diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,9354 @@
+{
+  "best_metric": 0.9857405703771849,
+  "best_model_checkpoint": "wav2vec2-present/checkpoint-13855",
+  "epoch": 9.996752192270218,
+  "eval_steps": 500,
+  "global_step": 15390,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9493177387914228e-07,
+      "loss": 2.5646,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8986354775828457e-07,
+      "loss": 2.5574,
+      "step": 20
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.847953216374269e-07,
+      "loss": 2.5648,
+      "step": 30
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.797270955165691e-07,
+      "loss": 2.5679,
+      "step": 40
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.746588693957115e-07,
+      "loss": 2.5632,
+      "step": 50
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1695906432748538e-06,
+      "loss": 2.5621,
+      "step": 60
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.364522417153996e-06,
+      "loss": 2.5622,
+      "step": 70
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5594541910331383e-06,
+      "loss": 2.5608,
+      "step": 80
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.7543859649122807e-06,
+      "loss": 2.5525,
+      "step": 90
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.949317738791423e-06,
+      "loss": 2.5571,
+      "step": 100
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1442495126705653e-06,
+      "loss": 2.5567,
+      "step": 110
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.3391812865497075e-06,
+      "loss": 2.5515,
+      "step": 120
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.5341130604288498e-06,
+      "loss": 2.5542,
+      "step": 130
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.729044834307992e-06,
+      "loss": 2.5487,
+      "step": 140
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9239766081871343e-06,
+      "loss": 2.5453,
+      "step": 150
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.1189083820662765e-06,
+      "loss": 2.5453,
+      "step": 160
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.313840155945419e-06,
+      "loss": 2.5492,
+      "step": 170
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.5087719298245615e-06,
+      "loss": 2.5339,
+      "step": 180
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.7037037037037037e-06,
+      "loss": 2.5269,
+      "step": 190
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.898635477582846e-06,
+      "loss": 2.5313,
+      "step": 200
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.093567251461988e-06,
+      "loss": 2.5306,
+      "step": 210
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.2884990253411305e-06,
+      "loss": 2.5212,
+      "step": 220
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.483430799220272e-06,
+      "loss": 2.5112,
+      "step": 230
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.678362573099415e-06,
+      "loss": 2.5022,
+      "step": 240
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.873294346978558e-06,
+      "loss": 2.4988,
+      "step": 250
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.0682261208576995e-06,
+      "loss": 2.503,
+      "step": 260
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.263157894736842e-06,
+      "loss": 2.4951,
+      "step": 270
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.458089668615984e-06,
+      "loss": 2.4562,
+      "step": 280
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.653021442495127e-06,
+      "loss": 2.4518,
+      "step": 290
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.8479532163742686e-06,
+      "loss": 2.4252,
+      "step": 300
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.042884990253411e-06,
+      "loss": 2.416,
+      "step": 310
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.237816764132553e-06,
+      "loss": 2.3758,
+      "step": 320
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.432748538011696e-06,
+      "loss": 2.3653,
+      "step": 330
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.627680311890838e-06,
+      "loss": 2.3687,
+      "step": 340
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.82261208576998e-06,
+      "loss": 2.3433,
+      "step": 350
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.017543859649123e-06,
+      "loss": 2.293,
+      "step": 360
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.212475633528265e-06,
+      "loss": 2.3328,
+      "step": 370
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.4074074074074075e-06,
+      "loss": 2.2264,
+      "step": 380
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.602339181286549e-06,
+      "loss": 2.1924,
+      "step": 390
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.797270955165692e-06,
+      "loss": 2.2046,
+      "step": 400
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.992202729044834e-06,
+      "loss": 2.1504,
+      "step": 410
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.187134502923976e-06,
+      "loss": 2.122,
+      "step": 420
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.38206627680312e-06,
+      "loss": 2.0498,
+      "step": 430
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.576998050682261e-06,
+      "loss": 2.093,
+      "step": 440
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.771929824561403e-06,
+      "loss": 2.02,
+      "step": 450
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.966861598440545e-06,
+      "loss": 1.9983,
+      "step": 460
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.161793372319688e-06,
+      "loss": 1.9804,
+      "step": 470
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.35672514619883e-06,
+      "loss": 1.9636,
+      "step": 480
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.551656920077972e-06,
+      "loss": 1.9058,
+      "step": 490
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.746588693957115e-06,
+      "loss": 1.8302,
+      "step": 500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.941520467836257e-06,
+      "loss": 1.7364,
+      "step": 510
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0136452241715399e-05,
+      "loss": 1.7473,
+      "step": 520
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0331384015594541e-05,
+      "loss": 1.7183,
+      "step": 530
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0526315789473684e-05,
+      "loss": 1.5826,
+      "step": 540
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0721247563352826e-05,
+      "loss": 1.6702,
+      "step": 550
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0916179337231968e-05,
+      "loss": 1.6055,
+      "step": 560
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.111111111111111e-05,
+      "loss": 1.5926,
+      "step": 570
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.1306042884990253e-05,
+      "loss": 1.5216,
+      "step": 580
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.1500974658869395e-05,
+      "loss": 1.4831,
+      "step": 590
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.1695906432748537e-05,
+      "loss": 1.478,
+      "step": 600
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.189083820662768e-05,
+      "loss": 1.4951,
+      "step": 610
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2066276803118908e-05,
+      "loss": 1.4149,
+      "step": 620
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.226120857699805e-05,
+      "loss": 1.4269,
+      "step": 630
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2456140350877193e-05,
+      "loss": 1.4186,
+      "step": 640
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2651072124756335e-05,
+      "loss": 1.4019,
+      "step": 650
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2846003898635477e-05,
+      "loss": 1.3041,
+      "step": 660
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.3040935672514619e-05,
+      "loss": 1.3429,
+      "step": 670
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.3235867446393762e-05,
+      "loss": 1.2323,
+      "step": 680
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3430799220272904e-05,
+      "loss": 1.3168,
+      "step": 690
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3625730994152046e-05,
+      "loss": 1.1683,
+      "step": 700
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3820662768031188e-05,
+      "loss": 1.2894,
+      "step": 710
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.4015594541910331e-05,
+      "loss": 1.1935,
+      "step": 720
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.4210526315789473e-05,
+      "loss": 1.2452,
+      "step": 730
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.43859649122807e-05,
+      "loss": 1.1743,
+      "step": 740
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.4580896686159844e-05,
+      "loss": 1.1788,
+      "step": 750
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.4775828460038986e-05,
+      "loss": 1.1124,
+      "step": 760
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.4970760233918128e-05,
+      "loss": 1.1868,
+      "step": 770
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.5165692007797271e-05,
+      "loss": 0.9854,
+      "step": 780
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.5360623781676413e-05,
+      "loss": 1.0849,
+      "step": 790
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.5555555555555555e-05,
+      "loss": 0.9514,
+      "step": 800
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.5750487329434697e-05,
+      "loss": 1.0227,
+      "step": 810
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.594541910331384e-05,
+      "loss": 1.0543,
+      "step": 820
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.614035087719298e-05,
+      "loss": 1.073,
+      "step": 830
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6335282651072126e-05,
+      "loss": 0.9335,
+      "step": 840
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6530214424951268e-05,
+      "loss": 0.918,
+      "step": 850
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.672514619883041e-05,
+      "loss": 0.9265,
+      "step": 860
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.692007797270955e-05,
+      "loss": 0.7876,
+      "step": 870
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7115009746588693e-05,
+      "loss": 1.0174,
+      "step": 880
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7290448343079922e-05,
+      "loss": 0.915,
+      "step": 890
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7485380116959064e-05,
+      "loss": 0.8503,
+      "step": 900
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.768031189083821e-05,
+      "loss": 0.8095,
+      "step": 910
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.787524366471735e-05,
+      "loss": 0.752,
+      "step": 920
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8070175438596493e-05,
+      "loss": 0.9394,
+      "step": 930
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8265107212475635e-05,
+      "loss": 0.8195,
+      "step": 940
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8460038986354776e-05,
+      "loss": 0.7987,
+      "step": 950
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8654970760233918e-05,
+      "loss": 0.7328,
+      "step": 960
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.884990253411306e-05,
+      "loss": 0.7044,
+      "step": 970
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9044834307992205e-05,
+      "loss": 0.8303,
+      "step": 980
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9239766081871347e-05,
+      "loss": 0.7132,
+      "step": 990
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.943469785575049e-05,
+      "loss": 0.7296,
+      "step": 1000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.962962962962963e-05,
+      "loss": 0.872,
+      "step": 1010
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9824561403508773e-05,
+      "loss": 0.794,
+      "step": 1020
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.0019493177387914e-05,
+      "loss": 0.6428,
+      "step": 1030
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.0214424951267056e-05,
+      "loss": 0.6758,
+      "step": 1040
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.04093567251462e-05,
+      "loss": 0.5788,
+      "step": 1050
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.0604288499025343e-05,
+      "loss": 0.6183,
+      "step": 1060
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.0799220272904485e-05,
+      "loss": 0.6845,
+      "step": 1070
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.0994152046783627e-05,
+      "loss": 0.614,
+      "step": 1080
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.118908382066277e-05,
+      "loss": 0.7144,
+      "step": 1090
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.138401559454191e-05,
+      "loss": 0.5781,
+      "step": 1100
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.1578947368421053e-05,
+      "loss": 0.587,
+      "step": 1110
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1773879142300194e-05,
+      "loss": 0.3923,
+      "step": 1120
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.196881091617934e-05,
+      "loss": 0.5006,
+      "step": 1130
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.216374269005848e-05,
+      "loss": 0.5502,
+      "step": 1140
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.2358674463937623e-05,
+      "loss": 0.4473,
+      "step": 1150
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.2553606237816765e-05,
+      "loss": 0.4308,
+      "step": 1160
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.2748538011695907e-05,
+      "loss": 0.8556,
+      "step": 1170
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.294346978557505e-05,
+      "loss": 0.5401,
+      "step": 1180
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.313840155945419e-05,
+      "loss": 0.5928,
+      "step": 1190
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3333333333333336e-05,
+      "loss": 0.5021,
+      "step": 1200
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3528265107212478e-05,
+      "loss": 0.4433,
+      "step": 1210
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.372319688109162e-05,
+      "loss": 0.6434,
+      "step": 1220
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.391812865497076e-05,
+      "loss": 0.5357,
+      "step": 1230
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.4113060428849903e-05,
+      "loss": 0.6115,
+      "step": 1240
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.4307992202729045e-05,
+      "loss": 0.5021,
+      "step": 1250
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.4502923976608187e-05,
+      "loss": 0.4228,
+      "step": 1260
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.4697855750487332e-05,
+      "loss": 0.3138,
+      "step": 1270
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.4892787524366474e-05,
+      "loss": 0.4118,
+      "step": 1280
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5087719298245616e-05,
+      "loss": 0.3991,
+      "step": 1290
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5282651072124758e-05,
+      "loss": 0.5021,
+      "step": 1300
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.54775828460039e-05,
+      "loss": 0.3522,
+      "step": 1310
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.567251461988304e-05,
+      "loss": 0.3366,
+      "step": 1320
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5867446393762183e-05,
+      "loss": 0.7333,
+      "step": 1330
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.6062378167641325e-05,
+      "loss": 0.4322,
+      "step": 1340
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.625730994152047e-05,
+      "loss": 0.4034,
+      "step": 1350
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6452241715399612e-05,
+      "loss": 0.4084,
+      "step": 1360
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6647173489278754e-05,
+      "loss": 0.4439,
+      "step": 1370
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6842105263157896e-05,
+      "loss": 0.3036,
+      "step": 1380
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7037037037037037e-05,
+      "loss": 0.4855,
+      "step": 1390
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.723196881091618e-05,
+      "loss": 0.3802,
+      "step": 1400
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.742690058479532e-05,
+      "loss": 0.3977,
+      "step": 1410
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7621832358674466e-05,
+      "loss": 0.4481,
+      "step": 1420
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7816764132553608e-05,
+      "loss": 0.3633,
+      "step": 1430
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.801169590643275e-05,
+      "loss": 0.2364,
+      "step": 1440
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8206627680311892e-05,
+      "loss": 0.4099,
+      "step": 1450
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8401559454191034e-05,
+      "loss": 0.5903,
+      "step": 1460
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8596491228070175e-05,
+      "loss": 0.2068,
+      "step": 1470
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8791423001949317e-05,
+      "loss": 0.5032,
+      "step": 1480
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8966861598440546e-05,
+      "loss": 0.5134,
+      "step": 1490
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9161793372319688e-05,
+      "loss": 0.3761,
+      "step": 1500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.935672514619883e-05,
+      "loss": 0.4829,
+      "step": 1510
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9551656920077972e-05,
+      "loss": 0.3205,
+      "step": 1520
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9746588693957114e-05,
+      "loss": 0.5946,
+      "step": 1530
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": 0.9645814167433303,
+      "eval_loss": 0.1871723085641861,
+      "eval_runtime": 11.9647,
+      "eval_samples_per_second": 181.701,
+      "eval_steps_per_second": 45.467,
+      "step": 1539
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9941520467836255e-05,
+      "loss": 0.3922,
+      "step": 1540
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.99848386398094e-05,
+      "loss": 0.3666,
+      "step": 1550
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9963179553822827e-05,
+      "loss": 0.5519,
+      "step": 1560
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9941520467836255e-05,
+      "loss": 0.6053,
+      "step": 1570
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.9919861381849688e-05,
+      "loss": 0.3558,
+      "step": 1580
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.9898202295863116e-05,
+      "loss": 0.3533,
+      "step": 1590
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.9876543209876545e-05,
+      "loss": 0.2868,
+      "step": 1600
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.9854884123889974e-05,
+      "loss": 0.2682,
+      "step": 1610
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.9833225037903402e-05,
+      "loss": 0.1893,
+      "step": 1620
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.981156595191683e-05,
+      "loss": 0.4738,
+      "step": 1630
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.9789906865930256e-05,
+      "loss": 0.3299,
+      "step": 1640
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.9768247779943685e-05,
+      "loss": 0.1656,
+      "step": 1650
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.9746588693957114e-05,
+      "loss": 0.3697,
+      "step": 1660
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.9724929607970546e-05,
+      "loss": 0.2047,
+      "step": 1670
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.9703270521983974e-05,
+      "loss": 0.5325,
+      "step": 1680
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.9681611435997403e-05,
+      "loss": 0.4451,
+      "step": 1690
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.9659952350010832e-05,
+      "loss": 0.3106,
+      "step": 1700
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.9638293264024257e-05,
+      "loss": 0.3059,
+      "step": 1710
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.9616634178037686e-05,
+      "loss": 0.3267,
+      "step": 1720
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.9594975092051114e-05,
+      "loss": 0.6133,
+      "step": 1730
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.9573316006064543e-05,
+      "loss": 0.429,
+      "step": 1740
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.9551656920077972e-05,
+      "loss": 0.4538,
+      "step": 1750
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.9529997834091404e-05,
+      "loss": 0.4357,
+      "step": 1760
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.9508338748104833e-05,
+      "loss": 0.4422,
+      "step": 1770
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.948667966211826e-05,
+      "loss": 0.4516,
+      "step": 1780
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.9465020576131687e-05,
+      "loss": 0.2641,
+      "step": 1790
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.9443361490145115e-05,
+      "loss": 0.2786,
+      "step": 1800
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.9421702404158544e-05,
+      "loss": 0.4257,
+      "step": 1810
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.9400043318171973e-05,
+      "loss": 0.2825,
+      "step": 1820
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.93783842321854e-05,
+      "loss": 0.3132,
+      "step": 1830
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.935672514619883e-05,
+      "loss": 0.5103,
+      "step": 1840
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.9335066060212262e-05,
+      "loss": 0.22,
+      "step": 1850
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.931340697422569e-05,
+      "loss": 0.304,
+      "step": 1860
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9291747888239116e-05,
+      "loss": 0.3983,
+      "step": 1870
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9270088802252545e-05,
+      "loss": 0.4939,
+      "step": 1880
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9248429716265973e-05,
+      "loss": 0.6642,
+      "step": 1890
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9226770630279402e-05,
+      "loss": 0.3381,
+      "step": 1900
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.920511154429283e-05,
+      "loss": 0.5406,
+      "step": 1910
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.918345245830626e-05,
+      "loss": 0.3367,
+      "step": 1920
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9161793372319688e-05,
+      "loss": 0.3969,
+      "step": 1930
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.914013428633312e-05,
+      "loss": 0.3194,
+      "step": 1940
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.9118475200346546e-05,
