diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,5812 @@
+{
+  "best_metric": 0.9068514851485149,
+  "best_model_checkpoint": "vit-base-patch16-224-in21k-food101-24-12/checkpoint-9468",
+  "epoch": 11.996198923028192,
+  "eval_steps": 500,
+  "global_step": 9468,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.279831045406548e-07,
+      "loss": 4.6147,
+      "step": 10
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0559662090813095e-06,
+      "loss": 4.6231,
+      "step": 20
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.583949313621964e-06,
+      "loss": 4.6228,
+      "step": 30
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.111932418162619e-06,
+      "loss": 4.6202,
+      "step": 40
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.6399155227032734e-06,
+      "loss": 4.612,
+      "step": 50
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.167898627243928e-06,
+      "loss": 4.6118,
+      "step": 60
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.6958817317845833e-06,
+      "loss": 4.6055,
+      "step": 70
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.223864836325238e-06,
+      "loss": 4.6041,
+      "step": 80
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.751847940865893e-06,
+      "loss": 4.6021,
+      "step": 90
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.279831045406547e-06,
+      "loss": 4.6001,
+      "step": 100
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.807814149947202e-06,
+      "loss": 4.5976,
+      "step": 110
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.335797254487856e-06,
+      "loss": 4.5877,
+      "step": 120
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.863780359028511e-06,
+      "loss": 4.5757,
+      "step": 130
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.3917634635691666e-06,
+      "loss": 4.565,
+      "step": 140
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.91974656810982e-06,
+      "loss": 4.5625,
+      "step": 150
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.447729672650476e-06,
+      "loss": 4.5472,
+      "step": 160
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.97571277719113e-06,
+      "loss": 4.5378,
+      "step": 170
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.503695881731786e-06,
+      "loss": 4.5255,
+      "step": 180
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.003167898627244e-05,
+      "loss": 4.516,
+      "step": 190
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.0559662090813093e-05,
+      "loss": 4.4896,
+      "step": 200
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.108764519535375e-05,
+      "loss": 4.4761,
+      "step": 210
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.1615628299894405e-05,
+      "loss": 4.4641,
+      "step": 220
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.2143611404435058e-05,
+      "loss": 4.4498,
+      "step": 230
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2671594508975712e-05,
+      "loss": 4.416,
+      "step": 240
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.3199577613516367e-05,
+      "loss": 4.3983,
+      "step": 250
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3727560718057022e-05,
+      "loss": 4.3886,
+      "step": 260
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4255543822597678e-05,
+      "loss": 4.3567,
+      "step": 270
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4783526927138333e-05,
+      "loss": 4.339,
+      "step": 280
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.531151003167899e-05,
+      "loss": 4.3166,
+      "step": 290
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.583949313621964e-05,
+      "loss": 4.2816,
+      "step": 300
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6367476240760296e-05,
+      "loss": 4.2569,
+      "step": 310
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6895459345300952e-05,
+      "loss": 4.2333,
+      "step": 320
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7423442449841605e-05,
+      "loss": 4.2023,
+      "step": 330
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.795142555438226e-05,
+      "loss": 4.1584,
+      "step": 340
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8479408658922915e-05,
+      "loss": 4.1363,
+      "step": 350
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.900739176346357e-05,
+      "loss": 4.0936,
+      "step": 360
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9535374868004224e-05,
+      "loss": 4.0692,
+      "step": 370
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.006335797254488e-05,
+      "loss": 4.0299,
+      "step": 380
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0591341077085534e-05,
+      "loss": 4.0009,
+      "step": 390
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.1119324181626187e-05,
+      "loss": 3.9694,
+      "step": 400
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.1647307286166843e-05,
+      "loss": 3.9451,
+      "step": 410
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.21752903907075e-05,
+      "loss": 3.9099,
+      "step": 420
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2703273495248153e-05,
+      "loss": 3.8657,
+      "step": 430
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.323125659978881e-05,
+      "loss": 3.8254,
+      "step": 440
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.3759239704329462e-05,
+      "loss": 3.8052,
+      "step": 450
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4287222808870115e-05,
+      "loss": 3.7684,
+      "step": 460
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4815205913410772e-05,
+      "loss": 3.7279,
+      "step": 470
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.5343189017951425e-05,
+      "loss": 3.6764,
+      "step": 480
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.587117212249208e-05,
+      "loss": 3.6822,
+      "step": 490
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.6399155227032734e-05,
+      "loss": 3.6328,
+      "step": 500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.6927138331573394e-05,
+      "loss": 3.5905,
+      "step": 510
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7455121436114044e-05,
+      "loss": 3.5687,
+      "step": 520
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7983104540654697e-05,
+      "loss": 3.5617,
+      "step": 530
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8511087645195357e-05,
+      "loss": 3.4733,
+      "step": 540
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.903907074973601e-05,
+      "loss": 3.4616,
+      "step": 550
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9567053854276666e-05,
+      "loss": 3.4352,
+      "step": 560
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.009503695881732e-05,
+      "loss": 3.3686,
+      "step": 570
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.062302006335798e-05,
+      "loss": 3.373,
+      "step": 580
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.115100316789863e-05,
+      "loss": 3.3427,
+      "step": 590
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.167898627243928e-05,
+      "loss": 3.3159,
+      "step": 600
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.220696937697994e-05,
+      "loss": 3.299,
+      "step": 610
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.273495248152059e-05,
+      "loss": 3.2549,
+      "step": 620
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.326293558606125e-05,
+      "loss": 3.2241,
+      "step": 630
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.3790918690601904e-05,
+      "loss": 3.1842,
+      "step": 640
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4318901795142554e-05,
+      "loss": 3.2073,
+      "step": 650
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.484688489968321e-05,
+      "loss": 3.1636,
+      "step": 660
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.537486800422387e-05,
+      "loss": 3.1403,
+      "step": 670
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.590285110876452e-05,
+      "loss": 3.095,
+      "step": 680
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.643083421330517e-05,
+      "loss": 3.0526,
+      "step": 690
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.695881731784583e-05,
+      "loss": 3.0167,
+      "step": 700
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.7486800422386486e-05,
+      "loss": 2.9658,
+      "step": 710
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.801478352692714e-05,
+      "loss": 2.9501,
+      "step": 720
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.85427666314678e-05,
+      "loss": 2.8755,
+      "step": 730
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.907074973600845e-05,
+      "loss": 2.9059,
+      "step": 740
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.9598732840549105e-05,
+      "loss": 2.8555,
+      "step": 750
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.012671594508976e-05,
+      "loss": 2.8292,
+      "step": 760
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.065469904963041e-05,
+      "loss": 2.8131,
+      "step": 770
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.118268215417107e-05,
+      "loss": 2.7927,
+      "step": 780
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": 0.7693069306930693,
+      "eval_loss": 2.5628888607025146,
+      "eval_runtime": 399.5969,
+      "eval_samples_per_second": 63.189,
+      "eval_steps_per_second": 2.635,
+      "step": 789
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1710665258711724e-05,
+      "loss": 2.7354,
+      "step": 790
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.2238648363252374e-05,
+      "loss": 2.6798,
+      "step": 800
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.276663146779304e-05,
+      "loss": 2.6512,
+      "step": 810
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.3294614572333687e-05,
+      "loss": 2.6179,
+      "step": 820
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.382259767687434e-05,
+      "loss": 2.6022,
+      "step": 830
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.4350580781415e-05,
+      "loss": 2.5843,
+      "step": 840
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.487856388595565e-05,
+      "loss": 2.5932,
+      "step": 850
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.5406546990496306e-05,
+      "loss": 2.48,
+      "step": 860
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.593453009503696e-05,
+      "loss": 2.4623,
+      "step": 870
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.646251319957762e-05,
+      "loss": 2.457,
+      "step": 880
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.699049630411827e-05,
+      "loss": 2.4305,
+      "step": 890
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.7518479408658925e-05,
+      "loss": 2.4053,
+      "step": 900
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.804646251319958e-05,
+      "loss": 2.3738,
+      "step": 910
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.857444561774023e-05,
+      "loss": 2.3922,
+      "step": 920
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.9102428722280894e-05,
+      "loss": 2.2832,
+      "step": 930
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.9630411826821544e-05,
+      "loss": 2.2324,
+      "step": 940
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.998239643234363e-05,
+      "loss": 2.2539,
+      "step": 950
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.992371787348903e-05,
+      "loss": 2.2092,
+      "step": 960
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.986503931463444e-05,
+      "loss": 2.1349,
+      "step": 970
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.980636075577984e-05,
+      "loss": 2.2159,
+      "step": 980
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.974768219692524e-05,
+      "loss": 2.1872,
+      "step": 990
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.968900363807065e-05,
+      "loss": 2.1451,
+      "step": 1000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.963032507921605e-05,
+      "loss": 2.0441,
+      "step": 1010
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.957164652036146e-05,
+      "loss": 2.1034,
+      "step": 1020
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.951296796150687e-05,
+      "loss": 2.034,
+      "step": 1030
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.945428940265227e-05,
+      "loss": 2.0642,
+      "step": 1040
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.939561084379768e-05,
+      "loss": 2.0603,
+      "step": 1050
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.933693228494308e-05,
+      "loss": 1.9997,
+      "step": 1060
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.927825372608849e-05,
+      "loss": 1.9559,
+      "step": 1070
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.921957516723389e-05,
+      "loss": 1.9944,
+      "step": 1080
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.91608966083793e-05,
+      "loss": 1.8963,
+      "step": 1090
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.91022180495247e-05,
+      "loss": 1.9476,
+      "step": 1100
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.904353949067011e-05,
+      "loss": 1.8979,
+      "step": 1110
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.898486093181552e-05,
+      "loss": 1.8654,
+      "step": 1120
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.892618237296092e-05,
+      "loss": 1.8325,
+      "step": 1130
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.886750381410633e-05,
+      "loss": 1.7831,
+      "step": 1140
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.880882525525173e-05,
+      "loss": 1.8714,
+      "step": 1150
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.8750146696397135e-05,
+      "loss": 1.7569,
+      "step": 1160
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.8691468137542544e-05,
+      "loss": 1.7879,
+      "step": 1170
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.8632789578687945e-05,
+      "loss": 1.7379,
+      "step": 1180
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.8574111019833354e-05,
+      "loss": 1.7148,
+      "step": 1190
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.851543246097876e-05,
+      "loss": 1.7197,
+      "step": 1200
