_per_second": 234.628,
+      "eval_steps_per_second": 29.339,
+      "step": 10000
+    },
+    {
+      "epoch": 5.0,
+      "step": 10000,
+      "total_flos": 6.194773019299738e+18,
+      "train_loss": 0.4213516190290451,
+      "train_runtime": 1040.551,
+      "train_samples_per_second": 76.858,
+      "train_steps_per_second": 9.61
+    }
+  ],
+  "max_steps": 10000,
+  "num_train_epochs": 5,
+  "total_flos": 6.194773019299738e+18,
+  "trial_name": null,
+  "trial_params": null
+}