+      "loss": 0.2396,
+      "step": 1950
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.9096816114359974e-05,
+      "loss": 0.3907,
+      "step": 1960
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.9075157028373403e-05,
+      "loss": 0.4338,
+      "step": 1970
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.905349794238683e-05,
+      "loss": 0.3385,
+      "step": 1980
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.903183885640026e-05,
+      "loss": 0.4055,
+      "step": 1990
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.901017977041369e-05,
+      "loss": 0.3263,
+      "step": 2000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8988520684427118e-05,
+      "loss": 0.2367,
+      "step": 2010
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8966861598440546e-05,
+      "loss": 0.3312,
+      "step": 2020
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8945202512453975e-05,
+      "loss": 0.3165,
+      "step": 2030
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.8923543426467404e-05,
+      "loss": 0.1925,
+      "step": 2040
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.8901884340480832e-05,
+      "loss": 0.2997,
+      "step": 2050
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.888022525449426e-05,
+      "loss": 0.1096,
+      "step": 2060
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.885856616850769e-05,
+      "loss": 0.3361,
+      "step": 2070
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.883690708252112e-05,
+      "loss": 0.5627,
+      "step": 2080
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.8815247996534547e-05,
+      "loss": 0.2367,
+      "step": 2090
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.8793588910547976e-05,
+      "loss": 0.2833,
+      "step": 2100
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.87719298245614e-05,
+      "loss": 0.1689,
+      "step": 2110
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.8750270738574833e-05,
+      "loss": 0.1547,
+      "step": 2120
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.8728611652588262e-05,
+      "loss": 0.2528,
+      "step": 2130
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.870695256660169e-05,
+      "loss": 0.2751,
+      "step": 2140
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.868529348061512e-05,
+      "loss": 0.2035,
+      "step": 2150
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.8663634394628548e-05,
+      "loss": 0.2216,
+      "step": 2160
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.8641975308641977e-05,
+      "loss": 0.68,
+      "step": 2170
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.8620316222655405e-05,
+      "loss": 0.489,
+      "step": 2180
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.859865713666883e-05,
+      "loss": 0.22,
+      "step": 2190
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.857699805068226e-05,
+      "loss": 0.3444,
+      "step": 2200
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.855533896469569e-05,
+      "loss": 0.3128,
+      "step": 2210
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.853367987870912e-05,
+      "loss": 0.4159,
+      "step": 2220
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.851202079272255e-05,
+      "loss": 0.1632,
+      "step": 2230
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.8490361706735977e-05,
+      "loss": 0.1909,
+      "step": 2240
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.8468702620749406e-05,
+      "loss": 0.2036,
+      "step": 2250
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.8447043534762835e-05,
+      "loss": 0.2216,
+      "step": 2260
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.842538444877626e-05,
+      "loss": 0.4232,
+      "step": 2270
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.840372536278969e-05,
+      "loss": 0.2262,
+      "step": 2280
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.8382066276803117e-05,
+      "loss": 0.1092,
+      "step": 2290
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.836040719081655e-05,
+      "loss": 0.4202,
+      "step": 2300
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.8338748104829978e-05,
+      "loss": 0.1182,
+      "step": 2310
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.8317089018843407e-05,
+      "loss": 0.4046,
+      "step": 2320
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.8295429932856836e-05,
+      "loss": 0.4827,
+      "step": 2330
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8273770846870264e-05,
+      "loss": 0.3009,
+      "step": 2340
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.825211176088369e-05,
+      "loss": 0.1604,
+      "step": 2350
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.8230452674897118e-05,
+      "loss": 0.4158,
+      "step": 2360
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.8208793588910547e-05,
+      "loss": 0.3007,
+      "step": 2370
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.8187134502923976e-05,
+      "loss": 0.3576,
+      "step": 2380
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.8165475416937408e-05,
+      "loss": 0.4811,
+      "step": 2390
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.8143816330950836e-05,
+      "loss": 0.2803,
+      "step": 2400
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.8122157244964265e-05,
+      "loss": 0.3015,
+      "step": 2410
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.8100498158977694e-05,
+      "loss": 0.1824,
+      "step": 2420
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.807883907299112e-05,
+      "loss": 0.1548,
+      "step": 2430
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.8057179987004548e-05,
+      "loss": 0.4265,
+      "step": 2440
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.8035520901017976e-05,
+      "loss": 0.0856,
+      "step": 2450
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.8013861815031405e-05,
+      "loss": 0.1945,
+      "step": 2460
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.7992202729044834e-05,
+      "loss": 0.2537,
+      "step": 2470
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.7970543643058266e-05,
+      "loss": 0.1947,
+      "step": 2480
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7948884557071695e-05,
+      "loss": 0.6065,
+      "step": 2490
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7927225471085123e-05,
+      "loss": 0.4276,
+      "step": 2500
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.790556638509855e-05,
+      "loss": 0.4066,
+      "step": 2510
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.7883907299111977e-05,
+      "loss": 0.069,
+      "step": 2520
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.7862248213125406e-05,
+      "loss": 0.488,
+      "step": 2530
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.7840589127138835e-05,
+      "loss": 0.418,
+      "step": 2540
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.7818930041152263e-05,
+      "loss": 0.2496,
+      "step": 2550
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.7797270955165692e-05,
+      "loss": 0.2981,
+      "step": 2560
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.7775611869179124e-05,
+      "loss": 0.3107,
+      "step": 2570
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.7753952783192553e-05,
+      "loss": 0.5043,
+      "step": 2580
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.7732293697205978e-05,
+      "loss": 0.4458,
+      "step": 2590
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.7710634611219407e-05,
+      "loss": 0.1837,
+      "step": 2600
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.7688975525232835e-05,
+      "loss": 0.1487,
+      "step": 2610
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.7667316439246264e-05,
+      "loss": 0.526,
+      "step": 2620
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.7645657353259693e-05,
+      "loss": 0.4615,
+      "step": 2630
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.762399826727312e-05,
+      "loss": 0.3477,
+      "step": 2640
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.760233918128655e-05,
+      "loss": 0.5001,
+      "step": 2650
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.758068009529998e-05,
+      "loss": 0.2388,
+      "step": 2660
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.7559021009313407e-05,
+      "loss": 0.4698,
+      "step": 2670
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.7537361923326836e-05,
+      "loss": 0.2285,
+      "step": 2680
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.7515702837340265e-05,
+      "loss": 0.2899,
+      "step": 2690
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.749620965995235e-05,
+      "loss": 0.2296,
+      "step": 2700
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.747455057396578e-05,
+      "loss": 0.2841,
+      "step": 2710
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.745289148797921e-05,
+      "loss": 0.4159,
+      "step": 2720
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.7431232401992638e-05,
+      "loss": 0.4595,
+      "step": 2730
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.7409573316006067e-05,
+      "loss": 0.484,
+      "step": 2740
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.7387914230019495e-05,
+      "loss": 0.3085,
+      "step": 2750
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.736625514403292e-05,
+      "loss": 0.1533,
+      "step": 2760
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.734459605804635e-05,
+      "loss": 0.3228,
+      "step": 2770
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.7322936972059778e-05,
+      "loss": 0.3396,
+      "step": 2780
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.7301277886073207e-05,
+      "loss": 0.0364,
+      "step": 2790
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.727961880008664e-05,
+      "loss": 0.3512,
+      "step": 2800
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.7257959714100068e-05,
+      "loss": 0.3361,
+      "step": 2810
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.7236300628113496e-05,
+      "loss": 0.3569,
+      "step": 2820
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.7214641542126925e-05,
+      "loss": 0.3248,
+      "step": 2830
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.719298245614035e-05,
+      "loss": 0.3594,
+      "step": 2840
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.717132337015378e-05,
+      "loss": 0.528,
+      "step": 2850
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.7149664284167208e-05,
+      "loss": 0.4059,
+      "step": 2860
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.7128005198180636e-05,
+      "loss": 0.3046,
+      "step": 2870
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.7106346112194065e-05,
+      "loss": 0.3214,
+      "step": 2880
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.7084687026207497e-05,
+      "loss": 0.2809,
+      "step": 2890
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.7063027940220926e-05,
+      "loss": 0.2166,
+      "step": 2900
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.7041368854234354e-05,
+      "loss": 0.4228,
+      "step": 2910
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.701970976824778e-05,
+      "loss": 0.2302,
+      "step": 2920
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.699805068226121e-05,
+      "loss": 0.1923,
+      "step": 2930
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.6976391596274637e-05,
+      "loss": 0.5294,
+      "step": 2940
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.6954732510288066e-05,
+      "loss": 0.5367,
+      "step": 2950
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.6933073424301494e-05,
+      "loss": 0.1474,
+      "step": 2960
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.6911414338314923e-05,
+      "loss": 0.3966,
+      "step": 2970
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.6889755252328355e-05,
+      "loss": 0.2654,
+      "step": 2980
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.6868096166341784e-05,
+      "loss": 0.3656,
+      "step": 2990
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.684643708035521e-05,
+      "loss": 0.4402,
+      "step": 3000
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.6824777994368638e-05,
+      "loss": 0.2606,
+      "step": 3010
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.6803118908382067e-05,
+      "loss": 0.2087,
+      "step": 3020
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.6781459822395495e-05,
+      "loss": 0.3498,
+      "step": 3030
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.6759800736408924e-05,
+      "loss": 0.204,
+      "step": 3040
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.6738141650422353e-05,
+      "loss": 0.252,
+      "step": 3050
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.671648256443578e-05,
+      "loss": 0.3591,
+      "step": 3060
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.6694823478449213e-05,
+      "loss": 0.1682,
+      "step": 3070
+    },
+    {
+      "epoch": 2.0,
+      "eval_accuracy": 0.9705611775528978,
+      "eval_loss": 0.14409060776233673,
+      "eval_runtime": 11.2406,
+      "eval_samples_per_second": 193.405,
+      "eval_steps_per_second": 48.396,
+      "step": 3079
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.667316439246264e-05,
+      "loss": 0.3395,
+      "step": 3080
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.6651505306476067e-05,
+      "loss": 0.3887,
+      "step": 3090
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.6629846220489496e-05,
+      "loss": 0.4136,
+      "step": 3100
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.6608187134502925e-05,
+      "loss": 0.193,
+      "step": 3110
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.6586528048516353e-05,
+      "loss": 0.2198,
+      "step": 3120
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.6564868962529782e-05,
+      "loss": 0.3296,
+      "step": 3130
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.654320987654321e-05,
+      "loss": 0.2935,
+      "step": 3140
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.652155079055664e-05,
+      "loss": 0.4362,
+      "step": 3150
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.6499891704570065e-05,
+      "loss": 0.2129,
+      "step": 3160
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.6478232618583497e-05,
+      "loss": 0.2338,
+      "step": 3170
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.6456573532596926e-05,
+      "loss": 0.4846,
+      "step": 3180
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.6434914446610354e-05,
+      "loss": 0.3385,
+      "step": 3190
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.6413255360623783e-05,
+      "loss": 0.3746,
+      "step": 3200
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.639159627463721e-05,
+      "loss": 0.2795,
+      "step": 3210
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.636993718865064e-05,
+      "loss": 0.3332,
+      "step": 3220
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.634827810266407e-05,
+      "loss": 0.3197,
+      "step": 3230
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.6326619016677494e-05,
+      "loss": 0.3093,
+      "step": 3240
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.6304959930690923e-05,
+      "loss": 0.0929,
+      "step": 3250
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.6283300844704355e-05,
+      "loss": 0.3409,
+      "step": 3260
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.6261641758717784e-05,
+      "loss": 0.2221,
+      "step": 3270
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.6239982672731212e-05,
+      "loss": 0.2957,
+      "step": 3280
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.621832358674464e-05,
+      "loss": 0.4028,
+      "step": 3290
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.619666450075807e-05,
+      "loss": 0.3095,
+      "step": 3300
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.61750054147715e-05,
+      "loss": 0.2265,
+      "step": 3310
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.6153346328784924e-05,
+      "loss": 0.1059,
+      "step": 3320
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.6131687242798352e-05,
+      "loss": 0.4659,
+      "step": 3330
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.611002815681178e-05,
+      "loss": 0.1842,
+      "step": 3340
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.6088369070825213e-05,
+      "loss": 0.3434,
+      "step": 3350
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.6066709984838642e-05,
+      "loss": 0.3828,
+      "step": 3360
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.604505089885207e-05,
+      "loss": 0.0725,
+      "step": 3370
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.60233918128655e-05,
+      "loss": 0.3887,
+      "step": 3380
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.6001732726878928e-05,
+      "loss": 0.3053,
+      "step": 3390
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.5980073640892353e-05,
+      "loss": 0.0642,
+      "step": 3400
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.5958414554905782e-05,
+      "loss": 0.2478,
+      "step": 3410
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.593675546891921e-05,
+      "loss": 0.1497,
+      "step": 3420
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.591509638293264e-05,
+      "loss": 0.5159,
+      "step": 3430
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.589343729694607e-05,
+      "loss": 0.2398,
+      "step": 3440
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.58717782109595e-05,
+      "loss": 0.5369,
+      "step": 3450
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.585011912497293e-05,
+      "loss": 0.554,
+      "step": 3460
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.5828460038986354e-05,
+      "loss": 0.12,
+      "step": 3470
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.5806800952999783e-05,
+      "loss": 0.2994,
+      "step": 3480
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.578514186701321e-05,
+      "loss": 0.1856,
+      "step": 3490
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.576348278102664e-05,
+      "loss": 0.3596,
+      "step": 3500
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.574182369504007e-05,
+      "loss": 0.1941,
+      "step": 3510
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.5720164609053497e-05,
+      "loss": 0.2724,
+      "step": 3520
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.569850552306693e-05,
+      "loss": 0.155,
+      "step": 3530
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.5676846437080358e-05,
+      "loss": 0.2923,
+      "step": 3540
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.5655187351093783e-05,
+      "loss": 0.2938,
+      "step": 3550
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.5633528265107212e-05,
+      "loss": 0.2757,
+      "step": 3560
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.561186917912064e-05,
+      "loss": 0.2673,
+      "step": 3570
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.559021009313407e-05,
+      "loss": 0.316,
+      "step": 3580
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5568551007147498e-05,
+      "loss": 0.1623,
+      "step": 3590
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.5546891921160927e-05,
+      "loss": 0.3536,
+      "step": 3600
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.5525232835174356e-05,
+      "loss": 0.1398,
+      "step": 3610
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.5503573749187788e-05,
+      "loss": 0.2394,
+      "step": 3620
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.5481914663201213e-05,
+      "loss": 0.2099,
+      "step": 3630
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.546025557721464e-05,
+      "loss": 0.5143,
+      "step": 3640
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.543859649122807e-05,
+      "loss": 0.5471,
+      "step": 3650
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.54169374052415e-05,
+      "loss": 0.2749,
+      "step": 3660
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.5395278319254928e-05,
+      "loss": 0.446,
+      "step": 3670
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.5373619233268356e-05,
+      "loss": 0.4234,
+      "step": 3680
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.5351960147281785e-05,
+      "loss": 0.1677,
+      "step": 3690
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.5330301061295214e-05,
+      "loss": 0.3338,
+      "step": 3700
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.5308641975308642e-05,
+      "loss": 0.5578,
+      "step": 3710
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.528698288932207e-05,
+      "loss": 0.128,
+      "step": 3720
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.52653238033355e-05,
+      "loss": 0.2384,
+      "step": 3730
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.524366471734893e-05,
+      "loss": 0.121,
+      "step": 3740
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.5222005631362357e-05,
+      "loss": 0.2308,
+      "step": 3750
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.5200346545375786e-05,
+      "loss": 0.2064,
+      "step": 3760
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.5178687459389215e-05,
+      "loss": 0.3473,
+      "step": 3770
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.5157028373402643e-05,
+      "loss": 0.408,
+      "step": 3780
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.513536928741607e-05,
+      "loss": 0.0684,
+      "step": 3790
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.51137102014295e-05,
+      "loss": 0.3948,
+      "step": 3800
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.509205111544293e-05,
+      "loss": 0.2305,
+      "step": 3810
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.5070392029456358e-05,
+      "loss": 0.3364,
+      "step": 3820
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.5048732943469787e-05,
+      "loss": 0.4581,
+      "step": 3830
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.5027073857483215e-05,
+      "loss": 0.2409,
+      "step": 3840
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.5005414771496644e-05,
+      "loss": 0.2305,
+      "step": 3850
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.4983755685510073e-05,
+      "loss": 0.3771,
+      "step": 3860
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.4962096599523498e-05,
+      "loss": 0.3271,
+      "step": 3870
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.4940437513536927e-05,
+      "loss": 0.1611,
+      "step": 3880
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.491877842755036e-05,
+      "loss": 0.2923,
+      "step": 3890
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.4897119341563787e-05,