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.8456753902124165e-05,
+      "loss": 1.685,
+      "step": 1210
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.839807534326957e-05,
+      "loss": 1.6869,
+      "step": 1220
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.8339396784414975e-05,
+      "loss": 1.6648,
+      "step": 1230
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.8280718225560384e-05,
+      "loss": 1.5702,
+      "step": 1240
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.8222039666705786e-05,
+      "loss": 1.6431,
+      "step": 1250
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.8163361107851194e-05,
+      "loss": 1.5766,
+      "step": 1260
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.8104682548996596e-05,
+      "loss": 1.6504,
+      "step": 1270
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.8046003990142005e-05,
+      "loss": 1.6049,
+      "step": 1280
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.7987325431287414e-05,
+      "loss": 1.5747,
+      "step": 1290
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.7928646872432815e-05,
+      "loss": 1.5264,
+      "step": 1300
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.7869968313578224e-05,
+      "loss": 1.4999,
+      "step": 1310
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.7811289754723626e-05,
+      "loss": 1.6153,
+      "step": 1320
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.775261119586903e-05,
+      "loss": 1.6118,
+      "step": 1330
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.7693932637014437e-05,
+      "loss": 1.5014,
+      "step": 1340
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.763525407815984e-05,
+      "loss": 1.4627,
+      "step": 1350
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.757657551930525e-05,
+      "loss": 1.4902,
+      "step": 1360
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.7517896960450656e-05,
+      "loss": 1.4635,
+      "step": 1370
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.7459218401596064e-05,
+      "loss": 1.3964,
+      "step": 1380
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.7400539842741466e-05,
+      "loss": 1.4411,
+      "step": 1390
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.734186128388687e-05,
+      "loss": 1.3983,
+      "step": 1400
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.728318272503228e-05,
+      "loss": 1.3967,
+      "step": 1410
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.722450416617768e-05,
+      "loss": 1.4788,
+      "step": 1420
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.716582560732309e-05,
+      "loss": 1.4407,
+      "step": 1430
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.710714704846849e-05,
+      "loss": 1.4165,
+      "step": 1440
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.70484684896139e-05,
+      "loss": 1.3067,
+      "step": 1450
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.6989789930759307e-05,
+      "loss": 1.407,
+      "step": 1460
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.693111137190471e-05,
+      "loss": 1.2977,
+      "step": 1470
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.687243281305012e-05,
+      "loss": 1.3355,
+      "step": 1480
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.681375425419552e-05,
+      "loss": 1.3263,
+      "step": 1490
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.675507569534092e-05,
+      "loss": 1.2952,
+      "step": 1500
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.669639713648633e-05,
+      "loss": 1.3236,
+      "step": 1510
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.663771857763173e-05,
+      "loss": 1.3544,
+      "step": 1520
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.657904001877714e-05,
+      "loss": 1.3005,
+      "step": 1530
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.652036145992255e-05,
+      "loss": 1.2282,
+      "step": 1540
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.646168290106796e-05,
+      "loss": 1.3016,
+      "step": 1550
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.640300434221336e-05,
+      "loss": 1.2905,
+      "step": 1560
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.634432578335876e-05,
+      "loss": 1.256,
+      "step": 1570
+    },
+    {
+      "epoch": 2.0,
+      "eval_accuracy": 0.8583366336633663,
+      "eval_loss": 0.9637454152107239,
+      "eval_runtime": 398.6344,
+      "eval_samples_per_second": 63.341,
+      "eval_steps_per_second": 2.642,
+      "step": 1578
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.628564722450417e-05,
+      "loss": 1.2161,
+      "step": 1580
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.622696866564957e-05,
+      "loss": 1.168,
+      "step": 1590
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.616829010679498e-05,
+      "loss": 1.1527,
+      "step": 1600
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.610961154794038e-05,
+      "loss": 1.0808,
+      "step": 1610
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.605093298908579e-05,
+      "loss": 1.1881,
+      "step": 1620
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.59922544302312e-05,
+      "loss": 1.1699,
+      "step": 1630
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.59335758713766e-05,
+      "loss": 1.1396,
+      "step": 1640
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.587489731252201e-05,
+      "loss": 1.2021,
+      "step": 1650
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.581621875366741e-05,
+      "loss": 1.1732,
+      "step": 1660
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.575754019481282e-05,
+      "loss": 1.1012,
+      "step": 1670
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.569886163595822e-05,
+      "loss": 1.1158,
+      "step": 1680
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.5640183077103624e-05,
+      "loss": 1.0248,
+      "step": 1690
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.558150451824903e-05,
+      "loss": 1.1069,
+      "step": 1700
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.552282595939444e-05,
+      "loss": 1.1218,
+      "step": 1710
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.546414740053985e-05,
+      "loss": 1.1292,
+      "step": 1720
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.540546884168525e-05,
+      "loss": 1.0524,
+      "step": 1730
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.5346790282830654e-05,
+      "loss": 1.0723,
+      "step": 1740
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.528811172397606e-05,
+      "loss": 1.1118,
+      "step": 1750
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.5229433165121465e-05,
+      "loss": 1.1128,
+      "step": 1760
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.517075460626687e-05,
+      "loss": 1.0061,
+      "step": 1770
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.5112076047412275e-05,
+      "loss": 1.1072,
+      "step": 1780
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.5053397488557684e-05,
+      "loss": 1.0507,
+      "step": 1790
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.499471892970309e-05,
+      "loss": 1.0552,
+      "step": 1800
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.4936040370848494e-05,
+      "loss": 1.0334,
+      "step": 1810
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.48773618119939e-05,
+      "loss": 1.0176,
+      "step": 1820
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.4818683253139305e-05,
+      "loss": 0.9674,
+      "step": 1830
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.4760004694284714e-05,
+      "loss": 1.012,
+      "step": 1840
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.4701326135430115e-05,
+      "loss": 1.0345,
+      "step": 1850
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.464264757657552e-05,
+      "loss": 1.0688,
+      "step": 1860
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.4583969017720926e-05,
+      "loss": 1.0523,
+      "step": 1870
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.452529045886633e-05,
+      "loss": 1.0648,
+      "step": 1880
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.4466611900011736e-05,
+      "loss": 0.9767,
+      "step": 1890
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.4407933341157145e-05,
+      "loss": 1.0085,
+      "step": 1900
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.434925478230255e-05,
+      "loss": 0.9688,
+      "step": 1910
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.4290576223447956e-05,
+      "loss": 1.0456,
+      "step": 1920
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.423189766459336e-05,
+      "loss": 0.9582,
+      "step": 1930
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.4173219105738766e-05,
+      "loss": 0.9926,
+      "step": 1940
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.411454054688417e-05,
+      "loss": 0.9717,
+      "step": 1950
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.405586198802958e-05,
+      "loss": 0.945,
+      "step": 1960
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.399718342917498e-05,
+      "loss": 1.0254,
+      "step": 1970
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.393850487032039e-05,
+      "loss": 1.0495,
+      "step": 1980
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.3879826311465796e-05,
+      "loss": 0.9646,
+      "step": 1990
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.38211477526112e-05,
+      "loss": 1.0159,
+      "step": 2000
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.3762469193756606e-05,
+      "loss": 0.915,
+      "step": 2010
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.370379063490201e-05,
+      "loss": 0.9125,
+      "step": 2020
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.364511207604741e-05,
+      "loss": 0.9624,
+      "step": 2030
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.358643351719282e-05,
+      "loss": 0.9378,
+      "step": 2040
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.352775495833822e-05,
+      "loss": 0.8956,
+      "step": 2050
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.346907639948363e-05,
+      "loss": 0.9906,
+      "step": 2060
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.341039784062904e-05,
+      "loss": 0.9479,
+      "step": 2070
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.335171928177445e-05,
+      "loss": 0.935,
+      "step": 2080
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.329304072291985e-05,
+      "loss": 0.8705,
+      "step": 2090
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.323436216406525e-05,
+      "loss": 0.9452,
+      "step": 2100
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.317568360521066e-05,
+      "loss": 0.8945,
+      "step": 2110
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.311700504635606e-05,
+      "loss": 0.9872,
+      "step": 2120
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.305832648750147e-05,
+      "loss": 0.8879,
+      "step": 2130
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.299964792864687e-05,
+      "loss": 0.8667,
+      "step": 2140
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.294096936979228e-05,
+      "loss": 0.8999,
+      "step": 2150
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.288229081093769e-05,
+      "loss": 0.9437,
+      "step": 2160
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.282361225208309e-05,
+      "loss": 0.9705,
+      "step": 2170
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.27649336932285e-05,
+      "loss": 0.8751,
+      "step": 2180
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.27062551343739e-05,
+      "loss": 0.9009,
+      "step": 2190
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.26475765755193e-05,
+      "loss": 0.908,
+      "step": 2200
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.258889801666471e-05,
+      "loss": 0.8844,
+      "step": 2210
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.2530219457810114e-05,
+      "loss": 0.9049,
+      "step": 2220
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.247154089895552e-05,
+      "loss": 0.8996,
+      "step": 2230
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.241286234010093e-05,
+      "loss": 0.8758,
+      "step": 2240
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.235418378124634e-05,
+      "loss": 0.8942,
+      "step": 2250
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.229550522239174e-05,
+      "loss": 0.919,
+      "step": 2260
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.2236826663537143e-05,
+      "loss": 0.7982,
+      "step": 2270
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.217814810468255e-05,
+      "loss": 0.8421,
+      "step": 2280
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.2119469545827954e-05,
+      "loss": 0.8665,
+      "step": 2290
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.206079098697336e-05,
+      "loss": 0.8542,
+      "step": 2300
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.2002112428118765e-05,
+      "loss": 0.9499,
+      "step": 2310
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.194343386926417e-05,
+      "loss": 0.8377,
+      "step": 2320
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.188475531040958e-05,
+      "loss": 0.9819,
+      "step": 2330
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.1826076751554984e-05,
+      "loss": 0.8884,
+      "step": 2340
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.176739819270039e-05,
+      "loss": 0.8134,
+      "step": 2350
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.1708719633845794e-05,
+      "loss": 0.94,
+      "step": 2360
+    },
+    {
+      "epoch": 3.0,
+      "eval_accuracy": 0.8816237623762376,
+      "eval_loss": 0.5866235494613647,
+      "eval_runtime": 397.7982,
+      "eval_samples_per_second": 63.474,
+      "eval_steps_per_second": 2.647,
+      "step": 2367
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.16500410749912e-05,
+      "loss": 0.7009,
+      "step": 2370
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.1591362516136605e-05,
+      "loss": 0.7794,
+      "step": 2380
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.153268395728201e-05,
+      "loss": 0.7997,
+      "step": 2390