+      "loss": 0.5307,
+      "step": 3900
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.4875460255577216e-05,
+      "loss": 0.1327,
+      "step": 3910
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.4853801169590645e-05,
+      "loss": 0.4278,
+      "step": 3920
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.4832142083604074e-05,
+      "loss": 0.2369,
+      "step": 3930
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.4810482997617502e-05,
+      "loss": 0.1881,
+      "step": 3940
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.4788823911630927e-05,
+      "loss": 0.2229,
+      "step": 3950
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.4767164825644356e-05,
+      "loss": 0.1225,
+      "step": 3960
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4745505739657785e-05,
+      "loss": 0.5652,
+      "step": 3970
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4723846653671217e-05,
+      "loss": 0.2312,
+      "step": 3980
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4702187567684646e-05,
+      "loss": 0.0249,
+      "step": 3990
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4680528481698074e-05,
+      "loss": 0.2958,
+      "step": 4000
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4658869395711503e-05,
+      "loss": 0.3693,
+      "step": 4010
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.463721030972493e-05,
+      "loss": 0.2373,
+      "step": 4020
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4615551223738357e-05,
+      "loss": 0.2235,
+      "step": 4030
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4593892137751786e-05,
+      "loss": 0.3099,
+      "step": 4040
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4572233051765214e-05,
+      "loss": 0.2254,
+      "step": 4050
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4550573965778643e-05,
+      "loss": 0.4968,
+      "step": 4060
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4528914879792075e-05,
+      "loss": 0.1395,
+      "step": 4070
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.4507255793805504e-05,
+      "loss": 0.2843,
+      "step": 4080
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.4485596707818932e-05,
+      "loss": 0.2342,
+      "step": 4090
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.446393762183236e-05,
+      "loss": 0.296,
+      "step": 4100
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.4442278535845786e-05,
+      "loss": 0.3318,
+      "step": 4110
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.4420619449859215e-05,
+      "loss": 0.3469,
+      "step": 4120
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.4398960363872644e-05,
+      "loss": 0.4605,
+      "step": 4130
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.4377301277886073e-05,
+      "loss": 0.4931,
+      "step": 4140
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.43556421918995e-05,
+      "loss": 0.259,
+      "step": 4150
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.4333983105912933e-05,
+      "loss": 0.3474,
+      "step": 4160
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.4312324019926362e-05,
+      "loss": 0.3581,
+      "step": 4170
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.429066493393979e-05,
+      "loss": 0.1622,
+      "step": 4180
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.4269005847953216e-05,
+      "loss": 0.3018,
+      "step": 4190
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.4247346761966645e-05,
+      "loss": 0.1869,
+      "step": 4200
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.4225687675980073e-05,
+      "loss": 0.3683,
+      "step": 4210
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.4204028589993502e-05,
+      "loss": 0.1742,
+      "step": 4220
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.418236950400693e-05,
+      "loss": 0.1894,
+      "step": 4230
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.416071041802036e-05,
+      "loss": 0.1319,
+      "step": 4240
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.413905133203379e-05,
+      "loss": 0.3434,
+      "step": 4250
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.411739224604722e-05,
+      "loss": 0.4275,
+      "step": 4260
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.4095733160060645e-05,
+      "loss": 0.3749,
+      "step": 4270
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.4074074074074074e-05,
+      "loss": 0.2838,
+      "step": 4280
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4052414988087503e-05,
+      "loss": 0.2925,
+      "step": 4290
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.403075590210093e-05,
+      "loss": 0.2661,
+      "step": 4300
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.400909681611436e-05,
+      "loss": 0.1254,
+      "step": 4310
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.398743773012779e-05,
+      "loss": 0.1943,
+      "step": 4320
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.3965778644141218e-05,
+      "loss": 0.3255,
+      "step": 4330
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.394411955815465e-05,
+      "loss": 0.237,
+      "step": 4340
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.3922460472168075e-05,
+      "loss": 0.1623,
+      "step": 4350
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.3900801386181504e-05,
+      "loss": 0.177,
+      "step": 4360
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.3879142300194932e-05,
+      "loss": 0.3462,
+      "step": 4370
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.385748321420836e-05,
+      "loss": 0.1004,
+      "step": 4380
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.383582412822179e-05,
+      "loss": 0.1144,
+      "step": 4390
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.381416504223522e-05,
+      "loss": 0.1841,
+      "step": 4400
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3792505956248647e-05,
+      "loss": 0.3915,
+      "step": 4410
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.3770846870262076e-05,
+      "loss": 0.3057,
+      "step": 4420
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.3749187784275504e-05,
+      "loss": 0.2484,
+      "step": 4430
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.3727528698288933e-05,
+      "loss": 0.2424,
+      "step": 4440
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.3705869612302362e-05,
+      "loss": 0.2855,
+      "step": 4450
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.368421052631579e-05,
+      "loss": 0.1492,
+      "step": 4460
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.366255144032922e-05,
+      "loss": 0.3557,
+      "step": 4470
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.3640892354342648e-05,
+      "loss": 0.2224,
+      "step": 4480
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.3619233268356076e-05,
+      "loss": 0.1159,
+      "step": 4490
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.3597574182369505e-05,
+      "loss": 0.2462,
+      "step": 4500
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.357591509638293e-05,
+      "loss": 0.2532,
+      "step": 4510
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3554256010396363e-05,
+      "loss": 0.465,
+      "step": 4520
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.353259692440979e-05,
+      "loss": 0.1533,
+      "step": 4530
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.351093783842322e-05,
+      "loss": 0.158,
+      "step": 4540
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.348927875243665e-05,
+      "loss": 0.1878,
+      "step": 4550
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.3467619666450077e-05,
+      "loss": 0.1176,
+      "step": 4560
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.3445960580463506e-05,
+      "loss": 0.3252,
+      "step": 4570
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.342430149447693e-05,
+      "loss": 0.2085,
+      "step": 4580
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.340264240849036e-05,
+      "loss": 0.3168,
+      "step": 4590
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.338098332250379e-05,
+      "loss": 0.4482,
+      "step": 4600
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.335932423651722e-05,
+      "loss": 0.2124,
+      "step": 4610
+    },
+    {
+      "epoch": 3.0,
+      "eval_accuracy": 0.9710211591536339,
+      "eval_loss": 0.14072078466415405,
+      "eval_runtime": 11.3617,
+      "eval_samples_per_second": 191.344,
+      "eval_steps_per_second": 47.88,
+      "step": 4618
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.333766515053065e-05,
+      "loss": 0.4011,
+      "step": 4620
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.3316006064544078e-05,
+      "loss": 0.3261,
+      "step": 4630
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.3294346978557507e-05,
+      "loss": 0.2479,
+      "step": 4640
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.3272687892570935e-05,
+      "loss": 0.0598,
+      "step": 4650
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.325102880658436e-05,
+      "loss": 0.2428,
+      "step": 4660
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.322936972059779e-05,
+      "loss": 0.4721,
+      "step": 4670
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.3207710634611218e-05,
+      "loss": 0.0608,
+      "step": 4680
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.3186051548624647e-05,
+      "loss": 0.227,
+      "step": 4690
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.316439246263808e-05,
+      "loss": 0.1291,
+      "step": 4700
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.3142733376651508e-05,
+      "loss": 0.2276,
+      "step": 4710
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.3121074290664936e-05,
+      "loss": 0.4605,
+      "step": 4720
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.3099415204678365e-05,
+      "loss": 0.2018,
+      "step": 4730
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.307775611869179e-05,
+      "loss": 0.0912,
+      "step": 4740
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.305609703270522e-05,
+      "loss": 0.2624,
+      "step": 4750
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.3034437946718648e-05,
+      "loss": 0.063,
+      "step": 4760
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.3012778860732076e-05,
+      "loss": 0.1739,
+      "step": 4770
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.2991119774745505e-05,
+      "loss": 0.3303,
+      "step": 4780
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.2969460688758937e-05,
+      "loss": 0.2697,
+      "step": 4790
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.2947801602772366e-05,
+      "loss": 0.2016,
+      "step": 4800
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.2926142516785794e-05,
+      "loss": 0.1915,
+      "step": 4810
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.290448343079922e-05,
+      "loss": 0.1271,
+      "step": 4820
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.288282434481265e-05,
+      "loss": 0.2432,
+      "step": 4830
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.2861165258826077e-05,
+      "loss": 0.1565,
+      "step": 4840
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.2839506172839506e-05,
+      "loss": 0.1643,
+      "step": 4850
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.2817847086852934e-05,
+      "loss": 0.3077,
+      "step": 4860
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.2796188000866363e-05,
+      "loss": 0.2494,
+      "step": 4870
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.2774528914879795e-05,
+      "loss": 0.1827,
+      "step": 4880
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.2752869828893224e-05,
+      "loss": 0.3153,
+      "step": 4890
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.273121074290665e-05,
+      "loss": 0.2205,
+      "step": 4900
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.2709551656920078e-05,
+      "loss": 0.3212,
+      "step": 4910
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.2687892570933507e-05,
+      "loss": 0.1531,
+      "step": 4920
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.266839939354559e-05,
+      "loss": 0.1684,
+      "step": 4930
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2.2646740307559022e-05,
+      "loss": 0.2157,
+      "step": 4940
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.262508122157245e-05,
+      "loss": 0.1079,
+      "step": 4950
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.260342213558588e-05,
+      "loss": 0.3359,
+      "step": 4960
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2.258176304959931e-05,
+      "loss": 0.1343,
+      "step": 4970
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2.2560103963612737e-05,
+      "loss": 0.3579,
+      "step": 4980
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2.2538444877626166e-05,
+      "loss": 0.3208,
+      "step": 4990
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2.251678579163959e-05,
+      "loss": 0.2924,
+      "step": 5000
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2.249512670565302e-05,
+      "loss": 0.074,
+      "step": 5010
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2.247346761966645e-05,
+      "loss": 0.4769,
+      "step": 5020
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2.245180853367988e-05,
+      "loss": 0.1781,
+      "step": 5030
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2.243014944769331e-05,
+      "loss": 0.1767,
+      "step": 5040
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2.2408490361706738e-05,
+      "loss": 0.1095,
+      "step": 5050
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2.2386831275720167e-05,
+      "loss": 0.2566,
+      "step": 5060
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2.2365172189733595e-05,
+      "loss": 0.4422,
+      "step": 5070
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2.234351310374702e-05,
+      "loss": 0.0663,
+      "step": 5080
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2.232185401776045e-05,
+      "loss": 0.1382,
+      "step": 5090
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2.2300194931773878e-05,
+      "loss": 0.1328,
+      "step": 5100
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2.2278535845787307e-05,
+      "loss": 0.3324,
+      "step": 5110
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2.225687675980074e-05,
+      "loss": 0.2265,
+      "step": 5120
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2.2235217673814167e-05,
+      "loss": 0.1227,
+      "step": 5130
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2.2213558587827596e-05,
+      "loss": 0.2375,
+      "step": 5140
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2.2191899501841025e-05,
+      "loss": 0.095,
+      "step": 5150
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2.217024041585445e-05,
+      "loss": 0.3678,
+      "step": 5160
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2.214858132986788e-05,
+      "loss": 0.164,
+      "step": 5170
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2.2126922243881307e-05,
+      "loss": 0.1916,
+      "step": 5180
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2.2105263157894736e-05,
+      "loss": 0.1649,
+      "step": 5190
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2.2083604071908165e-05,
+      "loss": 0.096,
+      "step": 5200
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2.2061944985921597e-05,
+      "loss": 0.252,
+      "step": 5210
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2.2040285899935026e-05,
+      "loss": 0.5103,
+      "step": 5220
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2.201862681394845e-05,
+      "loss": 0.3264,
+      "step": 5230
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2.199696772796188e-05,
+      "loss": 0.396,
+      "step": 5240
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2.1975308641975308e-05,
+      "loss": 0.4319,
+      "step": 5250
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2.1953649555988737e-05,
+      "loss": 0.1546,
+      "step": 5260
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2.1931990470002166e-05,
+      "loss": 0.28,
+      "step": 5270
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2.1910331384015594e-05,
+      "loss": 0.1436,
+      "step": 5280
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2.1888672298029023e-05,
+      "loss": 0.3984,
+      "step": 5290
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2.1867013212042455e-05,
+      "loss": 0.1428,
+      "step": 5300
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2.184535412605588e-05,
+      "loss": 0.3148,
+      "step": 5310
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2.182369504006931e-05,
+      "loss": 0.1803,
+      "step": 5320
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2.1802035954082738e-05,
+      "loss": 0.2355,
+      "step": 5330
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2.1780376868096166e-05,
+      "loss": 0.3809,
+      "step": 5340
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2.1758717782109595e-05,
+      "loss": 0.206,
+      "step": 5350
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2.1737058696123024e-05,
+      "loss": 0.0645,
+      "step": 5360
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2.1715399610136452e-05,
+      "loss": 0.2012,
+      "step": 5370
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2.169374052414988e-05,
+      "loss": 0.1244,
+      "step": 5380
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2.167208143816331e-05,
+      "loss": 0.184,
+      "step": 5390
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2.165042235217674e-05,
+      "loss": 0.3843,
+      "step": 5400
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2.1628763266190167e-05,
+      "loss": 0.4206,
+      "step": 5410
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2.1607104180203596e-05,
+      "loss": 0.3413,
+      "step": 5420
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2.1585445094217025e-05,
+      "loss": 0.2414,
+      "step": 5430
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2.1563786008230453e-05,
+      "loss": 0.2005,
+      "step": 5440
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2.1542126922243882e-05,
+      "loss": 0.2454,
+      "step": 5450
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2.152046783625731e-05,
+      "loss": 0.1222,
+      "step": 5460
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2.1498808750270736e-05,
+      "loss": 0.1544,
+      "step": 5470
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.1477149664284168e-05,
+      "loss": 0.2211,
+      "step": 5480
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2.1455490578297597e-05,
+      "loss": 0.1313,
+      "step": 5490
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2.1433831492311025e-05,
+      "loss": 0.256,
+      "step": 5500
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2.1412172406324454e-05,
+      "loss": 0.4428,
+      "step": 5510
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2.1390513320337883e-05,
+      "loss": 0.4144,
+      "step": 5520
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2.136885423435131e-05,
+      "loss": 0.3623,
+      "step": 5530
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2.134719514836474e-05,
+      "loss": 0.2175,
+      "step": 5540
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2.1325536062378165e-05,
+      "loss": 0.0741,
+      "step": 5550
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2.1303876976391594e-05,
+      "loss": 0.1698,
+      "step": 5560
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2.1282217890405026e-05,
+      "loss": 0.0258,
+      "step": 5570
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2.1260558804418455e-05,
+      "loss": 0.1483,
+      "step": 5580
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2.1238899718431884e-05,
+      "loss": 0.0344,
+      "step": 5590
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2.1217240632445312e-05,
+      "loss": 0.2812,
+      "step": 5600
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2.119558154645874e-05,
+      "loss": 0.4085,
+      "step": 5610
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2.117392246047217e-05,
+      "loss": 0.125,
+      "step": 5620
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.1152263374485595e-05,
+      "loss": 0.2444,
+      "step": 5630
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.1130604288499024e-05,
+      "loss": 0.0486,
+      "step": 5640
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.1108945202512452e-05,
+      "loss": 0.2719,
+      "step": 5650
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.1087286116525884e-05,
+      "loss": 0.2245,
+      "step": 5660
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.1065627030539313e-05,
+      "loss": 0.1015,
+      "step": 5670
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.1043967944552742e-05,
+      "loss": 0.2162,
+      "step": 5680
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.102230885856617e-05,
+      "loss": 0.2483,
+      "step": 5690
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.10006497725796e-05,
+      "loss": 0.3852,
+      "step": 5700
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.0978990686593024e-05,
+      "loss": 0.1349,
+      "step": 5710
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.0957331600606453e-05,
+      "loss": 0.239,
+      "step": 5720
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.0935672514619882e-05,
+      "loss": 0.1035,
+      "step": 5730
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.091401342863331e-05,
+      "loss": 0.1655,
+      "step": 5740
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.0892354342646743e-05,
+      "loss": 0.1174,
+      "step": 5750
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.087069525666017e-05,
+      "loss": 0.1455,
+      "step": 5760
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.08490361706736e-05,
+      "loss": 0.3763,
+      "step": 5770
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.082737708468703e-05,
+      "loss": 0.2423,
+      "step": 5780
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.0805717998700454e-05,
+      "loss": 0.2225,
+      "step": 5790
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.0784058912713883e-05,
+      "loss": 0.0747,
+      "step": 5800
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.076239982672731e-05,
+      "loss": 0.0322,
+      "step": 5810
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.074074074074074e-05,
+      "loss": 0.2534,
+      "step": 5820
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.071908165475417e-05,
+      "loss": 0.1408,
+      "step": 5830
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.06974225687676e-05,
+      "loss": 0.0425,
+      "step": 5840