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.1474005398427415e-05,
+      "loss": 0.8026,
+      "step": 2400
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.1415326839572824e-05,
+      "loss": 0.8137,
+      "step": 2410
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.135664828071823e-05,
+      "loss": 0.8701,
+      "step": 2420
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 4.1297969721863635e-05,
+      "loss": 0.7893,
+      "step": 2430
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 4.1239291163009036e-05,
+      "loss": 0.7492,
+      "step": 2440
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 4.1180612604154445e-05,
+      "loss": 0.8424,
+      "step": 2450
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 4.112193404529985e-05,
+      "loss": 0.7778,
+      "step": 2460
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 4.1063255486445256e-05,
+      "loss": 0.7002,
+      "step": 2470
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 4.100457692759066e-05,
+      "loss": 0.7029,
+      "step": 2480
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 4.0945898368736066e-05,
+      "loss": 0.7728,
+      "step": 2490
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 4.0887219809881475e-05,
+      "loss": 0.7492,
+      "step": 2500
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 4.082854125102688e-05,
+      "loss": 0.7589,
+      "step": 2510
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 4.0769862692172285e-05,
+      "loss": 0.7791,
+      "step": 2520
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 4.071118413331769e-05,
+      "loss": 0.7909,
+      "step": 2530
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 4.0652505574463096e-05,
+      "loss": 0.707,
+      "step": 2540
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 4.05938270156085e-05,
+      "loss": 0.7868,
+      "step": 2550
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 4.05351484567539e-05,
+      "loss": 0.768,
+      "step": 2560
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 4.047646989789931e-05,
+      "loss": 0.7974,
+      "step": 2570
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 4.041779133904472e-05,
+      "loss": 0.7644,
+      "step": 2580
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 4.0359112780190126e-05,
+      "loss": 0.812,
+      "step": 2590
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 4.030043422133553e-05,
+      "loss": 0.7008,
+      "step": 2600
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 4.024175566248093e-05,
+      "loss": 0.7948,
+      "step": 2610
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 4.018307710362634e-05,
+      "loss": 0.7443,
+      "step": 2620
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 4.012439854477174e-05,
+      "loss": 0.7368,
+      "step": 2630
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 4.006571998591715e-05,
+      "loss": 0.669,
+      "step": 2640
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 4.000704142706255e-05,
+      "loss": 0.7136,
+      "step": 2650
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.994836286820796e-05,
+      "loss": 0.7708,
+      "step": 2660
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.988968430935337e-05,
+      "loss": 0.753,
+      "step": 2670
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.983100575049877e-05,
+      "loss": 0.7805,
+      "step": 2680
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.977232719164418e-05,
+      "loss": 0.7353,
+      "step": 2690
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.971364863278958e-05,
+      "loss": 0.7871,
+      "step": 2700
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.965497007393499e-05,
+      "loss": 0.8076,
+      "step": 2710
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.959629151508039e-05,
+      "loss": 0.7463,
+      "step": 2720
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.953761295622579e-05,
+      "loss": 0.74,
+      "step": 2730
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.94789343973712e-05,
+      "loss": 0.7354,
+      "step": 2740
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.942025583851661e-05,
+      "loss": 0.7276,
+      "step": 2750
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.936157727966202e-05,
+      "loss": 0.7652,
+      "step": 2760
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.930289872080742e-05,
+      "loss": 0.7685,
+      "step": 2770
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.924422016195282e-05,
+      "loss": 0.7673,
+      "step": 2780
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.918554160309823e-05,
+      "loss": 0.7349,
+      "step": 2790
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.912686304424363e-05,
+      "loss": 0.6692,
+      "step": 2800
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.906818448538904e-05,
+      "loss": 0.7466,
+      "step": 2810
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.9009505926534443e-05,
+      "loss": 0.7739,
+      "step": 2820
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.895082736767985e-05,
+      "loss": 0.7284,
+      "step": 2830
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.8892148808825254e-05,
+      "loss": 0.687,
+      "step": 2840
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.883347024997066e-05,
+      "loss": 0.7028,
+      "step": 2850
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.877479169111607e-05,
+      "loss": 0.7053,
+      "step": 2860
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.871611313226147e-05,
+      "loss": 0.6891,
+      "step": 2870
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.865743457340688e-05,
+      "loss": 0.7159,
+      "step": 2880
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.8598756014552284e-05,
+      "loss": 0.7206,
+      "step": 2890
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.8540077455697686e-05,
+      "loss": 0.7356,
+      "step": 2900
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.8481398896843094e-05,
+      "loss": 0.7692,
+      "step": 2910
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.8422720337988496e-05,
+      "loss": 0.7041,
+      "step": 2920
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.8364041779133905e-05,
+      "loss": 0.6699,
+      "step": 2930
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.830536322027931e-05,
+      "loss": 0.7576,
+      "step": 2940
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.824668466142472e-05,
+      "loss": 0.6532,
+      "step": 2950
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.8188006102570124e-05,
+      "loss": 0.691,
+      "step": 2960
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.8129327543715526e-05,
+      "loss": 0.7073,
+      "step": 2970
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.8070648984860934e-05,
+      "loss": 0.7765,
+      "step": 2980
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.8011970426006336e-05,
+      "loss": 0.7148,
+      "step": 2990
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.7953291867151745e-05,
+      "loss": 0.7015,
+      "step": 3000
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.789461330829715e-05,
+      "loss": 0.6254,
+      "step": 3010
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.7835934749442556e-05,
+      "loss": 0.7135,
+      "step": 3020
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.7777256190587964e-05,
+      "loss": 0.6446,
+      "step": 3030
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.7718577631733366e-05,
+      "loss": 0.7521,
+      "step": 3040
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.7659899072878775e-05,
+      "loss": 0.6959,
+      "step": 3050
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.7601220514024177e-05,
+      "loss": 0.6661,
+      "step": 3060
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.754254195516958e-05,
+      "loss": 0.7255,
+      "step": 3070
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.748386339631499e-05,
+      "loss": 0.6324,
+      "step": 3080
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.742518483746039e-05,
+      "loss": 0.5931,
+      "step": 3090
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.73665062786058e-05,
+      "loss": 0.7603,
+      "step": 3100
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.7307827719751206e-05,
+      "loss": 0.6433,
+      "step": 3110
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.7249149160896615e-05,
+      "loss": 0.7281,
+      "step": 3120
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.719047060204202e-05,
+      "loss": 0.71,
+      "step": 3130
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.713179204318742e-05,
+      "loss": 0.7434,
+      "step": 3140
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.707311348433283e-05,
+      "loss": 0.6693,
+      "step": 3150
+    },
+    {
+      "epoch": 4.0,
+      "eval_accuracy": 0.8887524752475248,
+      "eval_loss": 0.4751880466938019,
+      "eval_runtime": 400.2361,
+      "eval_samples_per_second": 63.088,
+      "eval_steps_per_second": 2.631,
+      "step": 3157
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.701443492547823e-05,
+      "loss": 0.6859,
+      "step": 3160
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.695575636662364e-05,
+      "loss": 0.6522,
+      "step": 3170
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.689707780776904e-05,
+      "loss": 0.6272,
+      "step": 3180
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.683839924891445e-05,
+      "loss": 0.6705,
+      "step": 3190
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.677972069005986e-05,
+      "loss": 0.5972,
+      "step": 3200
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.672104213120526e-05,
+      "loss": 0.656,
+      "step": 3210
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.666236357235067e-05,
+      "loss": 0.5419,
+      "step": 3220
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.660368501349607e-05,
+      "loss": 0.7054,
+      "step": 3230
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.654500645464148e-05,
+      "loss": 0.6405,
+      "step": 3240
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.648632789578688e-05,
+      "loss": 0.6705,
+      "step": 3250
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.642764933693228e-05,
+      "loss": 0.5764,
+      "step": 3260
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.636897077807769e-05,
+      "loss": 0.6162,
+      "step": 3270
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.63102922192231e-05,
+      "loss": 0.6354,
+      "step": 3280
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.625161366036851e-05,
+      "loss": 0.581,
+      "step": 3290
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.619293510151391e-05,
+      "loss": 0.5885,
+      "step": 3300
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.613425654265931e-05,
+      "loss": 0.5734,
+      "step": 3310
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.607557798380472e-05,
+      "loss": 0.6123,
+      "step": 3320
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.601689942495012e-05,
+      "loss": 0.6978,
+      "step": 3330
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.595822086609553e-05,
+      "loss": 0.6384,
+      "step": 3340
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.589954230724093e-05,
+      "loss": 0.6202,
+      "step": 3350
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 3.584086374838634e-05,
+      "loss": 0.5992,
+      "step": 3360
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 3.578218518953175e-05,
+      "loss": 0.6408,
+      "step": 3370
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 3.572350663067715e-05,
+      "loss": 0.642,
+      "step": 3380
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 3.566482807182256e-05,
+      "loss": 0.5322,
+      "step": 3390
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 3.560614951296796e-05,
+      "loss": 0.7348,
+      "step": 3400
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 3.554747095411337e-05,
+      "loss": 0.6946,
+      "step": 3410
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 3.548879239525877e-05,
+      "loss": 0.6063,
+      "step": 3420
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 3.5430113836404175e-05,
+      "loss": 0.6099,
+      "step": 3430
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 3.5371435277549584e-05,
+      "loss": 0.6036,
+      "step": 3440
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 3.531275671869499e-05,
+      "loss": 0.6293,
+      "step": 3450
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 3.52540781598404e-05,
+      "loss": 0.6839,
+      "step": 3460
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.51953996009858e-05,
+      "loss": 0.5901,
+      "step": 3470
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.5136721042131205e-05,
+      "loss": 0.6573,
+      "step": 3480
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.507804248327661e-05,
+      "loss": 0.6118,
+      "step": 3490
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3.5019363924422015e-05,
+      "loss": 0.6166,
+      "step": 3500
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.4960685365567424e-05,
+      "loss": 0.5605,
+      "step": 3510
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.4902006806712826e-05,
+      "loss": 0.6295,
+      "step": 3520
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 3.4843328247858234e-05,
+      "loss": 0.6447,
+      "step": 3530
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.478464968900364e-05,
+      "loss": 0.5863,
+      "step": 3540
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.4725971130149045e-05,
+      "loss": 0.6358,
+      "step": 3550
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.4667292571294454e-05,
+      "loss": 0.5512,
+      "step": 3560
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3.4608614012439855e-05,
+      "loss": 0.6379,
+      "step": 3570
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 3.4549935453585264e-05,
+      "loss": 0.6645,
+      "step": 3580
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 3.4491256894730666e-05,
+      "loss": 0.6096,
+      "step": 3590
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.443257833587607e-05,
+      "loss": 0.6416,
+      "step": 3600
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.4373899777021477e-05,
+      "loss": 0.6241,
+      "step": 3610
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4315221218166885e-05,
+      "loss": 0.6421,
+      "step": 3620
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.4256542659312294e-05,
+      "loss": 0.5394,
+      "step": 3630
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.4197864100457696e-05,
+      "loss": 0.6055,
+      "step": 3640
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.4139185541603104e-05,
+      "loss": 0.5995,
+      "step": 3650
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.4080506982748506e-05,
+      "loss": 0.6037,
+      "step": 3660
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.402182842389391e-05,
+      "loss": 0.6083,
+      "step": 3670
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.396314986503932e-05,
+      "loss": 0.6421,
+      "step": 3680
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.390447130618472e-05,
+      "loss": 0.5975,
+      "step": 3690
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 3.384579274733013e-05,
+      "loss": 0.5539,
+      "step": 3700
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 3.3787114188475536e-05,
+      "loss": 0.6149,
+      "step": 3710
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 3.372843562962094e-05,
+      "loss": 0.6477,
+      "step": 3720
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.3669757070766347e-05,
+      "loss": 0.5987,
+      "step": 3730
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.361107851191175e-05,
+      "loss": 0.5889,
+      "step": 3740
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.355239995305716e-05,
+      "loss": 0.5686,
+      "step": 3750
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.349372139420256e-05,
+      "loss": 0.6086,
+      "step": 3760
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 3.343504283534796e-05,
+      "loss": 0.6452,
+      "step": 3770
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 3.337636427649337e-05,
+      "loss": 0.5817,
+      "step": 3780
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 3.331768571763877e-05,
+      "loss": 0.677,
+      "step": 3790
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 3.325900715878418e-05,
+      "loss": 0.631,
+      "step": 3800
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 3.320032859992959e-05,
+      "loss": 0.5995,
+      "step": 3810
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 3.3141650041075e-05,
+      "loss": 0.6198,
+      "step": 3820
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 3.30829714822204e-05,
+      "loss": 0.5988,
+      "step": 3830
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.30242929233658e-05,
+      "loss": 0.6271,
+      "step": 3840
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.296561436451121e-05,
+      "loss": 0.6977,
+      "step": 3850
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.290693580565661e-05,
+      "loss": 0.6019,
+      "step": 3860
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.284825724680202e-05,
+      "loss": 0.5495,
+      "step": 3870
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.278957868794742e-05,
+      "loss": 0.6431,
+      "step": 3880
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 3.273090012909283e-05,
+      "loss": 0.6352,
+      "step": 3890
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 3.267222157023824e-05,
+      "loss": 0.6219,
+      "step": 3900
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 3.261354301138364e-05,
+      "loss": 0.6272,
+      "step": 3910
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.255486445252905e-05,
+      "loss": 0.5295,
+      "step": 3920
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 3.249618589367445e-05,
+      "loss": 0.5747,
+      "step": 3930
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.243750733481986e-05,
+      "loss": 0.6337,
+      "step": 3940
+    },
+    {
+      "epoch": 5.0,
+      "eval_accuracy": 0.8940990099009901,
+      "eval_loss": 0.4281545579433441,
+      "eval_runtime": 399.633,
+      "eval_samples_per_second": 63.183,
+      "eval_steps_per_second": 2.635,
+      "step": 3946
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.237882877596526e-05,
+      "loss": 0.6057,
+      "step": 3950
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.2320150217110664e-05,
+      "loss": 0.5504,
+      "step": 3960
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.226147165825607e-05,
+      "loss": 0.5645,
+      "step": 3970
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.220279309940148e-05,
+      "loss": 0.5612,
+      "step": 3980
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.214411454054689e-05,
+      "loss": 0.549,
+      "step": 3990
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 3.208543598169229e-05,
+      "loss": 0.5561,
+      "step": 4000
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 3.2026757422837694e-05,
+      "loss": 0.514,
+      "step": 4010
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 3.19680788639831e-05,
+      "loss": 0.566,
+      "step": 4020
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 3.1909400305128505e-05,
+      "loss": 0.5244,
+      "step": 4030
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 3.185072174627391e-05,
+      "loss": 0.5024,
+      "step": 4040
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 3.1792043187419315e-05,
+      "loss": 0.5153,
+      "step": 4050
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 3.1733364628564724e-05,
+      "loss": 0.5276,
+      "step": 4060
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 3.167468606971013e-05,
+      "loss": 0.646,
+      "step": 4070
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 3.1616007510855534e-05,
+      "loss": 0.4689,
+      "step": 4080
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 3.155732895200094e-05,
+      "loss": 0.5348,
+      "step": 4090
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 3.1498650393146345e-05,
+      "loss": 0.5281,
+      "step": 4100
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 3.1439971834291754e-05,
+      "loss": 0.5507,
+      "step": 4110
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 3.1381293275437155e-05,
+      "loss": 0.5525,
+      "step": 4120
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 3.132261471658256e-05,
+      "loss": 0.5458,
+      "step": 4130
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 3.1263936157727966e-05,
+      "loss": 0.5005,
+      "step": 4140
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 3.1205257598873375e-05,
+      "loss": 0.5344,
+      "step": 4150
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 3.114657904001878e-05,
+      "loss": 0.5037,
+      "step": 4160
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 3.1087900481164185e-05,
+      "loss": 0.5588,
+      "step": 4170
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 3.102922192230959e-05,
+      "loss": 0.4994,
+      "step": 4180
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 3.0970543363454996e-05,
+      "loss": 0.5198,
+      "step": 4190
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 3.09118648046004e-05,
+      "loss": 0.5528,
+      "step": 4200
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 3.0853186245745806e-05,
+      "loss": 0.5593,
+      "step": 4210
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 3.079450768689121e-05,
+      "loss": 0.6275,
+      "step": 4220
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 3.073582912803662e-05,
+      "loss": 0.6394,
+      "step": 4230
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 3.0677150569182025e-05,
+      "loss": 0.5277,
+      "step": 4240
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 3.061847201032743e-05,
+      "loss": 0.502,
+      "step": 4250
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 3.0559793451472836e-05,
+      "loss": 0.5482,
+      "step": 4260
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 3.0501114892618238e-05,
+      "loss": 0.5681,
+      "step": 4270
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 3.0442436333763646e-05,
+      "loss": 0.565,
+      "step": 4280
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 3.038375777490905e-05,
+      "loss": 0.6308,
+      "step": 4290
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 3.0325079216054454e-05,
+      "loss": 0.5295,
+      "step": 4300
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 3.0266400657199862e-05,
+      "loss": 0.6258,
+      "step": 4310
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 3.0207722098345264e-05,
+      "loss": 0.5078,
+      "step": 4320
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 3.0149043539490673e-05,
+      "loss": 0.4899,
+      "step": 4330
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 3.0090364980636078e-05,
+      "loss": 0.5722,
+      "step": 4340
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 3.003168642178148e-05,
+      "loss": 0.5403,
+      "step": 4350
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 2.997300786292689e-05,
+      "loss": 0.5412,
+      "step": 4360
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 2.991432930407229e-05,
+      "loss": 0.4943,
+      "step": 4370
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 2.98556507452177e-05,
+      "loss": 0.5221,
+      "step": 4380
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 2.9796972186363104e-05,
+      "loss": 0.5392,
+      "step": 4390
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 2.9738293627508513e-05,
+      "loss": 0.5885,
+      "step": 4400
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 2.9679615068653915e-05,
+      "loss": 0.5397,
+      "step": 4410
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 2.962093650979932e-05,
+      "loss": 0.5002,
+      "step": 4420
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 2.956225795094473e-05,
+      "loss": 0.5483,
+      "step": 4430
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 2.950357939209013e-05,
+      "loss": 0.54,
+      "step": 4440
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 2.944490083323554e-05,
+      "loss": 0.5448,
+      "step": 4450
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 2.938622227438094e-05,
+      "loss": 0.5804,
+      "step": 4460
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 2.9327543715526347e-05,
+      "loss": 0.6176,
+      "step": 4470
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 2.9268865156671755e-05,
+      "loss": 0.5161,
+      "step": 4480
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 2.9210186597817157e-05,
+      "loss": 0.5581,
+      "step": 4490
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 2.9151508038962566e-05,
+      "loss": 0.5202,
+      "step": 4500
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 2.9092829480107968e-05,
+      "loss": 0.5369,
+      "step": 4510
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 2.903415092125338e-05,
+      "loss": 0.6328,
+      "step": 4520
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 2.897547236239878e-05,
+      "loss": 0.5194,
+      "step": 4530
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 2.8916793803544183e-05,
+      "loss": 0.5051,
+      "step": 4540
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 2.8858115244689592e-05,
+      "loss": 0.5527,
+      "step": 4550
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 2.8799436685834997e-05,
+      "loss": 0.5404,
+      "step": 4560
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 2.8740758126980406e-05,
+      "loss": 0.5163,
+      "step": 4570
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 2.8682079568125808e-05,
+      "loss": 0.5168,
+      "step": 4580
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 2.862340100927121e-05,
+      "loss": 0.5083,
+      "step": 4590
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 2.856472245041662e-05,
+      "loss": 0.6172,
+      "step": 4600
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 2.8506043891562024e-05,
+      "loss": 0.5151,
+      "step": 4610
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 2.8447365332707432e-05,
+      "loss": 0.4484,
+      "step": 4620
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 2.8388686773852834e-05,
+      "loss": 0.5801,
+      "step": 4630
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 2.8330008214998243e-05,
+      "loss": 0.4488,
+      "step": 4640
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 2.8271329656143648e-05,
+      "loss": 0.5608,
+      "step": 4650
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 2.821265109728905e-05,
+      "loss": 0.5098,
+      "step": 4660
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 2.815397253843446e-05,
+      "loss": 0.6013,
+      "step": 4670
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 2.809529397957986e-05,
+      "loss": 0.5053,
+      "step": 4680
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 2.803661542072527e-05,
+      "loss": 0.5561,
+      "step": 4690
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 2.7977936861870675e-05,
+      "loss": 0.5603,
+      "step": 4700
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 2.7919258303016076e-05,
+      "loss": 0.5952,
+      "step": 4710
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 2.7860579744161485e-05,
+      "loss": 0.5311,
+      "step": 4720
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 2.780190118530689e-05,
+      "loss": 0.5811,
+      "step": 4730
+    },
+    {
+      "epoch": 6.0,
+      "eval_accuracy": 0.8948514851485149,
+      "eval_loss": 0.4110336899757385,
+      "eval_runtime": 399.0734,
+      "eval_samples_per_second": 63.272,
+      "eval_steps_per_second": 2.639,
+      "step": 4735
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 2.77432226264523e-05,
+      "loss": 0.5915,
+      "step": 4740
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 2.76845440675977e-05,
+      "loss": 0.4746,
+      "step": 4750
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 2.7625865508743103e-05,
+      "loss": 0.443,
+      "step": 4760
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 2.756718694988851e-05,
+      "loss": 0.545,
+      "step": 4770
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 2.7508508391033917e-05,