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.067576348278103e-05,
+      "loss": 0.5257,
+      "step": 5850
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.0654104396794458e-05,
+      "loss": 0.065,
+      "step": 5860
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.0632445310807883e-05,
+      "loss": 0.3433,
+      "step": 5870
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.0610786224821312e-05,
+      "loss": 0.1285,
+      "step": 5880
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.058912713883474e-05,
+      "loss": 0.1402,
+      "step": 5890
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.056746805284817e-05,
+      "loss": 0.1291,
+      "step": 5900
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.0545808966861598e-05,
+      "loss": 0.3868,
+      "step": 5910
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.0524149880875027e-05,
+      "loss": 0.2525,
+      "step": 5920
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.050249079488846e-05,
+      "loss": 0.2196,
+      "step": 5930
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.0480831708901888e-05,
+      "loss": 0.3587,
+      "step": 5940
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.0459172622915313e-05,
+      "loss": 0.1055,
+      "step": 5950
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.043751353692874e-05,
+      "loss": 0.2379,
+      "step": 5960
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.041585445094217e-05,
+      "loss": 0.1997,
+      "step": 5970
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.03941953649556e-05,
+      "loss": 0.2737,
+      "step": 5980
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.0372536278969028e-05,
+      "loss": 0.2618,
+      "step": 5990
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.0350877192982456e-05,
+      "loss": 0.2753,
+      "step": 6000
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.0329218106995885e-05,
+      "loss": 0.3831,
+      "step": 6010
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.0307559021009317e-05,
+      "loss": 0.2203,
+      "step": 6020
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.0285899935022742e-05,
+      "loss": 0.3,
+      "step": 6030
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.026424084903617e-05,
+      "loss": 0.0537,
+      "step": 6040
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.02425817630496e-05,
+      "loss": 0.1554,
+      "step": 6050
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.022092267706303e-05,
+      "loss": 0.1027,
+      "step": 6060
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.0199263591076457e-05,
+      "loss": 0.1178,
+      "step": 6070
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.0177604505089886e-05,
+      "loss": 0.1043,
+      "step": 6080
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.0155945419103314e-05,
+      "loss": 0.2862,
+      "step": 6090
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.0134286333116743e-05,
+      "loss": 0.4109,
+      "step": 6100
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.0112627247130172e-05,
+      "loss": 0.1335,
+      "step": 6110
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.00909681611436e-05,
+      "loss": 0.1867,
+      "step": 6120
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.006930907515703e-05,
+      "loss": 0.394,
+      "step": 6130
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0047649989170458e-05,
+      "loss": 0.1857,
+      "step": 6140
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0025990903183887e-05,
+      "loss": 0.1806,
+      "step": 6150
+    },
+    {
+      "epoch": 4.0,
+      "eval_accuracy": 0.9793008279668813,
+      "eval_loss": 0.10733841359615326,
+      "eval_runtime": 11.4295,
+      "eval_samples_per_second": 190.209,
+      "eval_steps_per_second": 47.596,
+      "step": 6158
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0004331817197315e-05,
+      "loss": 0.1917,
+      "step": 6160
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9982672731210744e-05,
+      "loss": 0.3942,
+      "step": 6170
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9961013645224173e-05,
+      "loss": 0.1945,
+      "step": 6180
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9939354559237598e-05,
+      "loss": 0.2093,
+      "step": 6190
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.991769547325103e-05,
+      "loss": 0.0718,
+      "step": 6200
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.989603638726446e-05,
+      "loss": 0.2807,
+      "step": 6210
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.9874377301277887e-05,
+      "loss": 0.3303,
+      "step": 6220
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.9852718215291316e-05,
+      "loss": 0.1774,
+      "step": 6230
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.9831059129304745e-05,
+      "loss": 0.1302,
+      "step": 6240
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.9809400043318173e-05,
+      "loss": 0.2959,
+      "step": 6250
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.9787740957331602e-05,
+      "loss": 0.2284,
+      "step": 6260
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.9766081871345027e-05,
+      "loss": 0.0929,
+      "step": 6270
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.9744422785358456e-05,
+      "loss": 0.1274,
+      "step": 6280
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.9722763699371888e-05,
+      "loss": 0.1113,
+      "step": 6290
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.9701104613385317e-05,
+      "loss": 0.2069,
+      "step": 6300
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.9679445527398745e-05,
+      "loss": 0.3271,
+      "step": 6310
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.9657786441412174e-05,
+      "loss": 0.3148,
+      "step": 6320
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.9636127355425603e-05,
+      "loss": 0.2589,
+      "step": 6330
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.9614468269439028e-05,
+      "loss": 0.2478,
+      "step": 6340
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.9592809183452457e-05,
+      "loss": 0.202,
+      "step": 6350
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.9571150097465886e-05,
+      "loss": 0.1586,
+      "step": 6360
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.9549491011479314e-05,
+      "loss": 0.428,
+      "step": 6370
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.9527831925492746e-05,
+      "loss": 0.2479,
+      "step": 6380
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.9506172839506175e-05,
+      "loss": 0.1484,
+      "step": 6390
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.9484513753519604e-05,
+      "loss": 0.2138,
+      "step": 6400
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.9462854667533032e-05,
+      "loss": 0.016,
+      "step": 6410
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.9441195581546458e-05,
+      "loss": 0.2887,
+      "step": 6420
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.9419536495559886e-05,
+      "loss": 0.0289,
+      "step": 6430
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.9397877409573315e-05,
+      "loss": 0.3283,
+      "step": 6440
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.9376218323586744e-05,
+      "loss": 0.0934,
+      "step": 6450
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.9354559237600172e-05,
+      "loss": 0.1054,
+      "step": 6460
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.9332900151613604e-05,
+      "loss": 0.1365,
+      "step": 6470
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.9311241065627033e-05,
+      "loss": 0.1822,
+      "step": 6480
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.9289581979640462e-05,
+      "loss": 0.1934,
+      "step": 6490
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.9267922893653887e-05,
+      "loss": 0.3957,
+      "step": 6500
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.9246263807667316e-05,
+      "loss": 0.5288,
+      "step": 6510
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.9224604721680744e-05,
+      "loss": 0.348,
+      "step": 6520
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.9202945635694173e-05,
+      "loss": 0.0335,
+      "step": 6530
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.9181286549707602e-05,
+      "loss": 0.2466,
+      "step": 6540
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.915962746372103e-05,
+      "loss": 0.3088,
+      "step": 6550
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.9137968377734463e-05,
+      "loss": 0.0702,
+      "step": 6560
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.911630929174789e-05,
+      "loss": 0.3671,
+      "step": 6570
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.9094650205761317e-05,
+      "loss": 0.1374,
+      "step": 6580
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.9072991119774745e-05,
+      "loss": 0.0949,
+      "step": 6590
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.9051332033788174e-05,
+      "loss": 0.1785,
+      "step": 6600
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.9029672947801603e-05,
+      "loss": 0.1778,
+      "step": 6610
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.900801386181503e-05,
+      "loss": 0.2076,
+      "step": 6620
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.898635477582846e-05,
+      "loss": 0.2231,
+      "step": 6630
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.896469568984189e-05,
+      "loss": 0.2882,
+      "step": 6640
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.894303660385532e-05,
+      "loss": 0.2657,
+      "step": 6650
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.8921377517868746e-05,
+      "loss": 0.1255,
+      "step": 6660
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.8899718431882175e-05,
+      "loss": 0.3541,
+      "step": 6670
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.8878059345895603e-05,
+      "loss": 0.1747,
+      "step": 6680
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.8856400259909032e-05,
+      "loss": 0.2795,
+      "step": 6690
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.883474117392246e-05,
+      "loss": 0.0635,
+      "step": 6700
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.881308208793589e-05,
+      "loss": 0.3951,
+      "step": 6710
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.8791423001949318e-05,
+      "loss": 0.1272,
+      "step": 6720
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.8769763915962747e-05,
+      "loss": 0.0513,
+      "step": 6730
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.8748104829976176e-05,
+      "loss": 0.2131,
+      "step": 6740
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.8726445743989604e-05,
+      "loss": 0.1699,
+      "step": 6750
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.8704786658003033e-05,
+      "loss": 0.3455,
+      "step": 6760
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.868312757201646e-05,
+      "loss": 0.2892,
+      "step": 6770
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.866146848602989e-05,
+      "loss": 0.0785,
+      "step": 6780
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.863980940004332e-05,
+      "loss": 0.2403,
+      "step": 6790
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.8618150314056748e-05,
+      "loss": 0.3012,
+      "step": 6800
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.8596491228070176e-05,
+      "loss": 0.4696,
+      "step": 6810
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.85748321420836e-05,
+      "loss": 0.4096,
+      "step": 6820
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.8553173056097034e-05,
+      "loss": 0.0773,
+      "step": 6830
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.8531513970110462e-05,
+      "loss": 0.1343,
+      "step": 6840
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.850985488412389e-05,
+      "loss": 0.1368,
+      "step": 6850
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.848819579813732e-05,
+      "loss": 0.1681,
+      "step": 6860
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.846653671215075e-05,
+      "loss": 0.3219,
+      "step": 6870
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.8444877626164177e-05,
+      "loss": 0.2319,
+      "step": 6880
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.8423218540177606e-05,
+      "loss": 0.0276,
+      "step": 6890
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.840155945419103e-05,
+      "loss": 0.0599,
+      "step": 6900
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.837990036820446e-05,
+      "loss": 0.2562,
+      "step": 6910
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.8358241282217892e-05,
+      "loss": 0.1269,
+      "step": 6920
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.833658219623132e-05,
+      "loss": 0.1936,
+      "step": 6930
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.831492311024475e-05,
+      "loss": 0.2938,
+      "step": 6940
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.8293264024258178e-05,
+      "loss": 0.1068,
+      "step": 6950
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.8271604938271607e-05,
+      "loss": 0.3149,
+      "step": 6960
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.8249945852285035e-05,
+      "loss": 0.0323,
+      "step": 6970
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.822828676629846e-05,
+      "loss": 0.3232,
+      "step": 6980
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.820662768031189e-05,
+      "loss": 0.0495,
+      "step": 6990
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.8184968594325318e-05,
+      "loss": 0.2286,
+      "step": 7000
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.816330950833875e-05,
+      "loss": 0.1055,
+      "step": 7010
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.814165042235218e-05,
+      "loss": 0.253,
+      "step": 7020
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.8119991336365607e-05,
+      "loss": 0.1701,
+      "step": 7030
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.8098332250379036e-05,
+      "loss": 0.2601,
+      "step": 7040
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.8076673164392465e-05,
+      "loss": 0.2979,
+      "step": 7050
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.805501407840589e-05,
+      "loss": 0.2489,
+      "step": 7060
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.803335499241932e-05,
+      "loss": 0.1344,
+      "step": 7070
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.8011695906432747e-05,
+      "loss": 0.3532,
+      "step": 7080
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.7990036820446176e-05,
+      "loss": 0.1206,
+      "step": 7090
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.7968377734459608e-05,
+      "loss": 0.1082,
+      "step": 7100
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.7946718648473037e-05,
+      "loss": 0.1986,
+      "step": 7110
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.7925059562486466e-05,
+      "loss": 0.1924,
+      "step": 7120
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.7903400476499894e-05,
+      "loss": 0.3774,
+      "step": 7130
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.788174139051332e-05,
+      "loss": 0.1818,
+      "step": 7140
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.7860082304526748e-05,
+      "loss": 0.2385,
+      "step": 7150
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.7838423218540177e-05,
+      "loss": 0.1288,
+      "step": 7160
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.7816764132553606e-05,
+      "loss": 0.4253,
+      "step": 7170
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.7795105046567034e-05,
+      "loss": 0.2336,
+      "step": 7180
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.7773445960580466e-05,
+      "loss": 0.1051,
+      "step": 7190
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.7751786874593895e-05,
+      "loss": 0.1174,
+      "step": 7200
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.7730127788607324e-05,
+      "loss": 0.1564,
+      "step": 7210
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.770846870262075e-05,
+      "loss": 0.1049,
+      "step": 7220
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.7686809616634178e-05,
+      "loss": 0.0617,
+      "step": 7230
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.7665150530647606e-05,
+      "loss": 0.1691,
+      "step": 7240
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7643491444661035e-05,
+      "loss": 0.1695,
+      "step": 7250
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.7621832358674464e-05,
+      "loss": 0.3101,
+      "step": 7260
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.7600173272687892e-05,
+      "loss": 0.0719,
+      "step": 7270
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.7578514186701325e-05,
+      "loss": 0.012,
+      "step": 7280
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.755685510071475e-05,
+      "loss": 0.3291,
+      "step": 7290
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.753519601472818e-05,
+      "loss": 0.3846,
+      "step": 7300
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.7513536928741607e-05,
+      "loss": 0.2574,
+      "step": 7310
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.749404375135369e-05,
+      "loss": 0.1248,
+      "step": 7320
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.747238466536712e-05,
+      "loss": 0.1727,
+      "step": 7330
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.7450725579380552e-05,
+      "loss": 0.1932,
+      "step": 7340
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.742906649339398e-05,
+      "loss": 0.0928,
+      "step": 7350
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.740740740740741e-05,
+      "loss": 0.013,
+      "step": 7360
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.7385748321420838e-05,
+      "loss": 0.2181,
+      "step": 7370
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.7364089235434267e-05,
+      "loss": 0.2038,
+      "step": 7380
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.7342430149447692e-05,
+      "loss": 0.1948,
+      "step": 7390
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.732077106346112e-05,
+      "loss": 0.1502,
+      "step": 7400
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.729911197747455e-05,
+      "loss": 0.1275,
+      "step": 7410
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.7277452891487978e-05,
+      "loss": 0.128,
+      "step": 7420
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.725579380550141e-05,
+      "loss": 0.1376,
+      "step": 7430
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.723413471951484e-05,
+      "loss": 0.1634,
+      "step": 7440
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.7212475633528267e-05,
+      "loss": 0.1918,
+      "step": 7450
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.7190816547541696e-05,
+      "loss": 0.188,
+      "step": 7460
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.716915746155512e-05,
+      "loss": 0.2938,
+      "step": 7470
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.714749837556855e-05,
+      "loss": 0.2532,
+      "step": 7480
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.712583928958198e-05,
+      "loss": 0.3224,
+      "step": 7490
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.7104180203595407e-05,
+      "loss": 0.1205,
+      "step": 7500
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.7082521117608836e-05,
+      "loss": 0.26,
+      "step": 7510
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.7060862031622268e-05,
+      "loss": 0.3231,
+      "step": 7520
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.7039202945635697e-05,
+      "loss": 0.3138,
+      "step": 7530
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.7017543859649125e-05,
+      "loss": 0.061,
+      "step": 7540
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.699588477366255e-05,
+      "loss": 0.2703,
+      "step": 7550
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.697422568767598e-05,
+      "loss": 0.117,
+      "step": 7560
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.6952566601689408e-05,
+      "loss": 0.0645,
+      "step": 7570
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.6930907515702837e-05,
+      "loss": 0.198,
+      "step": 7580
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.6909248429716266e-05,
+      "loss": 0.3202,
+      "step": 7590
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.6887589343729694e-05,
+      "loss": 0.1262,
+      "step": 7600
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.6865930257743126e-05,
+      "loss": 0.1715,
+      "step": 7610
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.6844271171756555e-05,
+      "loss": 0.1517,
+      "step": 7620
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.682261208576998e-05,
+      "loss": 0.2405,
+      "step": 7630
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.680095299978341e-05,
+      "loss": 0.4678,
+      "step": 7640
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.6779293913796838e-05,
+      "loss": 0.192,
+      "step": 7650
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.6757634827810266e-05,
+      "loss": 0.2,
+      "step": 7660
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.6735975741823695e-05,
+      "loss": 0.2786,
+      "step": 7670
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.6714316655837124e-05,
+      "loss": 0.1117,
+      "step": 7680
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.6692657569850552e-05,
+      "loss": 0.131,
+      "step": 7690
+    },
+    {
+      "epoch": 5.0,
+      "eval_accuracy": 0.9825206991720331,
+      "eval_loss": 0.06987906247377396,
+      "eval_runtime": 11.6405,
+      "eval_samples_per_second": 186.762,
+      "eval_steps_per_second": 46.733,
+      "step": 7697
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.6670998483863984e-05,
+      "loss": 0.238,
+      "step": 7700
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.664933939787741e-05,
+      "loss": 0.1141,
+      "step": 7710
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.662768031189084e-05,
+      "loss": 0.3504,
+      "step": 7720
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.6606021225904267e-05,
+      "loss": 0.0974,
+      "step": 7730
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.6584362139917696e-05,
+      "loss": 0.1335,
+      "step": 7740
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.6562703053931124e-05,
+      "loss": 0.1289,
+      "step": 7750
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.6541043967944553e-05,
+      "loss": 0.2545,
+      "step": 7760
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.6519384881957982e-05,
+      "loss": 0.2367,