+      "loss": 0.4424,
+      "step": 4780
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 2.7449829832179325e-05,
+      "loss": 0.5913,
+      "step": 4790
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 2.7391151273324727e-05,
+      "loss": 0.4616,
+      "step": 4800
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 2.7332472714470136e-05,
+      "loss": 0.4867,
+      "step": 4810
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 2.727379415561554e-05,
+      "loss": 0.5181,
+      "step": 4820
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 2.7215115596760943e-05,
+      "loss": 0.4644,
+      "step": 4830
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 2.715643703790635e-05,
+      "loss": 0.5018,
+      "step": 4840
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 2.7097758479051754e-05,
+      "loss": 0.5754,
+      "step": 4850
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 2.7039079920197162e-05,
+      "loss": 0.4975,
+      "step": 4860
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 2.6980401361342567e-05,
+      "loss": 0.5089,
+      "step": 4870
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 2.692172280248797e-05,
+      "loss": 0.5003,
+      "step": 4880
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 2.6863044243633378e-05,
+      "loss": 0.5471,
+      "step": 4890
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 2.6804365684778783e-05,
+      "loss": 0.5086,
+      "step": 4900
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 2.6745687125924192e-05,
+      "loss": 0.5087,
+      "step": 4910
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 2.6687008567069594e-05,
+      "loss": 0.5217,
+      "step": 4920
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 2.6628330008215002e-05,
+      "loss": 0.5462,
+      "step": 4930
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 2.6569651449360404e-05,
+      "loss": 0.5394,
+      "step": 4940
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 2.651097289050581e-05,
+      "loss": 0.4623,
+      "step": 4950
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 2.6452294331651218e-05,
+      "loss": 0.4837,
+      "step": 4960
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 2.639361577279662e-05,
+      "loss": 0.5101,
+      "step": 4970
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 2.633493721394203e-05,
+      "loss": 0.4952,
+      "step": 4980
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 2.627625865508743e-05,
+      "loss": 0.4469,
+      "step": 4990
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 2.6217580096232836e-05,
+      "loss": 0.5455,
+      "step": 5000
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 2.6158901537378245e-05,
+      "loss": 0.509,
+      "step": 5010
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 2.6100222978523646e-05,
+      "loss": 0.4723,
+      "step": 5020
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 2.6041544419669055e-05,
+      "loss": 0.4704,
+      "step": 5030
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 2.598286586081446e-05,
+      "loss": 0.5105,
+      "step": 5040
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 2.5924187301959862e-05,
+      "loss": 0.5207,
+      "step": 5050
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 2.586550874310527e-05,
+      "loss": 0.4927,
+      "step": 5060
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 2.5806830184250673e-05,
+      "loss": 0.3989,
+      "step": 5070
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 2.574815162539608e-05,
+      "loss": 0.478,
+      "step": 5080
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 2.5689473066541487e-05,
+      "loss": 0.491,
+      "step": 5090
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 2.5630794507686895e-05,
+      "loss": 0.524,
+      "step": 5100
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 2.5572115948832297e-05,
+      "loss": 0.5145,
+      "step": 5110
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 2.5513437389977703e-05,
+      "loss": 0.4945,
+      "step": 5120
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 2.545475883112311e-05,
+      "loss": 0.4927,
+      "step": 5130
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 2.5396080272268513e-05,
+      "loss": 0.4799,
+      "step": 5140
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 2.5337401713413922e-05,
+      "loss": 0.4589,
+      "step": 5150
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 2.5278723154559324e-05,
+      "loss": 0.5086,
+      "step": 5160
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 2.522004459570473e-05,
+      "loss": 0.4755,
+      "step": 5170
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 2.5161366036850138e-05,
+      "loss": 0.5135,
+      "step": 5180
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 2.510268747799554e-05,
+      "loss": 0.5279,
+      "step": 5190
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 2.5044008919140948e-05,
+      "loss": 0.4465,
+      "step": 5200
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 2.4985330360286353e-05,
+      "loss": 0.4646,
+      "step": 5210
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 2.492665180143176e-05,
+      "loss": 0.491,
+      "step": 5220
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 2.4867973242577164e-05,
+      "loss": 0.4852,
+      "step": 5230
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 2.480929468372257e-05,
+      "loss": 0.5151,
+      "step": 5240
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 2.4750616124867974e-05,
+      "loss": 0.517,
+      "step": 5250
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 2.469193756601338e-05,
+      "loss": 0.4847,
+      "step": 5260
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 2.4633259007158785e-05,
+      "loss": 0.4843,
+      "step": 5270
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 2.457458044830419e-05,
+      "loss": 0.4689,
+      "step": 5280
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 2.4515901889449596e-05,
+      "loss": 0.5427,
+      "step": 5290
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 2.4457223330595004e-05,
+      "loss": 0.4895,
+      "step": 5300
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 2.4398544771740406e-05,
+      "loss": 0.5295,
+      "step": 5310
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 2.433986621288581e-05,
+      "loss": 0.4804,
+      "step": 5320
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 2.4281187654031217e-05,
+      "loss": 0.4771,
+      "step": 5330
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 2.4222509095176625e-05,
+      "loss": 0.5101,
+      "step": 5340
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 2.416383053632203e-05,
+      "loss": 0.4895,
+      "step": 5350
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 2.4105151977467436e-05,
+      "loss": 0.5428,
+      "step": 5360
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 2.4046473418612838e-05,
+      "loss": 0.4345,
+      "step": 5370
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 2.3987794859758246e-05,
+      "loss": 0.4833,
+      "step": 5380
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 2.392911630090365e-05,
+      "loss": 0.5177,
+      "step": 5390
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 2.3870437742049057e-05,
+      "loss": 0.4318,
+      "step": 5400
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 2.3811759183194462e-05,
+      "loss": 0.4722,
+      "step": 5410
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 2.3753080624339867e-05,
+      "loss": 0.5538,
+      "step": 5420
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 2.3694402065485273e-05,
+      "loss": 0.5049,
+      "step": 5430
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 2.3635723506630678e-05,
+      "loss": 0.4614,
+      "step": 5440
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 2.3577044947776083e-05,
+      "loss": 0.5144,
+      "step": 5450
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 2.351836638892149e-05,
+      "loss": 0.4543,
+      "step": 5460
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 2.3459687830066894e-05,
+      "loss": 0.5473,
+      "step": 5470
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 2.3401009271212302e-05,
+      "loss": 0.4606,
+      "step": 5480
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 2.3342330712357704e-05,
+      "loss": 0.4681,
+      "step": 5490
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 2.328365215350311e-05,
+      "loss": 0.4435,
+      "step": 5500
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 2.3224973594648515e-05,
+      "loss": 0.5062,
+      "step": 5510
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 2.3166295035793923e-05,
+      "loss": 0.4661,
+      "step": 5520
+    },
+    {
+      "epoch": 7.0,
+      "eval_accuracy": 0.899009900990099,
+      "eval_loss": 0.3875064253807068,
+      "eval_runtime": 398.0016,
+      "eval_samples_per_second": 63.442,
+      "eval_steps_per_second": 2.646,
+      "step": 5524
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 2.310761647693933e-05,
+      "loss": 0.451,
+      "step": 5530
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 2.3048937918084734e-05,
+      "loss": 0.4419,
+      "step": 5540
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 2.2990259359230136e-05,
+      "loss": 0.4326,
+      "step": 5550
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 2.2931580800375545e-05,
+      "loss": 0.4206,
+      "step": 5560
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 2.287290224152095e-05,
+      "loss": 0.4767,
+      "step": 5570
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 2.2814223682666355e-05,
+      "loss": 0.4463,
+      "step": 5580
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 2.275554512381176e-05,
+      "loss": 0.4266,
+      "step": 5590
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 2.2696866564957166e-05,
+      "loss": 0.4439,
+      "step": 5600
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 2.263818800610257e-05,
+      "loss": 0.4738,
+      "step": 5610
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 2.2579509447247976e-05,
+      "loss": 0.4568,
+      "step": 5620
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 2.252083088839338e-05,
+      "loss": 0.4254,
+      "step": 5630
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 2.2462152329538787e-05,
+      "loss": 0.4281,
+      "step": 5640
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 2.2403473770684195e-05,
+      "loss": 0.4449,
+      "step": 5650
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 2.2344795211829597e-05,
+      "loss": 0.4455,
+      "step": 5660
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 2.2286116652975002e-05,
+      "loss": 0.5091,
+      "step": 5670
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 2.2227438094120408e-05,
+      "loss": 0.4475,
+      "step": 5680
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 2.2168759535265816e-05,
+      "loss": 0.4746,
+      "step": 5690
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 2.211008097641122e-05,
+      "loss": 0.4871,
+      "step": 5700
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 2.2051402417556627e-05,
+      "loss": 0.4612,
+      "step": 5710
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 2.199272385870203e-05,
+      "loss": 0.4244,
+      "step": 5720
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 2.1934045299847437e-05,
+      "loss": 0.4546,
+      "step": 5730
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 2.1875366740992843e-05,
+      "loss": 0.5052,
+      "step": 5740
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 2.1816688182138248e-05,
+      "loss": 0.4638,
+      "step": 5750
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 2.1758009623283653e-05,
+      "loss": 0.4235,
+      "step": 5760
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 2.169933106442906e-05,
+      "loss": 0.4229,
+      "step": 5770
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 2.1640652505574464e-05,
+      "loss": 0.4181,
+      "step": 5780
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 2.158197394671987e-05,
+      "loss": 0.4372,
+      "step": 5790
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 2.1523295387865274e-05,
+      "loss": 0.4154,
+      "step": 5800
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 2.146461682901068e-05,
+      "loss": 0.4541,
+      "step": 5810
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 2.1405938270156088e-05,
+      "loss": 0.4455,
+      "step": 5820
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 2.1347259711301494e-05,
+      "loss": 0.4253,
+      "step": 5830
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 2.1288581152446895e-05,
+      "loss": 0.4025,
+      "step": 5840
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 2.12299025935923e-05,
+      "loss": 0.4508,
+      "step": 5850
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 2.117122403473771e-05,
+      "loss": 0.4527,
+      "step": 5860
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 2.1112545475883115e-05,
+      "loss": 0.4835,
+      "step": 5870
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 2.105386691702852e-05,
+      "loss": 0.4361,
+      "step": 5880
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 2.0995188358173925e-05,
+      "loss": 0.5039,
+      "step": 5890
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 2.093650979931933e-05,
+      "loss": 0.4428,
+      "step": 5900
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 2.0877831240464736e-05,
+      "loss": 0.4287,
+      "step": 5910
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 2.081915268161014e-05,
+      "loss": 0.4131,
+      "step": 5920
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 2.0760474122755546e-05,
+      "loss": 0.4163,
+      "step": 5930
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 2.070179556390095e-05,
+      "loss": 0.4156,
+      "step": 5940
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 2.0643117005046357e-05,
+      "loss": 0.4234,
+      "step": 5950
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 2.0584438446191762e-05,
+      "loss": 0.4639,
+      "step": 5960
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 2.0525759887337167e-05,
+      "loss": 0.4222,
+      "step": 5970
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 2.0467081328482573e-05,
+      "loss": 0.4369,
+      "step": 5980
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 2.0408402769627978e-05,
+      "loss": 0.4265,
+      "step": 5990
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 2.0349724210773387e-05,
+      "loss": 0.469,
+      "step": 6000
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 2.029104565191879e-05,
+      "loss": 0.4551,
+      "step": 6010
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 2.0232367093064194e-05,
+      "loss": 0.3986,
+      "step": 6020
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 2.01736885342096e-05,
+      "loss": 0.4173,
+      "step": 6030
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 2.0115009975355008e-05,
+      "loss": 0.4276,
+      "step": 6040
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 2.0056331416500413e-05,
+      "loss": 0.4404,
+      "step": 6050
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 1.9997652857645818e-05,
+      "loss": 0.4663,
+      "step": 6060
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 1.993897429879122e-05,
+      "loss": 0.4579,
+      "step": 6070
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 1.988029573993663e-05,
+      "loss": 0.5365,
+      "step": 6080
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 1.9821617181082034e-05,
+      "loss": 0.4825,
+      "step": 6090
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 1.976293862222744e-05,
+      "loss": 0.4686,
+      "step": 6100
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 1.9704260063372844e-05,
+      "loss": 0.4432,
+      "step": 6110
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 1.964558150451825e-05,
+      "loss": 0.4674,
+      "step": 6120
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 1.9586902945663655e-05,
+      "loss": 0.4021,
+      "step": 6130
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 1.952822438680906e-05,
+      "loss": 0.4318,
+      "step": 6140
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 1.9469545827954466e-05,
+      "loss": 0.4271,
+      "step": 6150
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 1.941086726909987e-05,
+      "loss": 0.4012,
+      "step": 6160
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 1.935218871024528e-05,
+      "loss": 0.4851,
+      "step": 6170
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 1.9293510151390685e-05,
+      "loss": 0.4893,
+      "step": 6180
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 1.9234831592536087e-05,
+      "loss": 0.4081,
+      "step": 6190
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.9176153033681492e-05,
+      "loss": 0.5092,
+      "step": 6200
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.91174744748269e-05,
+      "loss": 0.4617,
+      "step": 6210
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 1.9058795915972306e-05,
+      "loss": 0.5053,
+      "step": 6220
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 1.900011735711771e-05,
+      "loss": 0.4463,
+      "step": 6230
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 1.8941438798263113e-05,
+      "loss": 0.4513,
+      "step": 6240
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 1.888276023940852e-05,
+      "loss": 0.5137,
+      "step": 6250
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 1.8824081680553927e-05,
+      "loss": 0.4225,
+      "step": 6260
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 1.8765403121699332e-05,
+      "loss": 0.484,
+      "step": 6270
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 1.8706724562844737e-05,
+      "loss": 0.3924,
+      "step": 6280
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 1.8648046003990143e-05,
+      "loss": 0.4079,
+      "step": 6290
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 1.8589367445135548e-05,
+      "loss": 0.4435,
+      "step": 6300
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 1.8530688886280953e-05,
+      "loss": 0.4188,
+      "step": 6310
+    },
+    {
+      "epoch": 8.0,
+      "eval_accuracy": 0.901029702970297,
+      "eval_loss": 0.37761184573173523,
+      "eval_runtime": 400.9775,
+      "eval_samples_per_second": 62.971,
+      "eval_steps_per_second": 2.626,
+      "step": 6314
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.847201032742636e-05,
+      "loss": 0.4137,
+      "step": 6320
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.8413331768571764e-05,
+      "loss": 0.4184,
+      "step": 6330
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.8354653209717172e-05,
+      "loss": 0.4634,
+      "step": 6340
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 1.8295974650862578e-05,
+      "loss": 0.3844,
+      "step": 6350
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 1.823729609200798e-05,
+      "loss": 0.387,
+      "step": 6360
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 1.8178617533153385e-05,
+      "loss": 0.4523,
+      "step": 6370
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 1.8119938974298793e-05,
+      "loss": 0.4189,
+      "step": 6380
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 1.80612604154442e-05,
+      "loss": 0.4148,
+      "step": 6390
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 1.8002581856589604e-05,
+      "loss": 0.4416,
+      "step": 6400
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 1.794390329773501e-05,
+      "loss": 0.413,
+      "step": 6410
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 1.788522473888041e-05,
+      "loss": 0.3593,
+      "step": 6420
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 1.782654618002582e-05,
+      "loss": 0.4098,
+      "step": 6430
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 1.7767867621171225e-05,
+      "loss": 0.3993,
+      "step": 6440
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 1.770918906231663e-05,
+      "loss": 0.474,
+      "step": 6450
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 1.7650510503462036e-05,
+      "loss": 0.3588,
+      "step": 6460
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 1.759183194460744e-05,
+      "loss": 0.4746,
+      "step": 6470
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 1.7533153385752846e-05,
+      "loss": 0.4232,
+      "step": 6480
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 1.747447482689825e-05,
+      "loss": 0.3597,
+      "step": 6490
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 1.7415796268043657e-05,
+      "loss": 0.4162,
+      "step": 6500
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 1.7357117709189062e-05,
+      "loss": 0.4151,
+      "step": 6510
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 1.729843915033447e-05,
+      "loss": 0.3933,
+      "step": 6520
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 1.7239760591479876e-05,
+      "loss": 0.41,
+      "step": 6530
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 1.7181082032625278e-05,
+      "loss": 0.4193,
+      "step": 6540
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 1.7122403473770683e-05,
+      "loss": 0.4264,
+      "step": 6550
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 1.7063724914916092e-05,
+      "loss": 0.3798,
+      "step": 6560
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 1.7005046356061497e-05,
+      "loss": 0.4259,
+      "step": 6570
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 1.6946367797206902e-05,
+      "loss": 0.441,
+      "step": 6580
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 1.6887689238352304e-05,
+      "loss": 0.4133,
+      "step": 6590
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 1.6829010679497713e-05,
+      "loss": 0.3513,
+      "step": 6600
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 1.6770332120643118e-05,
+      "loss": 0.327,
+      "step": 6610
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 1.6711653561788523e-05,
+      "loss": 0.3869,
+      "step": 6620
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 1.665297500293393e-05,
+      "loss": 0.4627,
+      "step": 6630
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 1.6594296444079334e-05,
+      "loss": 0.3824,
+      "step": 6640
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 1.653561788522474e-05,
+      "loss": 0.4204,
+      "step": 6650
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 1.6476939326370144e-05,
+      "loss": 0.4003,
+      "step": 6660
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 1.641826076751555e-05,
+      "loss": 0.3647,
+      "step": 6670
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 1.6359582208660955e-05,
+      "loss": 0.3628,
+      "step": 6680
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 1.6300903649806364e-05,
+      "loss": 0.4142,
+      "step": 6690
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 1.624222509095177e-05,
+      "loss": 0.3856,
+      "step": 6700
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 1.618354653209717e-05,
+      "loss": 0.4179,
+      "step": 6710
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 1.6124867973242576e-05,
+      "loss": 0.3969,
+      "step": 6720
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 1.6066189414387985e-05,
+      "loss": 0.4278,
+      "step": 6730
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 1.600751085553339e-05,
+      "loss": 0.4499,
+      "step": 6740
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 1.5948832296678795e-05,
+      "loss": 0.3762,
+      "step": 6750
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 1.58901537378242e-05,
+      "loss": 0.3735,
+      "step": 6760
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 1.5831475178969606e-05,
+      "loss": 0.3661,
+      "step": 6770
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 1.577279662011501e-05,
+      "loss": 0.4218,
+      "step": 6780
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 1.5714118061260416e-05,
+      "loss": 0.461,
+      "step": 6790
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 1.565543950240582e-05,
+      "loss": 0.4403,
+      "step": 6800
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 1.5596760943551227e-05,
+      "loss": 0.4149,
+      "step": 6810
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 1.5538082384696635e-05,
+      "loss": 0.4092,
+      "step": 6820
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 1.5479403825842037e-05,
+      "loss": 0.4069,
+      "step": 6830
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 1.5420725266987443e-05,
+      "loss": 0.4136,
+      "step": 6840
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 1.5362046708132848e-05,
+      "loss": 0.442,
+      "step": 6850
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 1.5303368149278257e-05,
+      "loss": 0.3859,
+      "step": 6860
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 1.524468959042366e-05,
+      "loss": 0.4179,
+      "step": 6870
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 1.5186011031569064e-05,
+      "loss": 0.4012,
+      "step": 6880
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 1.512733247271447e-05,
+      "loss": 0.4547,
+      "step": 6890
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 1.5068653913859876e-05,
+      "loss": 0.418,
+      "step": 6900
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 1.5009975355005281e-05,
+      "loss": 0.4015,
+      "step": 6910
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 1.4951296796150688e-05,
+      "loss": 0.3834,
+      "step": 6920
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 1.4892618237296093e-05,
+      "loss": 0.3703,
+      "step": 6930
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 1.4833939678441497e-05,
+      "loss": 0.3912,
+      "step": 6940
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 1.4775261119586902e-05,
+      "loss": 0.4257,
+      "step": 6950
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 1.471658256073231e-05,
+      "loss": 0.3835,
+      "step": 6960
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 1.4657904001877714e-05,
+      "loss": 0.3571,
+      "step": 6970
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 1.4599225443023121e-05,
+      "loss": 0.4066,
+      "step": 6980
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 1.4540546884168527e-05,
+      "loss": 0.3714,
+      "step": 6990
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 1.448186832531393e-05,
+      "loss": 0.3878,
+      "step": 7000
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 1.4423189766459336e-05,
+      "loss": 0.3906,
+      "step": 7010
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 1.436451120760474e-05,
+      "loss": 0.4136,
+      "step": 7020
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 1.4305832648750148e-05,
+      "loss": 0.4145,
+      "step": 7030
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 1.4247154089895553e-05,
+      "loss": 0.4616,
+      "step": 7040
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 1.418847553104096e-05,
+      "loss": 0.4761,
+      "step": 7050
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 1.4129796972186362e-05,
+      "loss": 0.4617,
+      "step": 7060
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 1.4071118413331769e-05,
+      "loss": 0.4072,
+      "step": 7070
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 1.4012439854477174e-05,
+      "loss": 0.4289,
+      "step": 7080
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 1.3953761295622581e-05,
+      "loss": 0.4492,
+      "step": 7090
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 1.3895082736767986e-05,
+      "loss": 0.5045,
+      "step": 7100
+    },
+    {
+      "epoch": 9.0,
+      "eval_accuracy": 0.9048712871287129,
+      "eval_loss": 0.36326825618743896,
+      "eval_runtime": 399.7139,
+      "eval_samples_per_second": 63.17,
+      "eval_steps_per_second": 2.634,
+      "step": 7103
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.3836404177913392e-05,
+      "loss": 0.3706,
+      "step": 7110
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.3777725619058795e-05,
+      "loss": 0.4068,
+      "step": 7120
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.3719047060204202e-05,
+      "loss": 0.4351,
+      "step": 7130