+      "step": 7770
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.649772579597141e-05,
+      "loss": 0.1437,
+      "step": 7780
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.647606670998484e-05,
+      "loss": 0.1017,
+      "step": 7790
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.6454407623998268e-05,
+      "loss": 0.2109,
+      "step": 7800
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.6432748538011697e-05,
+      "loss": 0.2916,
+      "step": 7810
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.6411089452025125e-05,
+      "loss": 0.2007,
+      "step": 7820
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.6389430366038554e-05,
+      "loss": 0.2244,
+      "step": 7830
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.6367771280051983e-05,
+      "loss": 0.0468,
+      "step": 7840
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.634611219406541e-05,
+      "loss": 0.1024,
+      "step": 7850
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.632445310807884e-05,
+      "loss": 0.1671,
+      "step": 7860
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.6302794022092265e-05,
+      "loss": 0.1837,
+      "step": 7870
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.6281134936105697e-05,
+      "loss": 0.1667,
+      "step": 7880
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1.6259475850119126e-05,
+      "loss": 0.3294,
+      "step": 7890
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1.6237816764132555e-05,
+      "loss": 0.2275,
+      "step": 7900
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.6216157678145983e-05,
+      "loss": 0.1378,
+      "step": 7910
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.6194498592159412e-05,
+      "loss": 0.2246,
+      "step": 7920
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 1.617283950617284e-05,
+      "loss": 0.2126,
+      "step": 7930
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 1.615118042018627e-05,
+      "loss": 0.2265,
+      "step": 7940
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 1.6129521334199695e-05,
+      "loss": 0.2326,
+      "step": 7950
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 1.6107862248213123e-05,
+      "loss": 0.1189,
+      "step": 7960
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 1.6086203162226556e-05,
+      "loss": 0.3122,
+      "step": 7970
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 1.6064544076239984e-05,
+      "loss": 0.1709,
+      "step": 7980
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 1.6042884990253413e-05,
+      "loss": 0.2113,
+      "step": 7990
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 1.602122590426684e-05,
+      "loss": 0.269,
+      "step": 8000
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 1.599956681828027e-05,
+      "loss": 0.0644,
+      "step": 8010
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 1.5977907732293696e-05,
+      "loss": 0.4682,
+      "step": 8020
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 1.5956248646307124e-05,
+      "loss": 0.3193,
+      "step": 8030
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 1.5934589560320553e-05,
+      "loss": 0.1924,
+      "step": 8040
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 1.591293047433398e-05,
+      "loss": 0.2159,
+      "step": 8050
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 1.5891271388347414e-05,
+      "loss": 0.1123,
+      "step": 8060
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 1.5869612302360842e-05,
+      "loss": 0.2937,
+      "step": 8070
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 1.584795321637427e-05,
+      "loss": 0.0673,
+      "step": 8080
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 1.58262941303877e-05,
+      "loss": 0.1607,
+      "step": 8090
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 1.5804635044401125e-05,
+      "loss": 0.1242,
+      "step": 8100
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 1.5782975958414554e-05,
+      "loss": 0.1369,
+      "step": 8110
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 1.5761316872427982e-05,
+      "loss": 0.1622,
+      "step": 8120
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 1.573965778644141e-05,
+      "loss": 0.2463,
+      "step": 8130
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 1.571799870045484e-05,
+      "loss": 0.2972,
+      "step": 8140
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 1.5696339614468272e-05,
+      "loss": 0.287,
+      "step": 8150
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 1.56746805284817e-05,
+      "loss": 0.0314,
+      "step": 8160
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 1.565302144249513e-05,
+      "loss": 0.2538,
+      "step": 8170
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 1.5631362356508555e-05,
+      "loss": 0.2514,
+      "step": 8180
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 1.5609703270521983e-05,
+      "loss": 0.1212,
+      "step": 8190
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 1.5588044184535412e-05,
+      "loss": 0.1789,
+      "step": 8200
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 1.556638509854884e-05,
+      "loss": 0.0955,
+      "step": 8210
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 1.554472601256227e-05,
+      "loss": 0.2357,
+      "step": 8220
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 1.5523066926575698e-05,
+      "loss": 0.175,
+      "step": 8230
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 1.550140784058913e-05,
+      "loss": 0.1586,
+      "step": 8240
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 1.547974875460256e-05,
+      "loss": 0.2546,
+      "step": 8250
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 1.5458089668615984e-05,
+      "loss": 0.3342,
+      "step": 8260
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 1.5436430582629413e-05,
+      "loss": 0.159,
+      "step": 8270
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 1.541477149664284e-05,
+      "loss": 0.1267,
+      "step": 8280
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 1.539311241065627e-05,
+      "loss": 0.2811,
+      "step": 8290
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 1.53714533246697e-05,
+      "loss": 0.204,
+      "step": 8300
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 1.5349794238683127e-05,
+      "loss": 0.1254,
+      "step": 8310
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 1.5328135152696556e-05,
+      "loss": 0.1041,
+      "step": 8320
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 1.5306476066709988e-05,
+      "loss": 0.1072,
+      "step": 8330
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 1.5284816980723414e-05,
+      "loss": 0.3885,
+      "step": 8340
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 1.5263157894736842e-05,
+      "loss": 0.1205,
+      "step": 8350
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 1.5241498808750271e-05,
+      "loss": 0.1424,
+      "step": 8360
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 1.52198397227637e-05,
+      "loss": 0.2023,
+      "step": 8370
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 1.5198180636777128e-05,
+      "loss": 0.106,
+      "step": 8380
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 1.5176521550790557e-05,
+      "loss": 0.2613,
+      "step": 8390
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 1.5154862464803984e-05,
+      "loss": 0.2242,
+      "step": 8400
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 1.5133203378817413e-05,
+      "loss": 0.2247,
+      "step": 8410
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 1.5111544292830845e-05,
+      "loss": 0.0997,
+      "step": 8420
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 1.5089885206844273e-05,
+      "loss": 0.0919,
+      "step": 8430
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 1.50682261208577e-05,
+      "loss": 0.1301,
+      "step": 8440
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 1.5046567034871129e-05,
+      "loss": 0.3214,
+      "step": 8450
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 1.5024907948884558e-05,
+      "loss": 0.3217,
+      "step": 8460
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 1.5003248862897986e-05,
+      "loss": 0.2245,
+      "step": 8470
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 1.4981589776911413e-05,
+      "loss": 0.2558,
+      "step": 8480
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 1.4959930690924844e-05,
+      "loss": 0.1495,
+      "step": 8490
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 1.4938271604938272e-05,
+      "loss": 0.2304,
+      "step": 8500
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 1.4916612518951701e-05,
+      "loss": 0.2046,
+      "step": 8510
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 1.4894953432965128e-05,
+      "loss": 0.0935,
+      "step": 8520
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 1.4873294346978557e-05,
+      "loss": 0.1659,
+      "step": 8530
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 1.4851635260991987e-05,
+      "loss": 0.1549,
+      "step": 8540
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 1.4829976175005416e-05,
+      "loss": 0.2846,
+      "step": 8550
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 1.4808317089018843e-05,
+      "loss": 0.2267,
+      "step": 8560
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 1.4786658003032272e-05,
+      "loss": 0.1766,
+      "step": 8570
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 1.4764998917045702e-05,
+      "loss": 0.1734,
+      "step": 8580
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 1.474333983105913e-05,
+      "loss": 0.08,
+      "step": 8590
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 1.4721680745072558e-05,
+      "loss": 0.1066,
+      "step": 8600
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 1.4700021659085986e-05,
+      "loss": 0.1357,
+      "step": 8610
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 1.4678362573099415e-05,
+      "loss": 0.2295,
+      "step": 8620
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 1.4656703487112845e-05,
+      "loss": 0.2243,
+      "step": 8630
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 1.4635044401126272e-05,
+      "loss": 0.1363,
+      "step": 8640
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 1.4613385315139701e-05,
+      "loss": 0.0751,
+      "step": 8650
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.459172622915313e-05,
+      "loss": 0.1657,
+      "step": 8660
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.457006714316656e-05,
+      "loss": 0.077,
+      "step": 8670
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.4548408057179987e-05,
+      "loss": 0.0533,
+      "step": 8680
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.4526748971193416e-05,
+      "loss": 0.2045,
+      "step": 8690
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.4505089885206844e-05,
+      "loss": 0.1642,
+      "step": 8700
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.4483430799220273e-05,
+      "loss": 0.1926,
+      "step": 8710
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.4461771713233702e-05,
+      "loss": 0.1896,
+      "step": 8720
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.444011262724713e-05,
+      "loss": 0.3106,
+      "step": 8730
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.441845354126056e-05,
+      "loss": 0.0775,
+      "step": 8740
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.4396794455273988e-05,
+      "loss": 0.5138,
+      "step": 8750
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.4375135369287417e-05,
+      "loss": 0.1013,
+      "step": 8760
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.4353476283300845e-05,
+      "loss": 0.3561,
+      "step": 8770
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.4331817197314274e-05,
+      "loss": 0.1639,
+      "step": 8780
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.4310158111327703e-05,
+      "loss": 0.3318,
+      "step": 8790
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.428849902534113e-05,
+      "loss": 0.1585,
+      "step": 8800
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.426683993935456e-05,
+      "loss": 0.028,
+      "step": 8810
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.4245180853367989e-05,
+      "loss": 0.2677,
+      "step": 8820
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 1.4223521767381417e-05,
+      "loss": 0.3382,
+      "step": 8830
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 1.4201862681394844e-05,
+      "loss": 0.2413,
+      "step": 8840
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 1.4180203595408275e-05,
+      "loss": 0.0617,
+      "step": 8850
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 1.4158544509421703e-05,
+      "loss": 0.1472,
+      "step": 8860
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 1.4136885423435132e-05,
+      "loss": 0.246,
+      "step": 8870
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 1.4115226337448559e-05,
+      "loss": 0.3167,
+      "step": 8880
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 1.4093567251461988e-05,
+      "loss": 0.0828,
+      "step": 8890
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 1.4071908165475418e-05,
+      "loss": 0.1955,
+      "step": 8900
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 1.4050249079488847e-05,
+      "loss": 0.1167,
+      "step": 8910
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 1.4028589993502274e-05,
+      "loss": 0.2463,
+      "step": 8920
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 1.4006930907515703e-05,
+      "loss": 0.1647,
+      "step": 8930
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 1.3985271821529133e-05,
+      "loss": 0.1204,
+      "step": 8940
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 1.3963612735542562e-05,
+      "loss": 0.0943,
+      "step": 8950
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 1.3941953649555989e-05,
+      "loss": 0.1074,
+      "step": 8960
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 1.3920294563569417e-05,
+      "loss": 0.0525,
+      "step": 8970
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 1.3898635477582846e-05,
+      "loss": 0.2698,
+      "step": 8980
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 1.3876976391596276e-05,
+      "loss": 0.2737,
+      "step": 8990
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 1.3855317305609703e-05,
+      "loss": 0.1049,
+      "step": 9000
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 1.3833658219623132e-05,
+      "loss": 0.163,
+      "step": 9010
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 1.381199913363656e-05,
+      "loss": 0.0467,
+      "step": 9020
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 1.379034004764999e-05,
+      "loss": 0.1683,
+      "step": 9030
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 1.3768680961663418e-05,
+      "loss": 0.1002,
+      "step": 9040
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 1.3747021875676847e-05,
+      "loss": 0.312,
+      "step": 9050
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 1.3725362789690275e-05,
+      "loss": 0.0603,
+      "step": 9060
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 1.3703703703703704e-05,
+      "loss": 0.2659,
+      "step": 9070
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 1.3682044617717133e-05,
+      "loss": 0.0843,
+      "step": 9080
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 1.3660385531730562e-05,
+      "loss": 0.1908,
+      "step": 9090
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 1.363872644574399e-05,
+      "loss": 0.2324,
+      "step": 9100
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 1.3617067359757417e-05,
+      "loss": 0.0329,
+      "step": 9110
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 1.3595408273770848e-05,
+      "loss": 0.1185,
+      "step": 9120
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 1.3573749187784276e-05,
+      "loss": 0.3367,
+      "step": 9130
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 1.3552090101797705e-05,
+      "loss": 0.0803,
+      "step": 9140
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 1.3530431015811132e-05,
+      "loss": 0.3182,
+      "step": 9150
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 1.350877192982456e-05,
+      "loss": 0.1479,
+      "step": 9160
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 1.3487112843837991e-05,
+      "loss": 0.176,
+      "step": 9170
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 1.346545375785142e-05,
+      "loss": 0.2509,
+      "step": 9180
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 1.3443794671864847e-05,
+      "loss": 0.2482,
+      "step": 9190
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 1.3422135585878275e-05,
+      "loss": 0.104,
+      "step": 9200
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 1.3400476499891706e-05,
+      "loss": 0.1281,
+      "step": 9210
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 1.3378817413905134e-05,
+      "loss": 0.2495,
+      "step": 9220
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.3357158327918561e-05,
+      "loss": 0.2269,
+      "step": 9230
+    },
+    {
+      "epoch": 6.0,
+      "eval_accuracy": 0.9816007359705612,
+      "eval_loss": 0.09182796627283096,
+      "eval_runtime": 11.4083,
+      "eval_samples_per_second": 190.564,
+      "eval_steps_per_second": 47.685,
+      "step": 9237
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.333549924193199e-05,
+      "loss": 0.1493,
+      "step": 9240
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.3313840155945419e-05,
+      "loss": 0.2749,
+      "step": 9250
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.3292181069958849e-05,
+      "loss": 0.2435,
+      "step": 9260
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.3270521983972276e-05,
+      "loss": 0.3642,
+      "step": 9270
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.3248862897985705e-05,
+      "loss": 0.0964,
+      "step": 9280
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.3227203811999134e-05,
+      "loss": 0.1436,
+      "step": 9290
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.3205544726012564e-05,
+      "loss": 0.154,
+      "step": 9300
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 1.3183885640025991e-05,
+      "loss": 0.2918,
+      "step": 9310
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 1.316222655403942e-05,
+      "loss": 0.1256,
+      "step": 9320
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 1.3140567468052848e-05,
+      "loss": 0.1899,
+      "step": 9330
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 1.3118908382066277e-05,
+      "loss": 0.3193,
+      "step": 9340
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 1.3097249296079706e-05,
+      "loss": 0.0036,
+      "step": 9350
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 1.3075590210093134e-05,
+      "loss": 0.2896,
+      "step": 9360
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1.3053931124106563e-05,
+      "loss": 0.0959,
+      "step": 9370
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1.3032272038119992e-05,
+      "loss": 0.1718,
+      "step": 9380
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 1.301061295213342e-05,
+      "loss": 0.0043,
+      "step": 9390
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 1.2988953866146849e-05,
+      "loss": 0.061,
+      "step": 9400
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 1.2967294780160278e-05,
+      "loss": 0.1192,
+      "step": 9410
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 1.2945635694173706e-05,
+      "loss": 0.1323,
+      "step": 9420
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 1.2923976608187133e-05,
+      "loss": 0.1036,
+      "step": 9430
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 1.2902317522200564e-05,
+      "loss": 0.0225,
+      "step": 9440
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 1.2880658436213992e-05,
+      "loss": 0.3399,
+      "step": 9450
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 1.2858999350227421e-05,
+      "loss": 0.1479,
+      "step": 9460
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 1.2837340264240848e-05,
+      "loss": 0.2047,
+      "step": 9470
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 1.2815681178254279e-05,
+      "loss": 0.2818,
+      "step": 9480
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 1.2794022092267707e-05,
+      "loss": 0.1306,
+      "step": 9490
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 1.2772363006281136e-05,
+      "loss": 0.3348,
+      "step": 9500
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 1.2750703920294563e-05,
+      "loss": 0.2451,
+      "step": 9510
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 1.2729044834307992e-05,
+      "loss": 0.1546,
+      "step": 9520
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 1.2707385748321422e-05,
+      "loss": 0.002,
+      "step": 9530
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 1.268572666233485e-05,
+      "loss": 0.3015,
+      "step": 9540
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 1.2664067576348278e-05,
+      "loss": 0.1493,
+      "step": 9550
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 1.2642408490361706e-05,
+      "loss": 0.0039,
+      "step": 9560
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 1.2620749404375137e-05,
+      "loss": 0.2199,
+      "step": 9570
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 1.2599090318388565e-05,
+      "loss": 0.2471,
+      "step": 9580
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 1.2577431232401992e-05,
+      "loss": 0.0337,
+      "step": 9590
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 1.2555772146415421e-05,
+      "loss": 0.2533,
+      "step": 9600
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 1.253411306042885e-05,
+      "loss": 0.0732,
+      "step": 9610
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 1.251245397444228e-05,
+      "loss": 0.0396,
+      "step": 9620
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 1.2490794888455707e-05,
+      "loss": 0.1964,
+      "step": 9630
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 1.2469135802469136e-05,
+      "loss": 0.0912,
+      "step": 9640
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 1.2447476716482564e-05,
+      "loss": 0.1254,
+      "step": 9650
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 1.2425817630495995e-05,
+      "loss": 0.0018,
+      "step": 9660
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 1.2404158544509422e-05,
+      "loss": 0.1022,
+      "step": 9670
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 1.238249945852285e-05,
+      "loss": 0.2673,
+      "step": 9680
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 1.236084037253628e-05,
+      "loss": 0.2274,
+      "step": 9690
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 1.2339181286549708e-05,
+      "loss": 0.0415,
+      "step": 9700
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 1.2317522200563137e-05,
+      "loss": 0.1467,
+      "step": 9710
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 1.2295863114576565e-05,
+      "loss": 0.0534,
+      "step": 9720
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 1.2274204028589994e-05,
+      "loss": 0.4485,
+      "step": 9730
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 1.2252544942603423e-05,
+      "loss": 0.2953,
+      "step": 9740
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 1.2230885856616851e-05,
+      "loss": 0.3026,
+      "step": 9750
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 1.220922677063028e-05,
+      "loss": 0.2313,
+      "step": 9760
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 1.2187567684643709e-05,
+      "loss": 0.2533,
+      "step": 9770
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 1.2165908598657137e-05,
+      "loss": 0.0023,
+      "step": 9780
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 1.2146415421269223e-05,
+      "loss": 0.1422,
+      "step": 9790
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 1.2124756335282652e-05,
+      "loss": 0.1379,
+      "step": 9800
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 1.210309724929608e-05,
+      "loss": 0.1132,
+      "step": 9810
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 1.2081438163309509e-05,
+      "loss": 0.3578,
+      "step": 9820
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 1.2059779077322938e-05,
+      "loss": 0.003,
+      "step": 9830
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 1.2038119991336365e-05,
+      "loss": 0.0103,
+      "step": 9840
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 1.2016460905349795e-05,
+      "loss": 0.0631,
+      "step": 9850
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 1.1994801819363224e-05,
+      "loss": 0.2321,
+      "step": 9860
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1.1973142733376652e-05,
+      "loss": 0.2202,
+      "step": 9870
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.195148364739008e-05,
+      "loss": 0.1058,
+      "step": 9880
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.1929824561403508e-05,
+      "loss": 0.114,
+      "step": 9890
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1.1908165475416938e-05,
+      "loss": 0.303,
+      "step": 9900
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1.1886506389430367e-05,
+      "loss": 0.0623,
+      "step": 9910
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1.1864847303443794e-05,
+      "loss": 0.0327,
+      "step": 9920
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1.1843188217457223e-05,
+      "loss": 0.1601,
+      "step": 9930
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1.1821529131470653e-05,
+      "loss": 0.1608,
+      "step": 9940
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1.1799870045484082e-05,
+      "loss": 0.1325,
+      "step": 9950
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.1778210959497509e-05,
+      "loss": 0.0481,
+      "step": 9960
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 1.1756551873510938e-05,
+      "loss": 0.0913,
+      "step": 9970
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 1.1734892787524366e-05,
+      "loss": 0.1689,
+      "step": 9980
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.1713233701537797e-05,
+      "loss": 0.0024,
+      "step": 9990
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1.1691574615551224e-05,
+      "loss": 0.0703,
+      "step": 10000
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1.1669915529564652e-05,
+      "loss": 0.086,
+      "step": 10010
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.1648256443578081e-05,
+      "loss": 0.1641,
+      "step": 10020
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 1.162659735759151e-05,
+      "loss": 0.1482,
+      "step": 10030
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 1.1604938271604938e-05,
+      "loss": 0.3911,
+      "step": 10040
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1.1583279185618367e-05,
+      "loss": 0.0286,
+      "step": 10050
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1.1561620099631796e-05,
+      "loss": 0.0448,
+      "step": 10060
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.1539961013645224e-05,
+      "loss": 0.0228,
+      "step": 10070
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 1.1518301927658653e-05,
+      "loss": 0.179,
+      "step": 10080
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 1.1496642841672082e-05,
+      "loss": 0.1013,
+      "step": 10090
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.147498375568551e-05,
+      "loss": 0.002,
+      "step": 10100
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1.1453324669698939e-05,
+      "loss": 0.334,
+      "step": 10110
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1.1431665583712366e-05,
+      "loss": 0.2002,
+      "step": 10120
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.1410006497725796e-05,
+      "loss": 0.1392,
+      "step": 10130
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.1388347411739225e-05,
+      "loss": 0.0719,
+      "step": 10140
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.1366688325752654e-05,
+      "loss": 0.0365,
+      "step": 10150
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1.134502923976608e-05,
+      "loss": 0.1452,
+      "step": 10160
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1.1323370153779511e-05,
+      "loss": 0.0416,
+      "step": 10170
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1.130171106779294e-05,
+      "loss": 0.1576,
+      "step": 10180
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1.1280051981806369e-05,
+      "loss": 0.4723,
+      "step": 10190
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.1258392895819796e-05,
+      "loss": 0.0322,
+      "step": 10200
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.1236733809833224e-05,
+      "loss": 0.0891,
+      "step": 10210
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 1.1215074723846655e-05,
+      "loss": 0.0868,
+      "step": 10220
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 1.1193415637860083e-05,
+      "loss": 0.2185,
+      "step": 10230
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 1.117175655187351e-05,
+      "loss": 0.326,
+      "step": 10240
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.1150097465886939e-05,
+      "loss": 0.0354,
+      "step": 10250
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.112843837990037e-05,
+      "loss": 0.0886,
+      "step": 10260
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.1106779293913798e-05,
+      "loss": 0.1274,
+      "step": 10270
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 1.1085120207927225e-05,
+      "loss": 0.0696,
+      "step": 10280
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 1.1063461121940654e-05,
+      "loss": 0.0079,
+      "step": 10290
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 1.1041802035954082e-05,
+      "loss": 0.1214,
+      "step": 10300
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 1.1022308858566168e-05,
+      "loss": 0.1139,
+      "step": 10310
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 1.1000649772579598e-05,
+      "loss": 0.096,
+      "step": 10320
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 1.0978990686593027e-05,
+      "loss": 0.1664,
+      "step": 10330
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 1.0957331600606454e-05,
+      "loss": 0.2343,
+      "step": 10340
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 1.0935672514619883e-05,
+      "loss": 0.1013,
+      "step": 10350
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 1.0914013428633313e-05,
+      "loss": 0.2188,
+      "step": 10360
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 1.0892354342646742e-05,
+      "loss": 0.344,
+      "step": 10370
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 1.0870695256660169e-05,
+      "loss": 0.0253,
+      "step": 10380
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 1.0849036170673597e-05,
+      "loss": 0.0526,
+      "step": 10390
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 1.0827377084687026e-05,
+      "loss": 0.1539,
+      "step": 10400
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 1.0805717998700456e-05,
+      "loss": 0.0685,
+      "step": 10410
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 1.0784058912713883e-05,
+      "loss": 0.1032,
+      "step": 10420
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 1.0762399826727312e-05,
+      "loss": 0.1286,
+      "step": 10430
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 1.074074074074074e-05,
+      "loss": 0.173,
+      "step": 10440
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 1.0719081654754171e-05,
+      "loss": 0.0837,
+      "step": 10450
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 1.0697422568767598e-05,
+      "loss": 0.2937,
+      "step": 10460
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 1.0675763482781027e-05,
+      "loss": 0.1882,
+      "step": 10470
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 1.0654104396794456e-05,
+      "loss": 0.2534,
+      "step": 10480
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 1.0632445310807884e-05,
+      "loss": 0.0982,
+      "step": 10490
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 1.0610786224821313e-05,
+      "loss": 0.2651,
+      "step": 10500
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 1.0589127138834742e-05,
+      "loss": 0.3875,
+      "step": 10510
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 1.056746805284817e-05,
+      "loss": 0.0487,
+      "step": 10520
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 1.0545808966861597e-05,
+      "loss": 0.2208,
+      "step": 10530
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 1.0524149880875028e-05,
+      "loss": 0.2591,
+      "step": 10540
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 1.0502490794888456e-05,
+      "loss": 0.2439,
+      "step": 10550
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 1.0480831708901885e-05,
+      "loss": 0.1268,
+      "step": 10560
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 1.0459172622915312e-05,
+      "loss": 0.2885,
+      "step": 10570
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 1.043751353692874e-05,
+      "loss": 0.1322,
+      "step": 10580
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 1.0415854450942171e-05,
+      "loss": 0.1469,
+      "step": 10590
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 1.03941953649556e-05,
+      "loss": 0.3053,
+      "step": 10600
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 1.0372536278969027e-05,
+      "loss": 0.0234,
+      "step": 10610
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 1.0350877192982455e-05,
+      "loss": 0.1442,
+      "step": 10620
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 1.0329218106995886e-05,
+      "loss": 0.2114,
+      "step": 10630
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 1.0307559021009315e-05,
+      "loss": 0.1937,
+      "step": 10640
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 1.0285899935022741e-05,
+      "loss": 0.0363,
+      "step": 10650
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 1.026424084903617e-05,
+      "loss": 0.045,
+      "step": 10660
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 1.0242581763049599e-05,
+      "loss": 0.0268,
+      "step": 10670
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 1.022092267706303e-05,
+      "loss": 0.0029,
+      "step": 10680
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 1.0199263591076456e-05,
+      "loss": 0.0415,
+      "step": 10690
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 1.0177604505089885e-05,
+      "loss": 0.2679,
+      "step": 10700
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 1.0155945419103314e-05,
+      "loss": 0.0821,
+      "step": 10710
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 1.0134286333116744e-05,
+      "loss": 0.1114,
+      "step": 10720
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 1.0112627247130171e-05,
+      "loss": 0.125,
+      "step": 10730
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 1.00909681611436e-05,
+      "loss": 0.2086,
+      "step": 10740
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 1.0069309075157028e-05,
+      "loss": 0.1832,
+      "step": 10750
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 1.0047649989170457e-05,
+      "loss": 0.0859,
+      "step": 10760
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.0025990903183886e-05,
+      "loss": 0.1406,
+      "step": 10770
+    },
+    {
+      "epoch": 7.0,
+      "eval_accuracy": 0.9820607175712971,
+      "eval_loss": 0.1027088314294815,
+      "eval_runtime": 11.4247,
+      "eval_samples_per_second": 190.289,
+      "eval_steps_per_second": 47.616,
+      "step": 10776
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.0004331817197314e-05,
+      "loss": 0.1335,
+      "step": 10780
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.982672731210743e-06,
+      "loss": 0.2175,
+      "step": 10790
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.961013645224172e-06,
+      "loss": 0.1809,
+      "step": 10800
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.9393545592376e-06,
+      "loss": 0.2377,
+      "step": 10810
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.917695473251029e-06,
+      "loss": 0.0439,
+      "step": 10820
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.896036387264458e-06,
+      "loss": 0.0899,
+      "step": 10830
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.874377301277887e-06,
+      "loss": 0.1037,
+      "step": 10840
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.852718215291314e-06,
+      "loss": 0.1169,
+      "step": 10850
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.831059129304744e-06,
+      "loss": 0.0344,
+      "step": 10860
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.809400043318173e-06,
+      "loss": 0.34,
+      "step": 10870
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.787740957331601e-06,
+      "loss": 0.1443,
+      "step": 10880
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.766081871345028e-06,
+      "loss": 0.0902,
+      "step": 10890
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.744422785358459e-06,
+      "loss": 0.0097,
+      "step": 10900
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.722763699371887e-06,
+      "loss": 0.0021,
+      "step": 10910
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.701104613385316e-06,
+      "loss": 0.1804,
+      "step": 10920
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 9.679445527398743e-06,
+      "loss": 0.135,
+      "step": 10930
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 9.657786441412172e-06,
+      "loss": 0.1276,
+      "step": 10940
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 9.636127355425602e-06,
+      "loss": 0.0291,
+      "step": 10950
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 9.61446826943903e-06,
+      "loss": 0.0351,
+      "step": 10960
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 9.592809183452458e-06,
+      "loss": 0.0214,
+      "step": 10970
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 9.571150097465886e-06,
+      "loss": 0.1521,
+      "step": 10980
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 9.549491011479317e-06,
+      "loss": 0.0016,
+      "step": 10990
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 9.527831925492745e-06,
+      "loss": 0.0278,
+      "step": 11000
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 9.506172839506172e-06,
+      "loss": 0.1415,
+      "step": 11010
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 9.484513753519601e-06,
+      "loss": 0.3208,
+      "step": 11020
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 9.46285466753303e-06,
+      "loss": 0.0843,
+      "step": 11030
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 9.44119558154646e-06,
+      "loss": 0.043,
+      "step": 11040
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 9.419536495559887e-06,
+      "loss": 0.0166,
+      "step": 11050
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 9.397877409573316e-06,
+      "loss": 0.1233,
+      "step": 11060
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 9.376218323586745e-06,
+      "loss": 0.2636,
+      "step": 11070
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 9.354559237600175e-06,
+      "loss": 0.0911,
+      "step": 11080
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 9.332900151613602e-06,
+      "loss": 0.0851,
+      "step": 11090
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 9.31124106562703e-06,
+      "loss": 0.1619,
+      "step": 11100
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 9.28958197964046e-06,
+      "loss": 0.2722,
+      "step": 11110
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 9.267922893653888e-06,
+      "loss": 0.0219,
+      "step": 11120
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 9.246263807667317e-06,
+      "loss": 0.2887,
+      "step": 11130
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 9.224604721680745e-06,
+      "loss": 0.1434,
+      "step": 11140
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 9.202945635694174e-06,
+      "loss": 0.1464,
+      "step": 11150
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 9.181286549707603e-06,
+      "loss": 0.1582,
+      "step": 11160
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 9.159627463721031e-06,
+      "loss": 0.3014,
+      "step": 11170
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 9.13796837773446e-06,
+      "loss": 0.128,
+      "step": 11180
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 9.116309291747889e-06,
+      "loss": 0.0366,
+      "step": 11190
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 9.094650205761317e-06,
+      "loss": 0.0315,
+      "step": 11200
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 9.072991119774744e-06,
+      "loss": 0.1617,
+      "step": 11210
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 9.051332033788175e-06,
+      "loss": 0.14,
+      "step": 11220
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 9.029672947801604e-06,
+      "loss": 0.1742,
+      "step": 11230
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 9.008013861815032e-06,
+      "loss": 0.1645,
+      "step": 11240
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 8.98635477582846e-06,
+      "loss": 0.0676,
+      "step": 11250
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 8.96469568984189e-06,
+      "loss": 0.4134,
+      "step": 11260
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 8.943036603855318e-06,
+      "loss": 0.1698,
+      "step": 11270
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 8.921377517868747e-06,
+      "loss": 0.3438,
+      "step": 11280
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 8.899718431882174e-06,
+      "loss": 0.2901,
+      "step": 11290
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 8.878059345895603e-06,
+      "loss": 0.1935,
+      "step": 11300
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 8.856400259909033e-06,
+      "loss": 0.0899,
+      "step": 11310
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 8.834741173922462e-06,
+      "loss": 0.1031,
+      "step": 11320
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 8.813082087935889e-06,
+      "loss": 0.2409,
+      "step": 11330
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 8.791423001949317e-06,
+      "loss": 0.1731,
+      "step": 11340
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 8.769763915962748e-06,
+      "loss": 0.0284,
+      "step": 11350
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 8.748104829976176e-06,
+      "loss": 0.1007,
+      "step": 11360
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 8.726445743989603e-06,
+      "loss": 0.1167,
+      "step": 11370
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 8.704786658003032e-06,
+      "loss": 0.2883,
+      "step": 11380
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 8.68312757201646e-06,
+      "loss": 0.3288,
+      "step": 11390
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 8.661468486029891e-06,
+      "loss": 0.1414,
+      "step": 11400
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 8.639809400043318e-06,
+      "loss": 0.1391,
+      "step": 11410
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 8.618150314056747e-06,
+      "loss": 0.0034,
+      "step": 11420
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 8.596491228070176e-06,
+      "loss": 0.1287,
+      "step": 11430
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 8.574832142083606e-06,
+      "loss": 0.1158,
+      "step": 11440
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 8.553173056097033e-06,
+      "loss": 0.1224,
+      "step": 11450
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 8.531513970110462e-06,
+      "loss": 0.1694,
+      "step": 11460
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 8.50985488412389e-06,
+      "loss": 0.1216,
+      "step": 11470
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 8.488195798137319e-06,
+      "loss": 0.0039,
+      "step": 11480
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 8.466536712150748e-06,
+      "loss": 0.0075,
+      "step": 11490
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 8.444877626164176e-06,
+      "loss": 0.1198,
+      "step": 11500
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 8.423218540177605e-06,
+      "loss": 0.0194,
+      "step": 11510
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 8.401559454191034e-06,
+      "loss": 0.1048,
+      "step": 11520
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 8.379900368204462e-06,
+      "loss": 0.0203,
+      "step": 11530
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 8.358241282217891e-06,
+      "loss": 0.192,
+      "step": 11540
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 8.33658219623132e-06,
+      "loss": 0.1592,
+      "step": 11550
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 8.314923110244747e-06,
+      "loss": 0.2229,
+      "step": 11560
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 8.293264024258175e-06,
+      "loss": 0.1228,
+      "step": 11570
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 8.271604938271606e-06,
+      "loss": 0.1012,
+      "step": 11580
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 8.249945852285035e-06,
+      "loss": 0.1136,
+      "step": 11590
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 8.228286766298462e-06,
+      "loss": 0.0247,
+      "step": 11600