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.3660368501349607e-05,
+      "loss": 0.4385,
+      "step": 7140
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.3601689942495013e-05,
+      "loss": 0.4369,
+      "step": 7150
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 1.354301138364042e-05,
+      "loss": 0.408,
+      "step": 7160
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 1.3484332824785825e-05,
+      "loss": 0.386,
+      "step": 7170
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 1.3425654265931229e-05,
+      "loss": 0.3598,
+      "step": 7180
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 1.3366975707076634e-05,
+      "loss": 0.3675,
+      "step": 7190
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 1.330829714822204e-05,
+      "loss": 0.3872,
+      "step": 7200
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 1.3249618589367446e-05,
+      "loss": 0.4314,
+      "step": 7210
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 1.3190940030512853e-05,
+      "loss": 0.319,
+      "step": 7220
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 1.3132261471658255e-05,
+      "loss": 0.4001,
+      "step": 7230
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 1.3073582912803662e-05,
+      "loss": 0.3866,
+      "step": 7240
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 1.3014904353949067e-05,
+      "loss": 0.3192,
+      "step": 7250
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 1.2956225795094474e-05,
+      "loss": 0.4077,
+      "step": 7260
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 1.289754723623988e-05,
+      "loss": 0.4069,
+      "step": 7270
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 1.2838868677385285e-05,
+      "loss": 0.4144,
+      "step": 7280
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 1.2780190118530688e-05,
+      "loss": 0.3625,
+      "step": 7290
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 1.2721511559676093e-05,
+      "loss": 0.4387,
+      "step": 7300
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 1.26628330008215e-05,
+      "loss": 0.3916,
+      "step": 7310
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 1.2604154441966906e-05,
+      "loss": 0.3796,
+      "step": 7320
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 1.2545475883112313e-05,
+      "loss": 0.4104,
+      "step": 7330
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 1.2486797324257716e-05,
+      "loss": 0.3698,
+      "step": 7340
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 1.2428118765403123e-05,
+      "loss": 0.3828,
+      "step": 7350
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 1.2369440206548527e-05,
+      "loss": 0.338,
+      "step": 7360
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 1.2310761647693934e-05,
+      "loss": 0.3869,
+      "step": 7370
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 1.2252083088839339e-05,
+      "loss": 0.4145,
+      "step": 7380
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 1.2193404529984744e-05,
+      "loss": 0.4216,
+      "step": 7390
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 1.213472597113015e-05,
+      "loss": 0.3727,
+      "step": 7400
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.2076047412275555e-05,
+      "loss": 0.3338,
+      "step": 7410
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.201736885342096e-05,
+      "loss": 0.4048,
+      "step": 7420
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.1958690294566365e-05,
+      "loss": 0.3441,
+      "step": 7430
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.1900011735711772e-05,
+      "loss": 0.4916,
+      "step": 7440
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.1841333176857176e-05,
+      "loss": 0.3582,
+      "step": 7450
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.1782654618002583e-05,
+      "loss": 0.3774,
+      "step": 7460
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.1723976059147988e-05,
+      "loss": 0.3201,
+      "step": 7470
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.1665297500293393e-05,
+      "loss": 0.3788,
+      "step": 7480
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.1606618941438799e-05,
+      "loss": 0.3566,
+      "step": 7490
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.1547940382584206e-05,
+      "loss": 0.3326,
+      "step": 7500
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.1489261823729609e-05,
+      "loss": 0.4127,
+      "step": 7510
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.1430583264875014e-05,
+      "loss": 0.4035,
+      "step": 7520
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.1371904706020421e-05,
+      "loss": 0.3626,
+      "step": 7530
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.1313226147165825e-05,
+      "loss": 0.4444,
+      "step": 7540
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.1254547588311232e-05,
+      "loss": 0.4171,
+      "step": 7550
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.1195869029456637e-05,
+      "loss": 0.3903,
+      "step": 7560
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.1137190470602042e-05,
+      "loss": 0.4015,
+      "step": 7570
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.1078511911747448e-05,
+      "loss": 0.3856,
+      "step": 7580
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.1019833352892855e-05,
+      "loss": 0.3628,
+      "step": 7590
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.0961154794038258e-05,
+      "loss": 0.3791,
+      "step": 7600
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.0902476235183665e-05,
+      "loss": 0.3226,
+      "step": 7610
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.084379767632907e-05,
+      "loss": 0.4392,
+      "step": 7620
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.0785119117474476e-05,
+      "loss": 0.3796,
+      "step": 7630
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.0726440558619881e-05,
+      "loss": 0.3743,
+      "step": 7640
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.0667761999765286e-05,
+      "loss": 0.3533,
+      "step": 7650
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 1.0609083440910692e-05,
+      "loss": 0.3102,
+      "step": 7660
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 1.0550404882056097e-05,
+      "loss": 0.3815,
+      "step": 7670
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 1.0491726323201502e-05,
+      "loss": 0.4035,
+      "step": 7680
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 1.0433047764346907e-05,
+      "loss": 0.3654,
+      "step": 7690
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 1.0374369205492314e-05,
+      "loss": 0.347,
+      "step": 7700
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 1.0315690646637718e-05,
+      "loss": 0.4099,
+      "step": 7710
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 1.0257012087783125e-05,
+      "loss": 0.2906,
+      "step": 7720
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 1.019833352892853e-05,
+      "loss": 0.3875,
+      "step": 7730
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 1.0139654970073935e-05,
+      "loss": 0.3742,
+      "step": 7740
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 1.008097641121934e-05,
+      "loss": 0.3755,
+      "step": 7750
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 1.0022297852364746e-05,
+      "loss": 0.39,
+      "step": 7760
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 9.963619293510151e-06,
+      "loss": 0.3588,
+      "step": 7770
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 9.904940734655556e-06,
+      "loss": 0.342,
+      "step": 7780
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 9.846262175800963e-06,
+      "loss": 0.3335,
+      "step": 7790
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 9.787583616946367e-06,
+      "loss": 0.3922,
+      "step": 7800
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 9.728905058091774e-06,
+      "loss": 0.379,
+      "step": 7810
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 9.67022649923718e-06,
+      "loss": 0.3741,
+      "step": 7820
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 9.611547940382585e-06,
+      "loss": 0.3943,
+      "step": 7830
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 9.55286938152799e-06,
+      "loss": 0.4278,
+      "step": 7840
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 9.494190822673397e-06,
+      "loss": 0.3774,
+      "step": 7850
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 9.4355122638188e-06,
+      "loss": 0.3941,
+      "step": 7860
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 9.376833704964207e-06,
+      "loss": 0.4246,
+      "step": 7870
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 9.318155146109613e-06,
+      "loss": 0.3817,
+      "step": 7880
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 9.259476587255018e-06,
+      "loss": 0.3437,
+      "step": 7890
+    },
+    {
+      "epoch": 10.0,
+      "eval_accuracy": 0.9057821782178218,
+      "eval_loss": 0.3611043095588684,
+      "eval_runtime": 397.5837,
+      "eval_samples_per_second": 63.509,
+      "eval_steps_per_second": 2.648,
+      "step": 7892
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 9.200798028400423e-06,
+      "loss": 0.3838,
+      "step": 7900
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 9.142119469545828e-06,
+      "loss": 0.3519,
+      "step": 7910
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 9.083440910691234e-06,
+      "loss": 0.3564,
+      "step": 7920
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 9.024762351836639e-06,
+      "loss": 0.3672,
+      "step": 7930
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 8.966083792982046e-06,
+      "loss": 0.3838,
+      "step": 7940
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 8.90740523412745e-06,
+      "loss": 0.3876,
+      "step": 7950
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 8.848726675272856e-06,
+      "loss": 0.3398,
+      "step": 7960
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 8.790048116418262e-06,
+      "loss": 0.3985,
+      "step": 7970
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 8.731369557563667e-06,
+      "loss": 0.4459,
+      "step": 7980
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 8.672690998709072e-06,
+      "loss": 0.3988,
+      "step": 7990
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 8.614012439854477e-06,
+      "loss": 0.4287,
+      "step": 8000
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 8.555333880999883e-06,
+      "loss": 0.3405,
+      "step": 8010
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 8.496655322145288e-06,
+      "loss": 0.3314,
+      "step": 8020
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 8.437976763290693e-06,
+      "loss": 0.3314,
+      "step": 8030
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 8.379298204436099e-06,
+      "loss": 0.3549,
+      "step": 8040
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 8.320619645581505e-06,
+      "loss": 0.3997,
+      "step": 8050
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 8.261941086726909e-06,
+      "loss": 0.283,
+      "step": 8060
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 8.203262527872316e-06,
+      "loss": 0.3864,
+      "step": 8070
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 8.144583969017721e-06,
+      "loss": 0.3714,
+      "step": 8080
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 8.085905410163127e-06,
+      "loss": 0.3561,
+      "step": 8090
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 8.027226851308532e-06,
+      "loss": 0.3896,
+      "step": 8100
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 7.968548292453939e-06,
+      "loss": 0.4098,
+      "step": 8110
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 7.909869733599342e-06,
+      "loss": 0.3365,
+      "step": 8120
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 7.85119117474475e-06,
+      "loss": 0.3531,
+      "step": 8130
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 7.792512615890155e-06,
+      "loss": 0.3558,
+      "step": 8140
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 7.73383405703556e-06,
+      "loss": 0.3807,
+      "step": 8150
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 7.675155498180965e-06,
+      "loss": 0.3728,
+      "step": 8160
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 7.616476939326371e-06,
+      "loss": 0.2985,
+      "step": 8170
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 7.557798380471776e-06,
+      "loss": 0.3531,
+      "step": 8180
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 7.499119821617182e-06,
+      "loss": 0.3739,
+      "step": 8190
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 7.440441262762587e-06,
+      "loss": 0.4106,
+      "step": 8200
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 7.381762703907992e-06,
+      "loss": 0.3351,
+      "step": 8210
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 7.323084145053398e-06,
+      "loss": 0.3916,
+      "step": 8220
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 7.264405586198804e-06,
+      "loss": 0.3895,
+      "step": 8230
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 7.205727027344208e-06,
+      "loss": 0.3183,
+      "step": 8240
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 7.147048468489614e-06,
+      "loss": 0.3507,
+      "step": 8250
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 7.08836990963502e-06,
+      "loss": 0.3841,
+      "step": 8260
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 7.029691350780425e-06,
+      "loss": 0.3343,
+      "step": 8270
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 6.971012791925831e-06,
+      "loss": 0.3278,
+      "step": 8280
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 6.912334233071237e-06,
+      "loss": 0.312,
+      "step": 8290
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 6.8536556742166414e-06,
+      "loss": 0.3642,
+      "step": 8300
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 6.7949771153620476e-06,
+      "loss": 0.3633,
+      "step": 8310
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 6.736298556507452e-06,
+      "loss": 0.4392,
+      "step": 8320
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 6.677619997652858e-06,
+      "loss": 0.3309,
+      "step": 8330
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 6.618941438798263e-06,
+      "loss": 0.4279,
+      "step": 8340
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 6.560262879943669e-06,
+      "loss": 0.3741,
+      "step": 8350
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 6.501584321089074e-06,
+      "loss": 0.4051,
+      "step": 8360
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 6.44290576223448e-06,
+      "loss": 0.3219,
+      "step": 8370
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 6.3842272033798844e-06,
+      "loss": 0.3454,
+      "step": 8380
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 6.3255486445252906e-06,
+      "loss": 0.3832,
+      "step": 8390
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 6.266870085670697e-06,
+      "loss": 0.3234,
+      "step": 8400
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 6.208191526816102e-06,
+      "loss": 0.3343,
+      "step": 8410
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 6.149512967961507e-06,
+      "loss": 0.3536,
+      "step": 8420
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 6.0908344091069125e-06,
+      "loss": 0.3374,
+      "step": 8430
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 6.032155850252319e-06,
+      "loss": 0.3137,
+      "step": 8440
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 5.973477291397724e-06,
+      "loss": 0.3508,
+      "step": 8450
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 5.914798732543129e-06,
+      "loss": 0.3439,
+      "step": 8460
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 5.856120173688534e-06,
+      "loss": 0.3465,
+      "step": 8470
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 5.79744161483394e-06,
+      "loss": 0.3876,
+      "step": 8480
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 5.738763055979345e-06,
+      "loss": 0.3773,
+      "step": 8490
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 5.680084497124751e-06,
+      "loss": 0.312,
+      "step": 8500
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 5.621405938270156e-06,
+      "loss": 0.3878,
+      "step": 8510
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 5.562727379415562e-06,
+      "loss": 0.3474,
+      "step": 8520
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 5.504048820560968e-06,
+      "loss": 0.3077,
+      "step": 8530
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 5.445370261706373e-06,
+      "loss": 0.3076,
+      "step": 8540
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 5.386691702851778e-06,
+      "loss": 0.3651,
+      "step": 8550
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 5.3280131439971835e-06,
+      "loss": 0.402,
+      "step": 8560
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 5.26933458514259e-06,
+      "loss": 0.3682,
+      "step": 8570
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 5.210656026287995e-06,
+      "loss": 0.3735,
+      "step": 8580
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 5.1519774674334e-06,
+      "loss": 0.3675,
+      "step": 8590
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 5.093298908578805e-06,
+      "loss": 0.3724,
+      "step": 8600
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 5.034620349724211e-06,
+      "loss": 0.3823,
+      "step": 8610
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 4.975941790869616e-06,
+      "loss": 0.3348,
+      "step": 8620
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 4.917263232015022e-06,
+      "loss": 0.3704,
+      "step": 8630
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 4.858584673160427e-06,
+      "loss": 0.3299,
+      "step": 8640
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 4.799906114305833e-06,
+      "loss": 0.3261,
+      "step": 8650
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 4.741227555451239e-06,
+      "loss": 0.3746,
+      "step": 8660
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 4.682548996596644e-06,
+      "loss": 0.3931,
+      "step": 8670
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 4.623870437742049e-06,
+      "loss": 0.3494,
+      "step": 8680
+    },
+    {
+      "epoch": 11.0,
+      "eval_accuracy": 0.905980198019802,
+      "eval_loss": 0.3568229675292969,
+      "eval_runtime": 397.9365,
+      "eval_samples_per_second": 63.452,
+      "eval_steps_per_second": 2.646,
+      "step": 8681
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 4.565191878887455e-06,
+      "loss": 0.371,
+      "step": 8690
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 4.506513320032861e-06,
+      "loss": 0.3451,
+      "step": 8700
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 4.447834761178266e-06,
+      "loss": 0.3064,
+      "step": 8710
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 4.389156202323671e-06,
+      "loss": 0.3684,
+      "step": 8720
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 4.3304776434690765e-06,
+      "loss": 0.3499,
+      "step": 8730
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 4.271799084614482e-06,
+      "loss": 0.3823,
+      "step": 8740
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 4.213120525759887e-06,
+      "loss": 0.3393,
+      "step": 8750
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 4.154441966905293e-06,
+      "loss": 0.3612,
+      "step": 8760
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 4.095763408050698e-06,
+      "loss": 0.4089,
+      "step": 8770
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 4.037084849196104e-06,
+      "loss": 0.298,
+      "step": 8780
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 3.97840629034151e-06,
+      "loss": 0.3414,
+      "step": 8790
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 3.919727731486915e-06,
+      "loss": 0.2924,
+      "step": 8800
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 3.86104917263232e-06,
+      "loss": 0.364,
+      "step": 8810
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 3.802370613777726e-06,
+      "loss": 0.3016,
+      "step": 8820
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 3.7436920549231313e-06,
+      "loss": 0.3236,
+      "step": 8830
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 3.6850134960685365e-06,
+      "loss": 0.3938,
+      "step": 8840
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 3.6263349372139426e-06,
+      "loss": 0.3844,
+      "step": 8850
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 3.567656378359348e-06,
+      "loss": 0.2905,
+      "step": 8860
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 3.508977819504753e-06,
+      "loss": 0.3091,
+      "step": 8870
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 3.4502992606501584e-06,
+      "loss": 0.3235,
+      "step": 8880
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 3.391620701795564e-06,
+      "loss": 0.328,
+      "step": 8890
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 3.3329421429409694e-06,
+      "loss": 0.2829,
+      "step": 8900
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 3.2742635840863747e-06,
+      "loss": 0.3687,
+      "step": 8910
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 3.2155850252317808e-06,
+      "loss": 0.3096,
+      "step": 8920
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 3.156906466377186e-06,
+      "loss": 0.3464,
+      "step": 8930
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 3.0982279075225913e-06,
+      "loss": 0.3444,
+      "step": 8940
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 3.0395493486679966e-06,
+      "loss": 0.3806,
+      "step": 8950
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 2.9808707898134023e-06,
+      "loss": 0.3588,
+      "step": 8960
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 2.9221922309588076e-06,
+      "loss": 0.3099,
+      "step": 8970
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 2.8635136721042132e-06,
+      "loss": 0.3905,
+      "step": 8980
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 2.804835113249619e-06,
+      "loss": 0.2535,
+      "step": 8990
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 2.746156554395024e-06,
+      "loss": 0.3194,
+      "step": 9000
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 2.6874779955404295e-06,
+      "loss": 0.3438,
+      "step": 9010
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 2.628799436685835e-06,
+      "loss": 0.3744,
+      "step": 9020
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 2.5701208778312404e-06,
+      "loss": 0.3083,
+      "step": 9030
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 2.511442318976646e-06,
+      "loss": 0.2938,
+      "step": 9040
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 2.4527637601220514e-06,
+      "loss": 0.398,
+      "step": 9050
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 2.394085201267457e-06,
+      "loss": 0.3838,
+      "step": 9060
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 2.3354066424128624e-06,
+      "loss": 0.3276,
+      "step": 9070
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 2.2767280835582676e-06,
+      "loss": 0.3572,
+      "step": 9080
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 2.2180495247036733e-06,
+      "loss": 0.3137,
+      "step": 9090
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 2.159370965849079e-06,
+      "loss": 0.3818,
+      "step": 9100
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 2.1006924069944843e-06,
+      "loss": 0.3616,
+      "step": 9110
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 2.04201384813989e-06,
+      "loss": 0.3378,
+      "step": 9120
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 1.9833352892852952e-06,
+      "loss": 0.3852,
+      "step": 9130
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 1.9246567304307005e-06,
+      "loss": 0.3564,
+      "step": 9140
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 1.8659781715761062e-06,
+      "loss": 0.3587,
+      "step": 9150
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 1.8072996127215115e-06,
+      "loss": 0.3007,
+      "step": 9160
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 1.7486210538669172e-06,
+      "loss": 0.3093,
+      "step": 9170
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 1.6899424950123226e-06,
+      "loss": 0.3176,
+      "step": 9180
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 1.631263936157728e-06,
+      "loss": 0.4034,
+      "step": 9190
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 1.5725853773031336e-06,
+      "loss": 0.3128,
+      "step": 9200
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 1.513906818448539e-06,
+      "loss": 0.3264,
+      "step": 9210
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 1.4552282595939443e-06,
+      "loss": 0.3578,
+      "step": 9220
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 1.3965497007393498e-06,
+      "loss": 0.4155,
+      "step": 9230
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 1.3378711418847555e-06,
+      "loss": 0.4095,
+      "step": 9240
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 1.2791925830301608e-06,
+      "loss": 0.3355,
+      "step": 9250
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 1.2205140241755663e-06,
+      "loss": 0.3755,
+      "step": 9260
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 1.1618354653209717e-06,
+      "loss": 0.3389,
+      "step": 9270
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 1.1031569064663772e-06,
+      "loss": 0.3319,
+      "step": 9280
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 1.0444783476117827e-06,
+      "loss": 0.3664,
+      "step": 9290
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 9.857997887571882e-07,
+      "loss": 0.3523,
+      "step": 9300
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 9.271212299025936e-07,
+      "loss": 0.3214,
+      "step": 9310
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 8.684426710479991e-07,
+      "loss": 0.3444,
+      "step": 9320
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 8.097641121934046e-07,
+      "loss": 0.3018,
+      "step": 9330
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 7.5108555333881e-07,
+      "loss": 0.3467,
+      "step": 9340
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 6.924069944842155e-07,
+      "loss": 0.3421,
+      "step": 9350
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 6.33728435629621e-07,
+      "loss": 0.3314,
+      "step": 9360
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 5.750498767750264e-07,
+      "loss": 0.39,
+      "step": 9370
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 5.163713179204318e-07,
+      "loss": 0.3077,
+      "step": 9380
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 4.5769275906583735e-07,
+      "loss": 0.3459,
+      "step": 9390
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 3.990142002112428e-07,
+      "loss": 0.3685,
+      "step": 9400
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 3.403356413566483e-07,
+      "loss": 0.3392,
+      "step": 9410
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 2.816570825020538e-07,
+      "loss": 0.3384,
+      "step": 9420
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 2.2297852364745924e-07,
+      "loss": 0.3421,
+      "step": 9430
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 1.642999647928647e-07,
+      "loss": 0.3268,
+      "step": 9440
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 1.0562140593827016e-07,
+      "loss": 0.3693,
+      "step": 9450
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 4.6942847083675625e-08,
+      "loss": 0.3381,
+      "step": 9460
+    },
+    {
+      "epoch": 12.0,
+      "eval_accuracy": 0.9068514851485149,
+      "eval_loss": 0.3532540202140808,
+      "eval_runtime": 400.3011,
+      "eval_samples_per_second": 63.078,
+      "eval_steps_per_second": 2.631,
+      "step": 9468
+    },
+    {
+      "epoch": 12.0,
+      "step": 9468,
+      "total_flos": 7.048177916397006e+19,
+      "train_loss": 0.9189980125245939,
+      "train_runtime": 36697.4055,
+      "train_samples_per_second": 24.77,
+      "train_steps_per_second": 0.258
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 9468,
+  "num_train_epochs": 12,
+  "save_steps": 500,
+  "total_flos": 7.048177916397006e+19,
+  "trial_name": null,
+  "trial_params": null
+}