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 8.20662768031189e-06,
+      "loss": 0.3162,
+      "step": 11610
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 8.184968594325319e-06,
+      "loss": 0.0228,
+      "step": 11620
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 8.16330950833875e-06,
+      "loss": 0.3069,
+      "step": 11630
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 8.141650422352176e-06,
+      "loss": 0.1401,
+      "step": 11640
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 8.119991336365605e-06,
+      "loss": 0.3756,
+      "step": 11650
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 8.098332250379034e-06,
+      "loss": 0.1575,
+      "step": 11660
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 8.076673164392464e-06,
+      "loss": 0.1175,
+      "step": 11670
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 8.055014078405891e-06,
+      "loss": 0.142,
+      "step": 11680
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 8.03335499241932e-06,
+      "loss": 0.1816,
+      "step": 11690
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 8.011695906432748e-06,
+      "loss": 0.0936,
+      "step": 11700
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 7.990036820446177e-06,
+      "loss": 0.1407,
+      "step": 11710
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 7.968377734459606e-06,
+      "loss": 0.0172,
+      "step": 11720
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 7.946718648473034e-06,
+      "loss": 0.0791,
+      "step": 11730
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 7.925059562486463e-06,
+      "loss": 0.1454,
+      "step": 11740
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 7.903400476499892e-06,
+      "loss": 0.1292,
+      "step": 11750
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 7.88174139051332e-06,
+      "loss": 0.129,
+      "step": 11760
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 7.860082304526749e-06,
+      "loss": 0.2011,
+      "step": 11770
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 7.838423218540178e-06,
+      "loss": 0.0254,
+      "step": 11780
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 7.816764132553607e-06,
+      "loss": 0.1652,
+      "step": 11790
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 7.795105046567034e-06,
+      "loss": 0.0554,
+      "step": 11800
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 7.773445960580464e-06,
+      "loss": 0.1178,
+      "step": 11810
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 7.751786874593893e-06,
+      "loss": 0.0668,
+      "step": 11820
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 7.730127788607321e-06,
+      "loss": 0.0845,
+      "step": 11830
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 7.708468702620748e-06,
+      "loss": 0.0826,
+      "step": 11840
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 7.686809616634179e-06,
+      "loss": 0.375,
+      "step": 11850
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 7.665150530647607e-06,
+      "loss": 0.097,
+      "step": 11860
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 7.643491444661036e-06,
+      "loss": 0.0551,
+      "step": 11870
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 7.621832358674464e-06,
+      "loss": 0.3063,
+      "step": 11880
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 7.600173272687892e-06,
+      "loss": 0.0019,
+      "step": 11890
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 7.578514186701322e-06,
+      "loss": 0.1901,
+      "step": 11900
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 7.55685510071475e-06,
+      "loss": 0.1215,
+      "step": 11910
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 7.535196014728179e-06,
+      "loss": 0.039,
+      "step": 11920
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 7.513536928741606e-06,
+      "loss": 0.0444,
+      "step": 11930
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 7.491877842755036e-06,
+      "loss": 0.2903,
+      "step": 11940
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 7.470218756768465e-06,
+      "loss": 0.1258,
+      "step": 11950
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 7.448559670781893e-06,
+      "loss": 0.0981,
+      "step": 11960
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 7.426900584795321e-06,
+      "loss": 0.0701,
+      "step": 11970
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 7.405241498808751e-06,
+      "loss": 0.1823,
+      "step": 11980
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 7.3835824128221785e-06,
+      "loss": 0.1369,
+      "step": 11990
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 7.361923326835608e-06,
+      "loss": 0.098,
+      "step": 12000
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 7.340264240849036e-06,
+      "loss": 0.187,
+      "step": 12010
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 7.3186051548624654e-06,
+      "loss": 0.2385,
+      "step": 12020
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 7.296946068875893e-06,
+      "loss": 0.0271,
+      "step": 12030
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 7.275286982889323e-06,
+      "loss": 0.1092,
+      "step": 12040
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 7.253627896902751e-06,
+      "loss": 0.0626,
+      "step": 12050
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 7.231968810916179e-06,
+      "loss": 0.0399,
+      "step": 12060
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 7.210309724929608e-06,
+      "loss": 0.1757,
+      "step": 12070
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 7.188650638943037e-06,
+      "loss": 0.0704,
+      "step": 12080
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 7.166991552956465e-06,
+      "loss": 0.0874,
+      "step": 12090
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 7.145332466969894e-06,
+      "loss": 0.182,
+      "step": 12100
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 7.123673380983323e-06,
+      "loss": 0.2369,
+      "step": 12110
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 7.1020142949967514e-06,
+      "loss": 0.2545,
+      "step": 12120
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 7.08035520901018e-06,
+      "loss": 0.0413,
+      "step": 12130
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 7.058696123023609e-06,
+      "loss": 0.1008,
+      "step": 12140
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 7.037037037037037e-06,
+      "loss": 0.2507,
+      "step": 12150
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 7.015377951050466e-06,
+      "loss": 0.005,
+      "step": 12160
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.993718865063894e-06,
+      "loss": 0.1919,
+      "step": 12170
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.9720597790773235e-06,
+      "loss": 0.0591,
+      "step": 12180
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.950400693090751e-06,
+      "loss": 0.0858,
+      "step": 12190
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.928741607104181e-06,
+      "loss": 0.0661,
+      "step": 12200
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.907082521117609e-06,
+      "loss": 0.1969,
+      "step": 12210
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.885423435131038e-06,
+      "loss": 0.161,
+      "step": 12220
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.863764349144466e-06,
+      "loss": 0.0023,
+      "step": 12230
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.842105263157895e-06,
+      "loss": 0.0021,
+      "step": 12240
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.8204461771713235e-06,
+      "loss": 0.0316,
+      "step": 12250
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.798787091184752e-06,
+      "loss": 0.1438,
+      "step": 12260
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.777128005198181e-06,
+      "loss": 0.0997,
+      "step": 12270
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.7554689192116095e-06,
+      "loss": 0.1109,
+      "step": 12280
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.733809833225038e-06,
+      "loss": 0.0012,
+      "step": 12290
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.712150747238467e-06,
+      "loss": 0.0729,
+      "step": 12300
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.690491661251896e-06,
+      "loss": 0.0865,
+      "step": 12310
+    },
+    {
+      "epoch": 8.0,
+      "eval_accuracy": 0.983440662373505,
+      "eval_loss": 0.10050991922616959,
+      "eval_runtime": 11.339,
+      "eval_samples_per_second": 191.728,
+      "eval_steps_per_second": 47.976,
+      "step": 12316
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.668832575265324e-06,
+      "loss": 0.0695,
+      "step": 12320
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 6.647173489278752e-06,
+      "loss": 0.1402,
+      "step": 12330
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 6.625514403292182e-06,
+      "loss": 0.1792,
+      "step": 12340
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 6.6038553173056095e-06,
+      "loss": 0.057,
+      "step": 12350
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 6.582196231319039e-06,
+      "loss": 0.1237,
+      "step": 12360
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 6.560537145332467e-06,
+      "loss": 0.1454,
+      "step": 12370
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 6.538878059345896e-06,
+      "loss": 0.3794,
+      "step": 12380
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 6.517218973359324e-06,
+      "loss": 0.0513,
+      "step": 12390
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 6.495559887372754e-06,
+      "loss": 0.1367,
+      "step": 12400
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 6.473900801386182e-06,
+      "loss": 0.1089,
+      "step": 12410
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 6.45224171539961e-06,
+      "loss": 0.1266,
+      "step": 12420
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 6.430582629413039e-06,
+      "loss": 0.0314,
+      "step": 12430
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 6.408923543426468e-06,
+      "loss": 0.187,
+      "step": 12440
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 6.387264457439896e-06,
+      "loss": 0.1011,
+      "step": 12450
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 6.365605371453325e-06,
+      "loss": 0.104,
+      "step": 12460
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 6.343946285466754e-06,
+      "loss": 0.1843,
+      "step": 12470
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 6.322287199480182e-06,
+      "loss": 0.1027,
+      "step": 12480
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 6.300628113493611e-06,
+      "loss": 0.0929,
+      "step": 12490
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 6.27896902750704e-06,
+      "loss": 0.1537,
+      "step": 12500
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 6.257309941520468e-06,
+      "loss": 0.21,
+      "step": 12510
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 6.235650855533897e-06,
+      "loss": 0.2373,
+      "step": 12520
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 6.213991769547325e-06,
+      "loss": 0.2052,
+      "step": 12530
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 6.1923326835607545e-06,
+      "loss": 0.0222,
+      "step": 12540
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 6.170673597574182e-06,
+      "loss": 0.1403,
+      "step": 12550
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 6.149014511587612e-06,
+      "loss": 0.0783,
+      "step": 12560
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 6.12735542560104e-06,
+      "loss": 0.1133,
+      "step": 12570
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 6.105696339614468e-06,
+      "loss": 0.078,
+      "step": 12580
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 6.084037253627897e-06,
+      "loss": 0.1056,
+      "step": 12590
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 6.062378167641326e-06,
+      "loss": 0.0064,
+      "step": 12600
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 6.0407190816547545e-06,
+      "loss": 0.1774,
+      "step": 12610
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 6.019059995668182e-06,
+      "loss": 0.29,
+      "step": 12620
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.997400909681612e-06,
+      "loss": 0.0621,
+      "step": 12630
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.97574182369504e-06,
+      "loss": 0.1282,
+      "step": 12640
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.954082737708469e-06,
+      "loss": 0.1614,
+      "step": 12650
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.932423651721897e-06,
+      "loss": 0.0925,
+      "step": 12660
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.9107645657353266e-06,
+      "loss": 0.0266,
+      "step": 12670
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.889105479748754e-06,
+      "loss": 0.0502,
+      "step": 12680
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.867446393762183e-06,
+      "loss": 0.2139,
+      "step": 12690
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.845787307775612e-06,
+      "loss": 0.1442,
+      "step": 12700
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.8241282217890405e-06,
+      "loss": 0.1269,
+      "step": 12710
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.802469135802469e-06,
+      "loss": 0.0462,
+      "step": 12720
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.780810049815898e-06,
+      "loss": 0.0381,
+      "step": 12730
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.7591509638293265e-06,
+      "loss": 0.0016,
+      "step": 12740
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.737491877842755e-06,
+      "loss": 0.0688,
+      "step": 12750
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.715832791856183e-06,
+      "loss": 0.0817,
+      "step": 12760
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.6941737058696126e-06,
+      "loss": 0.1522,
+      "step": 12770
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.67251461988304e-06,
+      "loss": 0.0022,
+      "step": 12780
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.65085553389647e-06,
+      "loss": 0.1226,
+      "step": 12790
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.629196447909898e-06,
+      "loss": 0.1626,
+      "step": 12800
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.607537361923327e-06,
+      "loss": 0.126,
+      "step": 12810
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.585878275936755e-06,
+      "loss": 0.2298,
+      "step": 12820
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.564219189950185e-06,
+      "loss": 0.0408,
+      "step": 12830
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.5425601039636125e-06,
+      "loss": 0.0436,
+      "step": 12840
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.520901017977041e-06,
+      "loss": 0.0041,
+      "step": 12850
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.49924193199047e-06,
+      "loss": 0.0011,
+      "step": 12860
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.477582846003899e-06,
+      "loss": 0.0818,
+      "step": 12870
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.455923760017327e-06,
+      "loss": 0.1121,
+      "step": 12880
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.434264674030756e-06,
+      "loss": 0.0018,
+      "step": 12890
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.412605588044185e-06,
+      "loss": 0.0731,
+      "step": 12900
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.390946502057613e-06,
+      "loss": 0.1524,
+      "step": 12910
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.369287416071042e-06,
+      "loss": 0.0035,
+      "step": 12920
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 5.347628330084471e-06,
+      "loss": 0.1107,
+      "step": 12930
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.3259692440978985e-06,
+      "loss": 0.1086,
+      "step": 12940
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.304310158111328e-06,
+      "loss": 0.201,
+      "step": 12950
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.282651072124756e-06,
+      "loss": 0.0546,
+      "step": 12960
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.2609919861381854e-06,
+      "loss": 0.1694,
+      "step": 12970
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.239332900151613e-06,
+      "loss": 0.0113,
+      "step": 12980
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.217673814165043e-06,
+      "loss": 0.001,
+      "step": 12990
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.196014728178471e-06,
+      "loss": 0.1616,
+      "step": 13000
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 5.1743556421919e-06,
+      "loss": 0.0243,
+      "step": 13010
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.152696556205328e-06,
+      "loss": 0.2819,
+      "step": 13020
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.131037470218757e-06,
+      "loss": 0.1329,
+      "step": 13030
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 5.109378384232185e-06,
+      "loss": 0.0679,
+      "step": 13040
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 5.087719298245614e-06,
+      "loss": 0.1569,
+      "step": 13050
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 5.066060212259043e-06,
+      "loss": 0.1033,
+      "step": 13060
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 5.0444011262724714e-06,
+      "loss": 0.1112,
+      "step": 13070
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 5.0227420402859e-06,
+      "loss": 0.0879,
+      "step": 13080
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 5.001082954299329e-06,
+      "loss": 0.0839,
+      "step": 13090
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.9794238683127575e-06,
+      "loss": 0.0216,
+      "step": 13100
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.957764782326186e-06,
+      "loss": 0.1372,
+      "step": 13110
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.936105696339614e-06,
+      "loss": 0.224,
+      "step": 13120
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.9144466103530435e-06,
+      "loss": 0.0735,
+      "step": 13130
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.892787524366471e-06,
+      "loss": 0.0528,
+      "step": 13140
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.871128438379901e-06,
+      "loss": 0.1845,
+      "step": 13150
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.849469352393329e-06,
+      "loss": 0.2739,
+      "step": 13160
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.827810266406758e-06,
+      "loss": 0.1506,
+      "step": 13170
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.806151180420186e-06,
+      "loss": 0.0357,
+      "step": 13180
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.784492094433616e-06,
+      "loss": 0.1289,
+      "step": 13190
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.7628330084470435e-06,
+      "loss": 0.2116,
+      "step": 13200
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.741173922460472e-06,
+      "loss": 0.0194,
+      "step": 13210
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.719514836473901e-06,
+      "loss": 0.0915,
+      "step": 13220
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.6978557504873295e-06,
+      "loss": 0.0661,
+      "step": 13230
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.676196664500758e-06,
+      "loss": 0.1411,
+      "step": 13240
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.654537578514187e-06,
+      "loss": 0.0583,
+      "step": 13250
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.632878492527616e-06,
+      "loss": 0.0014,
+      "step": 13260
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.611219406541044e-06,
+      "loss": 0.0801,
+      "step": 13270
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.589560320554473e-06,
+      "loss": 0.0949,
+      "step": 13280
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.567901234567902e-06,
+      "loss": 0.2094,
+      "step": 13290
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.5462421485813295e-06,
+      "loss": 0.1486,
+      "step": 13300
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.524583062594759e-06,
+      "loss": 0.1623,
+      "step": 13310
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.502923976608187e-06,
+      "loss": 0.0248,
+      "step": 13320
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.481264890621616e-06,
+      "loss": 0.045,
+      "step": 13330
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.459605804635044e-06,
+      "loss": 0.062,
+      "step": 13340
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.437946718648474e-06,
+      "loss": 0.1095,
+      "step": 13350
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 4.416287632661902e-06,
+      "loss": 0.0382,
+      "step": 13360
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 4.394628546675331e-06,
+      "loss": 0.1162,
+      "step": 13370
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 4.372969460688759e-06,
+      "loss": 0.3412,
+      "step": 13380
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 4.351310374702188e-06,
+      "loss": 0.1247,
+      "step": 13390
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 4.329651288715616e-06,
+      "loss": 0.1992,
+      "step": 13400
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 4.307992202729045e-06,
+      "loss": 0.0013,
+      "step": 13410
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 4.286333116742474e-06,
+      "loss": 0.0017,
+      "step": 13420
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 4.264674030755902e-06,
+      "loss": 0.1953,
+      "step": 13430
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 4.243014944769331e-06,
+      "loss": 0.0127,
+      "step": 13440
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 4.22135585878276e-06,
+      "loss": 0.1538,
+      "step": 13450
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 4.1996967727961885e-06,
+      "loss": 0.0453,
+      "step": 13460
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 4.178037686809617e-06,
+      "loss": 0.1036,
+      "step": 13470
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 4.156378600823045e-06,
+      "loss": 0.0737,
+      "step": 13480
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 4.1347195148364745e-06,
+      "loss": 0.0063,
+      "step": 13490
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 4.113060428849902e-06,
+      "loss": 0.0592,
+      "step": 13500
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 4.091401342863332e-06,
+      "loss": 0.0254,
+      "step": 13510
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 4.06974225687676e-06,
+      "loss": 0.0824,
+      "step": 13520
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 4.048083170890189e-06,
+      "loss": 0.2222,
+      "step": 13530
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 4.026424084903617e-06,
+      "loss": 0.1123,
+      "step": 13540
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 4.004764998917046e-06,
+      "loss": 0.1423,
+      "step": 13550
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.9831059129304745e-06,
+      "loss": 0.3386,
+      "step": 13560
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.961446826943903e-06,
+      "loss": 0.0493,
+      "step": 13570
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.939787740957332e-06,
+      "loss": 0.2119,
+      "step": 13580
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.9181286549707605e-06,
+      "loss": 0.1862,
+      "step": 13590
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.896469568984189e-06,
+      "loss": 0.1311,
+      "step": 13600
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.874810482997618e-06,
+      "loss": 0.012,
+      "step": 13610
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.8531513970110466e-06,
+      "loss": 0.0577,
+      "step": 13620
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.831492311024474e-06,
+      "loss": 0.1025,
+      "step": 13630
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.809833225037903e-06,
+      "loss": 0.1233,
+      "step": 13640
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.788174139051332e-06,
+      "loss": 0.1714,
+      "step": 13650
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.7665150530647605e-06,
+      "loss": 0.1979,
+      "step": 13660
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.7448559670781896e-06,
+      "loss": 0.0014,
+      "step": 13670
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.723196881091618e-06,
+      "loss": 0.0485,
+      "step": 13680
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.7015377951050465e-06,
+      "loss": 0.1494,
+      "step": 13690
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.679878709118475e-06,
+      "loss": 0.0244,
+      "step": 13700
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.658219623131904e-06,
+      "loss": 0.2303,
+      "step": 13710
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.6365605371453326e-06,
+      "loss": 0.0834,
+      "step": 13720
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.6149014511587613e-06,
+      "loss": 0.1842,
+      "step": 13730
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.59324236517219e-06,
+      "loss": 0.1489,
+      "step": 13740
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.5715832791856186e-06,
+      "loss": 0.2743,
+      "step": 13750
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.5499241931990473e-06,
+      "loss": 0.2174,
+      "step": 13760
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.5282651072124756e-06,
+      "loss": 0.2781,
+      "step": 13770
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.5066060212259043e-06,
+      "loss": 0.0518,
+      "step": 13780
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.484946935239333e-06,
+      "loss": 0.0033,
+      "step": 13790
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.4632878492527616e-06,
+      "loss": 0.047,
+      "step": 13800
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.4416287632661903e-06,
+      "loss": 0.1389,
+      "step": 13810
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.419969677279619e-06,
+      "loss": 0.0018,
+      "step": 13820
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.3983105912930477e-06,
+      "loss": 0.1181,
+      "step": 13830
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.3766515053064764e-06,
+      "loss": 0.2105,
+      "step": 13840
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.354992419319905e-06,
+      "loss": 0.0806,
+      "step": 13850
+    },
+    {
+      "epoch": 9.0,
+      "eval_accuracy": 0.9857405703771849,
+      "eval_loss": 0.08672840148210526,
+      "eval_runtime": 11.3236,
+      "eval_samples_per_second": 191.988,
+      "eval_steps_per_second": 48.041,
+      "step": 13855
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 0.1626,
+      "step": 13860
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.311674247346762e-06,
+      "loss": 0.0091,
+      "step": 13870
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3.2900151613601907e-06,
+      "loss": 0.0625,
+      "step": 13880
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3.2683560753736194e-06,
+      "loss": 0.0985,
+      "step": 13890
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.246696989387048e-06,
+      "loss": 0.122,
+      "step": 13900
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.2250379034004767e-06,
+      "loss": 0.0019,
+      "step": 13910
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.2033788174139054e-06,
+      "loss": 0.0174,
+      "step": 13920
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.181719731427334e-06,
+      "loss": 0.0745,
+      "step": 13930
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.160060645440763e-06,
+      "loss": 0.0324,
+      "step": 13940
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.138401559454191e-06,
+      "loss": 0.0783,
+      "step": 13950
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.1167424734676198e-06,
+      "loss": 0.2326,
+      "step": 13960
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.0950833874810484e-06,
+      "loss": 0.114,
+      "step": 13970
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 3.073424301494477e-06,
+      "loss": 0.0503,
+      "step": 13980
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 3.051765215507906e-06,
+      "loss": 0.0775,
+      "step": 13990
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 3.0301061295213345e-06,
+      "loss": 0.2055,
+      "step": 14000
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 3.008447043534763e-06,
+      "loss": 0.1802,
+      "step": 14010
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 2.986787957548192e-06,
+      "loss": 0.1149,
+      "step": 14020
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 2.9651288715616205e-06,
+      "loss": 0.0518,
+      "step": 14030
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 2.943469785575049e-06,
+      "loss": 0.2972,
+      "step": 14040
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 2.9218106995884775e-06,
+      "loss": 0.0239,
+      "step": 14050
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 2.900151613601906e-06,
+      "loss": 0.0443,
+      "step": 14060
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 2.878492527615335e-06,
+      "loss": 0.1549,
+      "step": 14070
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 2.8568334416287635e-06,
+      "loss": 0.2885,
+      "step": 14080
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 2.8351743556421922e-06,
+      "loss": 0.1295,
+      "step": 14090
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 2.813515269655621e-06,
+      "loss": 0.0811,
+      "step": 14100
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 2.7918561836690496e-06,
+      "loss": 0.0655,
+      "step": 14110
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 2.770197097682478e-06,
+      "loss": 0.1558,
+      "step": 14120
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 2.7485380116959066e-06,
+      "loss": 0.0635,
+      "step": 14130
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 2.7268789257093352e-06,
+      "loss": 0.1408,
+      "step": 14140
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 2.705219839722764e-06,
+      "loss": 0.0554,
+      "step": 14150
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 2.683560753736192e-06,
+      "loss": 0.1879,
+      "step": 14160
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 2.661901667749621e-06,
+      "loss": 0.0448,
+      "step": 14170
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 2.6402425817630496e-06,
+      "loss": 0.0305,
+      "step": 14180
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 2.6185834957764782e-06,
+      "loss": 0.1915,
+      "step": 14190
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 2.596924409789907e-06,
+      "loss": 0.0839,
+      "step": 14200
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 2.575265323803335e-06,
+      "loss": 0.0395,
+      "step": 14210
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 2.553606237816764e-06,
+      "loss": 0.1343,
+      "step": 14220
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 2.5319471518301926e-06,
+      "loss": 0.0914,
+      "step": 14230
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 2.5102880658436212e-06,
+      "loss": 0.027,
+      "step": 14240
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 2.48862897985705e-06,
+      "loss": 0.0023,
+      "step": 14250
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 2.4669698938704786e-06,
+      "loss": 0.0972,
+      "step": 14260
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 2.4453108078839073e-06,
+      "loss": 0.0163,
+      "step": 14270
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 2.423651721897336e-06,
+      "loss": 0.1409,
+      "step": 14280
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 2.4019926359107647e-06,
+      "loss": 0.0516,
+      "step": 14290
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 2.380333549924193e-06,
+      "loss": 0.037,
+      "step": 14300
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 2.3586744639376216e-06,
+      "loss": 0.1219,
+      "step": 14310
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 2.3370153779510503e-06,
+      "loss": 0.0027,
+      "step": 14320
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.315356291964479e-06,
+      "loss": 0.0009,
+      "step": 14330
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.2936972059779077e-06,
+      "loss": 0.0663,
+      "step": 14340
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 2.2720381199913364e-06,
+      "loss": 0.1464,
+      "step": 14350
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.250379034004765e-06,
+      "loss": 0.2329,
+      "step": 14360
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.2287199480181937e-06,
+      "loss": 0.1316,
+      "step": 14370
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 2.2070608620316224e-06,
+      "loss": 0.0296,
+      "step": 14380
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 2.1854017760450507e-06,
+      "loss": 0.0342,
+      "step": 14390
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 2.1637426900584794e-06,
+      "loss": 0.002,
+      "step": 14400
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 2.142083604071908e-06,
+      "loss": 0.0018,
+      "step": 14410
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 2.1204245180853367e-06,
+      "loss": 0.1846,
+      "step": 14420
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 2.0987654320987654e-06,
+      "loss": 0.1093,
+      "step": 14430
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 2.077106346112194e-06,
+      "loss": 0.2368,
+      "step": 14440
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 2.0554472601256228e-06,
+      "loss": 0.1422,
+      "step": 14450
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 2.0337881741390515e-06,
+      "loss": 0.0017,
+      "step": 14460
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 2.0121290881524797e-06,
+      "loss": 0.0543,
+      "step": 14470
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.9904700021659084e-06,
+      "loss": 0.1462,
+      "step": 14480
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.968810916179337e-06,
+      "loss": 0.096,
+      "step": 14490
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.9471518301927658e-06,
+      "loss": 0.1368,
+      "step": 14500
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.9254927442061945e-06,
+      "loss": 0.1442,
+      "step": 14510
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.9038336582196232e-06,
+      "loss": 0.0019,
+      "step": 14520
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.8821745722330518e-06,
+      "loss": 0.2085,
+      "step": 14530
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.8605154862464803e-06,
+      "loss": 0.0484,
+      "step": 14540
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.838856400259909e-06,
+      "loss": 0.0886,
+      "step": 14550
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.8171973142733377e-06,
+      "loss": 0.1149,
+      "step": 14560
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.7955382282867664e-06,
+      "loss": 0.222,
+      "step": 14570
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.7738791423001948e-06,
+      "loss": 0.0016,
+      "step": 14580
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.7522200563136235e-06,
+      "loss": 0.2297,
+      "step": 14590
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.7305609703270522e-06,
+      "loss": 0.1013,
+      "step": 14600
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.708901884340481e-06,
+      "loss": 0.0219,
+      "step": 14610
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.6872427983539096e-06,
+      "loss": 0.147,
+      "step": 14620
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.665583712367338e-06,
+      "loss": 0.2539,
+      "step": 14630
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.6439246263807667e-06,
+      "loss": 0.0522,
+      "step": 14640
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.6222655403941954e-06,
+      "loss": 0.0186,
+      "step": 14650
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.6006064544076241e-06,
+      "loss": 0.0528,
+      "step": 14660
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.5789473684210526e-06,
+      "loss": 0.1753,
+      "step": 14670
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.5572882824344813e-06,
+      "loss": 0.0821,
+      "step": 14680
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.53562919644791e-06,
+      "loss": 0.1442,
+      "step": 14690
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.5139701104613386e-06,
+      "loss": 0.0285,
+      "step": 14700
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.4923110244747673e-06,
+      "loss": 0.0058,
+      "step": 14710
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.4706519384881958e-06,
+      "loss": 0.0685,
+      "step": 14720
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.4489928525016245e-06,
+      "loss": 0.2457,
+      "step": 14730
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.4273337665150532e-06,
+      "loss": 0.0037,
+      "step": 14740
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.4056746805284819e-06,
+      "loss": 0.0015,
+      "step": 14750
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.3840155945419103e-06,
+      "loss": 0.0339,
+      "step": 14760
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.362356508555339e-06,
+      "loss": 0.0442,
+      "step": 14770
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.3406974225687677e-06,
+      "loss": 0.1403,
+      "step": 14780
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.3190383365821964e-06,
+      "loss": 0.178,
+      "step": 14790
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.2973792505956249e-06,
+      "loss": 0.1028,
+      "step": 14800
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.2757201646090535e-06,
+      "loss": 0.2271,
+      "step": 14810
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.2540610786224822e-06,
+      "loss": 0.0674,
+      "step": 14820
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.232401992635911e-06,
+      "loss": 0.0746,
+      "step": 14830
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.2107429066493396e-06,
+      "loss": 0.0413,
+      "step": 14840
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.189083820662768e-06,
+      "loss": 0.086,
+      "step": 14850
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.1674247346761968e-06,
+      "loss": 0.1245,
+      "step": 14860
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.1457656486896254e-06,
+      "loss": 0.0683,
+      "step": 14870
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.1241065627030541e-06,
+      "loss": 0.0196,
+      "step": 14880
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.1024474767164826e-06,
+      "loss": 0.0653,
+      "step": 14890
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.0807883907299113e-06,
+      "loss": 0.0737,
+      "step": 14900
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.05912930474334e-06,
+      "loss": 0.0026,
+      "step": 14910
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.0374702187567687e-06,
+      "loss": 0.0501,
+      "step": 14920
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 1.0158111327701971e-06,
+      "loss": 0.0339,
+      "step": 14930
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 9.941520467836256e-07,
+      "loss": 0.0773,
+      "step": 14940
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 9.724929607970543e-07,
+      "loss": 0.0015,
+      "step": 14950
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 9.508338748104831e-07,
+      "loss": 0.237,
+      "step": 14960
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 9.291747888239117e-07,
+      "loss": 0.1892,
+      "step": 14970
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 9.075157028373403e-07,
+      "loss": 0.1219,
+      "step": 14980
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 8.858566168507689e-07,
+      "loss": 0.0323,
+      "step": 14990
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 8.641975308641976e-07,
+      "loss": 0.001,
+      "step": 15000
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 8.425384448776262e-07,
+      "loss": 0.1438,
+      "step": 15010
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 8.208793588910549e-07,
+      "loss": 0.1015,
+      "step": 15020
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 7.992202729044834e-07,
+      "loss": 0.2792,
+      "step": 15030
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 7.775611869179121e-07,
+      "loss": 0.0015,
+      "step": 15040
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 7.559021009313408e-07,
+      "loss": 0.0411,
+      "step": 15050
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 7.342430149447694e-07,
+      "loss": 0.0828,
+      "step": 15060
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 7.12583928958198e-07,
+      "loss": 0.1128,
+      "step": 15070
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 6.909248429716266e-07,
+      "loss": 0.0015,
+      "step": 15080
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 6.692657569850552e-07,
+      "loss": 0.1064,
+      "step": 15090
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 6.476066709984838e-07,
+      "loss": 0.0865,
+      "step": 15100
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 6.259475850119125e-07,
+      "loss": 0.1159,
+      "step": 15110
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 6.042884990253411e-07,
+      "loss": 0.0683,
+      "step": 15120
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.826294130387698e-07,
+      "loss": 0.1512,
+      "step": 15130
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.609703270521983e-07,
+      "loss": 0.0468,
+      "step": 15140
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.39311241065627e-07,
+      "loss": 0.3834,
+      "step": 15150
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 5.176521550790556e-07,
+      "loss": 0.112,
+      "step": 15160
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 4.959930690924843e-07,
+      "loss": 0.0028,
+      "step": 15170
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 4.74333983105913e-07,
+      "loss": 0.0806,
+      "step": 15180
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 4.5267489711934156e-07,
+      "loss": 0.1619,
+      "step": 15190
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 4.310158111327702e-07,
+      "loss": 0.1508,
+      "step": 15200
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 4.093567251461988e-07,
+      "loss": 0.2488,
+      "step": 15210
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.876976391596275e-07,
+      "loss": 0.1282,
+      "step": 15220
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.6603855317305614e-07,
+      "loss": 0.1004,
+      "step": 15230
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.443794671864848e-07,
+      "loss": 0.0315,
+      "step": 15240
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 3.227203811999134e-07,
+      "loss": 0.1472,
+      "step": 15250
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 3.01061295213342e-07,
+      "loss": 0.1534,
+      "step": 15260
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 2.794022092267706e-07,
+      "loss": 0.0519,
+      "step": 15270
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 2.5774312324019925e-07,
+      "loss": 0.1376,
+      "step": 15280
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 2.3608403725362788e-07,
+      "loss": 0.1193,
+      "step": 15290
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 2.1442495126705652e-07,
+      "loss": 0.0607,
+      "step": 15300
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 1.9276586528048517e-07,
+      "loss": 0.0742,
+      "step": 15310
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.711067792939138e-07,
+      "loss": 0.1991,
+      "step": 15320
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.4944769330734244e-07,
+      "loss": 0.1263,
+      "step": 15330
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.2778860732077107e-07,
+      "loss": 0.1593,
+      "step": 15340
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.061295213341997e-07,
+      "loss": 0.1776,
+      "step": 15350
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 8.447043534762833e-08,
+      "loss": 0.116,
+      "step": 15360
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 6.281134936105695e-08,
+      "loss": 0.1269,
+      "step": 15370
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 4.115226337448559e-08,
+      "loss": 0.1151,
+      "step": 15380
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.949317738791423e-08,
+      "loss": 0.06,
+      "step": 15390
+    },
+    {
+      "epoch": 10.0,
+      "eval_accuracy": 0.984820607175713,
+      "eval_loss": 0.0890473797917366,
+      "eval_runtime": 11.413,
+      "eval_samples_per_second": 190.485,
+      "eval_steps_per_second": 47.665,
+      "step": 15390
+    },
+    {
+      "epoch": 10.0,
+      "step": 15390,
+      "total_flos": 1.1936517061793697e+18,
+      "train_loss": 0.30538128049362784,
+      "train_runtime": 2387.2875,
+      "train_samples_per_second": 51.59,
+      "train_steps_per_second": 6.447
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 15390,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 500,
+  "total_flos": 1.1936517061793697e+18,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}